DESIGN AND PERFORMANCE OF JOINT REPLACEMENTS

Among the earliest methods for treating arthritis (212)
was interposition arthroplasty using soft and flexible materials, but
the strength of these materials was inadequate. Rigid materials such as
metal and glass in the form of condylar shapes attached to one of the
joint surfaces provided some success, but there was still a problem
with the apposing joint surface. In the knee, geometric inaccuracy
leading to poor kinematics and abnormal soft-tissue tensions was also a
problem. An interesting finding with condylar components, such as cup
arthroplasty or MacIntosh tibial plateaus, was the formation of a
fibrous membrane adjacent to the component, with a new bone plate
beneath. This tissue modeling is now recognized to develop due to
interface micromotion and to the stresses acting on the exposed
trabecular ends. Even now, implants are designed with relatively smooth
surfaces interfacing the bone; however, there is a higher incidence of
pain, migration, and loosening with these implants compared with the
more rigid methods of fixation. The success of more invasive components
with uncemented intramedullary stems, such as hemiarthroplasties and
knee hinges, depended largely on obtaining an acceptably low level of
stem–bone micromotion and interface stresses on the bone (33).

Modern-day joint replacement began in the early 1960s,
when Charnley introduced cemented metal-polyethylene components for the
hip, and in the late 1960s, when this technology was transferred to the
knee by Gunston (95). The principles proposed
by Charnley were rigid fixation of the components to the bone,
resurfacing of both joint surfaces, and the use of materials with low
friction and wear. These principles, embodied in cemented
metal-on-plastic components, have stood the test of time to this day (100,112).

In the 1970s, there was a rapid expansion in the number
of total joints used. A number of other hip designs were introduced
that were fundamentally based on the Charnley design. An exception was
the use of metal-on-metal in the McKee-Farrar design. There was a great
deal of attention to surgical technique, especially obtaining good
cement–bone apposition with high shear strength. Some total knees
proved to be successful in the long term, but others fell by the
wayside because of the lack of recognition of the kinematics, the role
of the cruciates, the patellofemoral mechanics, and the requirements
for adequate fixation. Most of the design forms used today, namely
unicompartmentals, condylar replacements with or without cruciate
retention, mobile bearing knees, stabilized condylars, and fixed and
rotating hinges, were all introduced before the end of the decade.
Ceramic-on-polyethylene and ceramic-on-ceramic replacements for the hip
had been introduced before 1980.

The 1980s saw two substantial areas of development: more
sophisticated instrumentation, especially for the knee, and uncemented
components with porous coatings intended for indefinite fixation. In
the knee, improved consistency of surgical technique was achieved as a
result of more accurate alignment and soft-tissue balancing. Concerning
porous coating, those designs in which rigid initial fixation was
achieved obtained sufficient bone ingrowth for long-lasting results.
Certain uncemented designs of hip stem, acetabular component, and
femoral component of the knee have shown survivorship values (54) superior to those of cemented components. A hydroxyapatite (HA) coating (104,203)
has similarly shown high durability from 5 to 10 years follow-up. The
situation today is that a number of designs of hips, knees, and other
joints have been shown to have a survivorship of greater than 90% at 10
years, such that the large majority of elderly patients can be treated
confidently. Today’s hip and knee systems offer a large variety of
sizes and modular augments to deal with virtually every situation
encountered in primary and revision surgery. The instrumentation
systems are elaborate, usually well engineered, and often too complex,
but they allow the surgeon to achieve accurate component placement and
limb alignment.

The main limitations of total hip and knee replacements
are excessive wear of the ultra-high molecular weight polyethylene
(UHMWPE) (126,236) and
loosening. These two problems are related to some extent in that the
accumulation of small particles causes a tissue response that, in turn,
produces bone resorption. However, it is now evident that wear can be
reduced in a number of ways. The main problem has been that
gamma-irradiation of UHMWPE in air, followed by gradual oxidation
either on the shelf or in the patient, has led to a degradation of
mechanical properties and an increase in the wear rate (18,140,180).

Components that have been directly molded from UHMWPE
have shown a reduced susceptibility to oxidation, which has resulted in
reduced surface wear and a much lower incidence of the destructive
delamination seen in total knee replacements (TKRs) (27).
Gamma-irradiation and storage in an inert atmosphere, as well as
enhanced cross-linking and stabilization to minimize subsequent
oxidation, have resulted in reduced wear rates. For the hip joint,
ceramic-on-UHMWPE can reduce wear by as much as 50%, whereas
ceramic-on-ceramic and metal-on-metal produce minimal wear debris (65,82).
If such improved bearings are combined with superior cement techniques
or porous and bioactive coatings, much greater durability can be
expected compared with that achieved in the past 2 decades.

In the knee, however, delamination due to high
subsurface stresses is still a threat. Computer modeling of this
process has now shown how appropriate surface geometry can
substantially reduce the likelihood of delamination without
compromising the freedom of motion (189). New
advances will undoubtedly be made in the performance of TKR, especially
for younger patients. The mobile bearing concept is one such approach
for minimizing wear and improving function. However, further advances
are likely in the form of guided motion knees providing optimal muscle
lever arms and high flexion ranges. The superior functional performance
of unicompartmental replacement performed on one or both sides of the
joint opens up the possibilities of minimally invasive surgery (84),
with the option of using computer-assisted technologies. The boundary
between biologic treatments and total joints will become an issue, and
with both approaches improving and developing, the relative merits and
applications will become an area for extensive research and evaluation (59,70,72,79,141,142,152,162,165,177).

The motion in a single plane is first considered. Figure 100.1 shows the knee schematically with an origin A fixed in the femur and a contact point B on the tibial surface. The knee is set in a global axis system ZY with origin O (the X axis is used for the lateral-medial direction for a right knee). The femur is first given a rotation F about A. The femur is next given a displacement or translation
p in the negative Z direction. Finally, the femur is given an
additional displacement d along the negative Y direction. The tibia has
remained fixed with respect to the global axes, and the femoral motions
could be described with respect to an axis system fixed in the tibia.
The values F, p, and d are vectors in that they have magnitude and direction, often denoted with a line above or in bold type.

From a kinematics point of view, the final position of the femur can be obtained by a single rotation F about an axis D (Fig. 100.2). An instant center of rotation (207) is defined for small motions. The centers for a succession of small motions is the instant center pathway,
not to be confused with the geometric centers of local radii of
curvature. If the instant center is at the center of curvature, the
motion at the contact point is called pure sliding (124). If the center is at the contact point, the motion is pure rolling (31). A center in between the two produces sliding and rolling combined.

In general, to visualize motion, it is usually easier to
define rotations and displacements with respect to axes, rather than
using centers of rotation (238) by the following principles:

The importance of these principles is seen in visualizing three-dimensional (3-D) motion. In Figure 100.3,
the femur is shown in its initial position on the tibia at zero flexion
with contact points L and M. The femur is first rotated by 90° (flexed)
about a transverse axis through A. Next the femur is rotated about a
vertical axis through A fixed relative to the tibial axes. The lateral
contact point moves posteriorly by a and the medial anteriorly by a. The femur is next translated posteriorly by a. The lateral contact point has now displaced posteriorly by a total of 2a,
but the medial condyle is in its original position. In orthopaedic
terms, as the knee has flexed, there has been an internal rotation of
the tibia and a posterior translation of the femur.

Note that, depending on the order of the motions, the
posterior displacement of the lateral contact point can occur on an
unrotated or rotated tibia, producing a slightly different end result.
The entire motion could have been achieved by a single rotation about
and a displacement along a screw axis (235),
but this is more difficult to visualize conceptually than the sequence
of rotations and displacements described. In addition, the internal
rotation and posterior displacement could have been achieved by a
rotation about a vertical axis through the medial tibial condyle.
Hence, it is important to define the method of describing the motion
and which axes are used, to avoid misunderstanding and confusion.

For the hip joint, the origin of the femur is the center
of the femoral head and the vertical axis is through the center of the
knee (Fig. 100.4). The anteroposterior (AP) and
mediolateral (ML) axes are self-evident, and all axes are mutually
perpendicular. The axes in the acetabulum and femur are conveniently
defined as being coincident with the hip in the neutral position. The
motion of the femur is defined in relation to a fixed acetabulum as
follows:

The importance of the orders of the rotation can be seen by reversing the order of the first two.

The motion of the knee is more complex in that it involves six degrees of freedom—three rotations, and three translations (56) (Fig. 100.5).
The transverse axis in the femur can be chosen through the epicondyles.
The other two axes are mutually perpendicular. For convenience, the
axes of the tibia in the initial reference position at zero flexion can
be taken to be the same as for the femur. Ordered motions are defined
for the femoral axis system relative to the tibial axis system.

A modification in the way the axes are defined has been
used to define a set of order-independent motions of the knee. This is
the Grood-Suntay system (93), which has the advantage of ease of visualization.

For a joint with uniaxial motion, such as the elbow (67),
it is useful to define the axes with respect to the axis of rotation,
rather than using the long axes of the bones. The description of the
motion can then be simplified to a single degree of freedom by the
choice of the axis, which

bisects the carrying angle (Fig. 100.6).
This axis becomes a common axis in the humerus and ulna, the other two
axes being mutually perpendicular as shown. In this case, the long axes
of the bones are not coincident with the reference axes.

The discussion in the previous section deals with
different motions and the paths from the initial position to the final
position. Successive rotations about three axes as shown in the
preceding figures are called eulerian or cardan angles (6,235). Transformation matrices
consisting of sines and cosines are used to convert the coordinates of
points on the object from the initial to the final position with
respect to one of the defined axis systems. A major interest in
artificial joints is the continuum of motion during dynamic activities,
an aspect that will be discussed later.

Forces, like motions, are vector quantities. In this
case, they have magnitude, line of action, and point of application.
Exactly the same axis systems used for the bones can be used to define
the forces. For the hip in the neutral position, there are three components of force along each of the three axes, which can be combined into a single resultant force (Fig. 100.7). The forces on the femoral head are exactly equal and opposite
to those on the acetabulum. If the femur is now moved to any arbitrary
position, the resultant forces on the femoral head and in the
acetabulum will still be equal and opposite. However, the three
components of force along each of the femoral axes are now different
from those along the acetabular axes. Hence, when defining the forces
across a joint, it is important to specify the axis system to which the
forces are referred. In reality, of course, the resultant force is
representative of the pressure acting over a large surface area.

In the knee, to represent the forces acting on the joint surfaces as well as along the axes, a moment in the frontal plane can be considered (Fig. 100.8). A moment has the units of force × distance, for example, N·m or N·mm. In the case shown, the main force is axial compression FY acting in the negative YT direction. An anterior shear force FZ is shown acting along positive ZT, and a varus moment MZ is shown acting about ZT. This set of forces and moments is shown distributed
to the upper tibia. The compressive force and the varus moment produce
a larger resultant force on the medial condyle than on the lateral
condyle. The forces themselves are representative of the contact
pressures and act at the center of pressure
of each contact area. The shear force is transmitted to the tibia by
the contact areas being on the upward anterior slopes of the tibial
plateaus and by a force in the posterior cruciate ligament (PCL).

A method for the analysis of forces is shown in Figure 100.9,

in which the problem is to calculate the forces on the face of the glenoid with the arm abducted and a weight in the hand (174). The steps are:

Assuming that the geometry is known (e.g., from
radiographs), there are three unknowns, namely FX, FY, and FM. Three
equations are thus required to obtain a solution.

Even though the resultant forces across joints are
depicted with lines of action, in reality the forces are transmitted
across areas on the joint surfaces. This results in a contact pressure, or contact stress, acting on the surface (124). The density of the subchondral bone plate will provide an indication of the pressure distribution (161).

The units of pressure are pascals, equal to newtons per square meter (N/m²). Because this is such a small quantity, mega-pascals (mPa), equal to newtons per square millimeter (N/mm²)
are used. Whereas static analysis, as shown in the previous section,
can be used to calculate resultant forces, to calculate pressures and
stresses in an implant-bone system, more complex analyses are required.
For simple geometries, elasticity (e.g., hertzian) equations can be
used to calculate the contact areas and stresses in terms of the radii
and the elastic properties of the materials. However, for realistic
bone and implant shapes, finite element analysis (FEA) (15,16,44,175) is the most appropriate technique.

If a sphere is loaded onto a flat surface, a circular contact area is formed due to deformation of both surfaces (Fig. 100.10). The pressure distribution
is hemispherical, with the maximum pressure at the center being 1.5
times the mean pressure. The same situation occurs if the lower surface
is flat, a convex spherical surface, or a concave spherical surface.

The contact area increases as any one or a combination of the following conditions occurs:

The contact pressure increases as any one or a combination of the following conditions occurs:

The conformity has a major influence on the pressure, the relevant parameter being the relative radius of curvature R defined by:

where R₁ and R₂ are the radii of
curvature for the two surfaces, and the radius of curvature is taken as
positive for convex surfaces and negative for concave surfaces.

Thus, any two surfaces can be represented by a surface with radius R
on a flat surface. The average contact pressures calculated for a given
force (1000 N) and material properties (metal-on-plastic) are shown in Table 100.2. This table shows how the relative radius and the contact

pressure change dramatically as the surfaces approach full conformity.
The range covers a nonconforming knee to a hip joint with a small
femoral–acetabular clearance (15). Note, however, that the standard hertzian equations lose accuracy when the contact radius (R₂) approaches the surface radius (R₁). Another factor is that the contact pressure does not necessarily predict the wear rate, which is so often assumed (148).

A cylinder on a flat surface (with an infinite radius of
curvature) has a much larger contact area than does a sphere on the
same flat surface (Fig. 100.10). Examples of
arthroplasties with such contacts include some knees, elbows, ankles,
and fingers. Today, most geometries in condylar knees are toroidal,
between spherical and cylindrical, producing elliptical contact areas.
This has the advantages of reducing contact stresses and avoiding
“digging-in” at the sides if tilting occurs. The stresses beneath the
surface in all types of contacts are extremely important in both
natural and artificial joints, in terms of producing subsurface damage.

High pressures can act across implant–bone interfaces (39) (Fig. 100.11).
(Note that the word “stress” can be substituted for the word “pressure”
in this context.) In the case of metallic cemented or uncemented hips,
the contact pressures are highest in the proximal-medial region but
with high pressures distal-lateral also. The magnitudes of the
pressures depend also on stem geometry and head offset. For uncemented
stems, in which a “perfect” metal–bone fit is not achieved, local
contact pressures can be very high due to the stiffnesses of the
materials. An additional interface effect is due to friction between
the metal and the bone, causing shear forces to act as shown in Figure 100.11 (101,198).
If there is bonding across the interface from cement or from a porous
or HA coating, the shear forces will be higher, decreasing the
interface contact pressures. The interface shear stress is defined as:

In designing implant components for whatever location, the design goals for the interface stresses are:

The implications of goals 2 and 3 are to minimize or
avoid intramedullary fixation, or to design the component to minimize
tension and shear.

On the outside surface of a structure on which no direct
forces are acting, in general, there will be stresses acting in the
plane of the surface (Fig. 100.12). If a small square is drawn on the surface, there will be one direct stress (tensile or compressive) acting on opposite faces and another direct stress acting on the other pair of faces. There will be a single value of shear stress
acting to distort the square. If the square is now rotated, an
orientation can be found at which the shear stress is zero. The direct
stresses acting at this orientation are called the principal stresses.
One of these is the maximum and the other the minimum. Tension is taken
to be positive; compression negative. However, the principal stresses
can be both negative, both positive, or one negative and one positive,
depending on

the loading conditions around the square of surface. The maximum shear stress is at 45° to the principal stresses.

In three dimensions, the general principles are the
same. Here, a small cube of material is taken, which has three direct
stresses and three shear stresses. At a certain orientation, the shear
stresses are zero and the direct stresses are again called the
principal stresses.

Stress calculations of bones and implant components often quote values of Von Mises stress.
This term is convenient in that it describes the stress state at a
point with a single number. In fact, it is calculated from the three
principal stresses, and is a value that quantifies the likelihood of
yielding, synonymous with the yield stress in direct tension.

Another often-used quantity is strain energy density,
which is the amount of energy (force × distance) stored in the material
due to the applied stresses deforming the material. This quantity is
considered to be relevant to bone remodeling, whereby the response of
osteoblasts within the bone material is proportionate to the amount of
deformation and hence strain energy density. This is an “elastic”
quantity recoverable on removal of the stresses, but yielding of the
material is an additional consideration.

Material properties relate only to the material itself
and not to the shape of the object, whether it be a bone or an implant
component. A fundamental property is modulus of elasticity E, which can be determined by a simple tensile or compressive test on a cylinder (Fig. 100.13):

where stress = force ÷ area = F/A₀, and A₀ is the original cross-sectional area.

The change in a linear dimension is described by strain:

For stiffer materials, such as bone and metals, the
strain is small, usually measured in microstrain. A typical strain
value for the cortex of a long bone under peak load is on the order of
2000 to 3000 microstrain. For materials such as cartilage and ligament,
however, the strain can be 10% or more, and hence the true stress after load application must be related to the deformed area σ_T = F/A₁, where A₁ is the area with the force applied. Poissons ratio ν defines the strain in directions perpendicular to the longitudinal strain:

For metals ν equals 0.3, for plastics ν equals 0.3 to
0.4, and for an incompressible material such as a fluid or rubber, ν
equals 0.5.

For metals, the stress-strain relation is linear (Fig. 100.14A). The elastic modulus, defined as the slope of the stress-strain curve, is constant. Ligament and tendon are strain stiffening due to the straightening of collagen fibrils (86). This means that the elastic modulus is not constant

but increases steadily with strain. Rubbers usually exhibit strain
stiffening also. Conversely, materials such as polyethylene exhibit
strain-softening behavior, in which the elastic modulus appears to
reduce with strain. For these nonlinear materials, modulus of
elasticity has to be defined as a tangent modulus at a particular value of strain, as shown in Figure 100.14A.

Materials for which the strain recovers immediately on removal of the force are called elastic (Fig. 100.14B). If there is a time delay in recovery and a hysteresis loop is formed, the material is defined as viscoelastic, and some energy is absorbed in the cycle. A viscous material, such as synovial fluid, does not recover the strain (or only a very small amount) after force removal.

In some materials, if a force is applied suddenly, there
is an immediate strain, followed by a continued strain at a decreasing
rate, reaching an asymptotic level (Fig. 100.14C). The behavior is called creep or cold flow.
When the force is removed, the reverse process occurs. Polyethylene
behaves in this way, as does articular cartilage, tendon, and ligament.
Stretching exercises before a sporting activity put such tissues
through viscoelastic and creep cycles. Acrylic cement, when subjected
to sustained loading in a fluid environment at 37°C, shows some creep,
and this may be relevant to the slight sinkage of polished hip stems
over time.

Materials for which the properties are the same in all directions are called isotropic. Most artificial materials used in joints are of this type, although a number of experimental composite materials consisting of aligned fibers embedded in a polymeric matrix are nonisotropic. These materials are called orthotropic
if a property such as modulus or strength is much greater in only one
particular direction. Most biologic materials are orthotropic, which
results from their structure, either aligned collagen fibers in the
case of ligament and tendon, or aligned osteons in the case of bone (128).
Articular cartilage is a special case with complex properties due to
its triphasic composition of collagen fibers, mucopolysaccharide
matrix, and fluid. The values of modulus of elasticity for the
materials discussed cover a wide range (Table 100.3),
and consequently, the transfer of forces between different structures
is complex and can involve regions of stress concentration.

Material failure in a single load episode can occur in
either tension, compression, or shear. In general, tensile failure is
the most common. If a material is stretched and fails suddenly, that is
termed brittle fracture. Ceramics and acrylic cement behave in this way. Metals and polyethylene, on the other hand, are ductile in that there is a region of plastic deformation whereby the material elongates at essentially the same applied load. If the load is released, there is a permanent deformation,
but the material can continue to be structurally useful in that
condition. Permanent deformation in plastic tibial and patellar
components is an example. Tendons and ligaments behave as if they were
numerous fascicles in combination (41,241),
with rupture of the tightest fascicles occurring first. If the load is
released and sufficient fibers are intact, the structure can still
function. Bone is close to being brittle, although there is some
plastic deformation due to pull-out of osteons from the matrix (58,128,159).

Failure of a material can occur at stress levels below
the level at which single cycle yield or fracture occurs due to
repetitive cycling of forces, a process known as fatigue. This process has been observed on many orthopaedic devices including hip stems (93) (Fig. 100.15), tibial trays (1), and plastic liners (Fig. 100.16). The fatigue strength or fatigue limit is the stress below which no failure would occur no matter how many cycles of loading occurred (Table 100.3).
This value is generally about two thirds of the failure stress for a
single-load application. Fatigue failure initiates at the location of
the highest stresses, and the crack gradually propagates through the
section until the stress reaches the level at which complete failure
occurs at a single load. The starting point of the crack can be a small
defect in the material structure, such as an intergranular defect in
metal or plastic (205). Even a reference

number of a component etched onto a metal surface in a highly stressed
region has been the initial location of a fatigue crack. Such defects
are called stress concentrations or stress raisers in that they raise the stress level above the overall average level in that region (106).
Sharp edges, corners, or grooves in implant components are common sites
of stress concentration. Even defects just within the material, such as
small bubbles in cement, can act as stress concentrations in the same
way. Generally, biologic materials do not experience fatigue failure
because the resting periods between loading episodes allow for
restoration and remodeling. However, one theory for the development of
osteoarthritis has been advanced: healed trabecular microfractures,
failed in fatigue, produce an overly stiff supporting structure to the
articular cartilage.

The stresses are complex in actual structures and can
change rapidly from section to section, especially in the presence of
implants. Combinations of bone and implant are called composite structures. An example of a component that acts in series with the bone is a metal-backed tibial plateau (Fig. 100.17).
It is assumed that the resultant forces on the plastic surfaces are
similar for the all-plastic and metal-backed components. However, at
the resection level of the upper tibia, although the forces transmitted
across this section are the same, the distribution is quite different.
In the case of the plastic, there are high stresses directly under the
contact areas. In contrast, the metal plate spreads the load over the
entire upper tibia. The actual pressure distribution largely reflects
the foundation stiffness of the underlying trabecular bone (154). At an individual trabecular level, the stresses are largely compressive in a vertical direction in the normal knee, but are

varied and complex beneath an artificial knee, depending on the
penetration and closeness of contact of the cement. At the level of the
cortex, below the implant level, the stresses in the prosthetic case
are similar to those of the normal knee.

In the normal intact hip, there are forces on the femoral head H and in the abductors G, such as to produce both compression and bending in the upper femur (34) (Fig. 100.18). This can be depicted at a level just below the lesser trochanter, where there is a resultant compressive force C acting down the center, together with a moment M. Force C produces a mean compressive stress of C/A at the section, where A is the area of bone. Moment M produces tensile stresses on the lateral half of the femur and compressive stresses on the medial half. These are called bending stresses.
These two stress distributions are combined to produce the resultant
stresses. The bending stresses usually dominate over direct stresses
such that the lateral side is still in tension and the medial side is
in increased compression.

In a total hip replacement, a composite structure, the hip stem, and the bone act in parallel. It is assumed that the forces H and G are the same before and after hip replacement. At the section shown in Figure 100.18, compressive force C and bending moment M are now shared between the bone and the stem. For C, the proportions are A_B.E_B:A_S.E_S, where A_B and A_S are areas and E_B and E_S are the elastic moduli of the bone and stem, respectively. For M, the proportions are E_B.I_B:E_S.I_S where I_B and I_S are the second moments of area or section moduli of the bone and stem, respectively. For a circular stem diameter D, I_S = πD⁴/64. For a hollow femur with outer and inner diameters D_o and D_i, respectively, I_B = π(D_o⁴ – D_i⁴)/64. The important effect of diameter can be seen. The product of E.I
is called the bending stiffness. In practice, a hip stem has a much
higher EI value than the bone for a narrow cortical thickness and large
canal diameter. In this case, the proportion of the bending carried by
the stem is high, and the bone can be seriously stress protected. At the other extreme, with a thick cortex and a narrow canal, the bone carries a high proportion of the bending.

Typical angular rotations at the hip joint for a walking cycle are shown in Figure 100.20 (120).
On heel-strike, there is about 30° of flexion, and at toe-off, about
10° of extension. The other two rotations are approximate because of
the difficulty of separating femoral from pelvic motion. However, the
range of abduction to adduction is about 11°, and for internal-external
rotation, the range is about 8°. These rotations produce motions of
individual points on the femoral head that traverse curved paths in the
socket and cross over one another. This aspect is discussed further in
the section dealing with hip-simulating machines.

The forces acting across the normal intact hip have been
determined indirectly by using gait analysis and directly in total
joints by using telemetry (20,21,187).
The latter is more applicable to the subject of this chapter. These
forces have been depicted as acting on the femoral head during
different activities. Figure 100.21 shows that
the line of action of the force during level walking changes much less
than the angles of motion of the femur relative to the acetabulum. The
projection on the acetabelum of these forces during walking results in
a similar pattern but with larger angles of excursion. This means that
the force vector moves somewhat more with the femur than it does with
the acetabulum (74,75). The reason for this is that approximately two thirds of the hip force is produced by the abductors (10), which tend to apply their force parallel to the long axis of the femur (74,75).

The directions of the resultant force on the joint are
important to the function of total hips. For this purpose, it is useful
to consider the forces relative to axes based on the long axis of the
femur. In the frontal plane, the force makes an angle of 15° to 27° to
the long axis of the femur during stance. This produces axial
compression, a varus moment, and a medial-to-lateral force. In the
sagittal plane, there is an important component peaking at
approximately 0.75 BW acting from an anterior to posterior direction on
the femoral head, which results in torsion (21,99).
The latter is considered to be an important contributor to the
compressive failure of trabecular bone in uncemented stems, and results
in stem fractures initiating from an antero-lateral corner.

The peak forces during a range of different activities are shown in Figure 100.22.
The exceedingly high force of 8 BW encountered accidentally in a
stumble is important because a single force such as this, along with
its torsional component (21), could lead to debonding at the implant–bone, cement–bone or cement–stem interfaces, or to cracking of the cement mantle (22,150,210). A factor that influences the mechanics of the hip is the offset of the femoral head from the femoral axis (167,184).
This is frequently reduced from normal in a total hip replacement
(THR). The effect is an increase in the force required in the abductors
(10), leading to a higher resultant joint
force, a more vertical resultant, and sometimes a gait abnormality. An
increase in offset reduces forces but causes an increase in the bending
moment on the stem. Failure to achieve the

normal anteversion of the head and neck increases the axial torsion, which is undesirable (21).

Cemented femoral components have changed relatively
little since the original Charnley design, although paradoxically, even
a small change in design can result in a major change in performance.
Fundamentally, the stem shape parallels the canal shape (167),
leaving space for a cement mantle of 2 to 4 mm in thickness. In early
designs such as the Charnley, Exeter, Stanmore, Muüller, and T28, the
cement was not intended to bond to the stem. In contrast, at the
cement–bone interface, these designs aimed for an intimate mechanical
contact with rough bone or trabecular bone. Compressive, shear, and
tensile stresses are transmitted to the bone surface (113), an unnatural situation, but one that is tolerated for long time periods.

It is well known that the stresses in the femur are
changed from normal due to the presence of the stem, especially
proximomedially, where the longitudinal compressive stresses are
reduced to 20% to 30% of normal levels (110,223) (Fig. 100.23).
Progressing distally, the stresses become closer to normal, reaching
normal levels below the level of the stem tip. The reduction in a
stress parameter, such as the Von Mises stress, particularly when it
results in a reduction of bone density or volume, is called stress protection or stress shielding (30).
Over a time period of approximately 5 to 10 years, this results in
longitudinal removal of bone at the neck cut level, osteopenia, and
resorption of bone away from the cement. If this process becomes
extreme, the interface stresses around the remaining stem become
excessive (150), leading to “clinical” loosening (94).
In addition, wear particles can be more easily transported down the
interface, producing lysis at locations all around the stem, especially
distally (98). A further consequence of
proximal bone loss is that the stem is only rigidly fixed in its distal
half, increasing the bending stresses and increasing the possibility of
fatigue fracture (94).

Dual energy x-ray absorptiometry (DEXA) scanning has now
become the standard method for measuring and monitoring changes in bone
density over time (201,202).
In this technique, in the frontal plane, for example, an x-ray beam is
scanned across the field, and measures of x-ray absorption are made,
giving a measure of bone mass at a matrix of points in the field. Major
changes in bone mass usually occur slowly, and it has been found that
bone loss may be only 10% to 20% at 2 years, but that this loss can
increase to as much as 50% in the proximal-medial region at 5 years.

The allowable level of stresses at the interface are
known only approximately, but the goal is to minimize these stresses
and to avoid local stress concentrations. Stresses within the bone
should be as close to normal as

possible.
Cement and stem stresses, especially tensile, should be minimized and
should not exceed the fatigue limit under normal loading conditions.
All of these goals cannot be obtained simultaneously, but a number of
guidelines for hip stem design exist:

Concerning the ideal surface finish for a stem,
long-term clinical data suggest that a smooth surface or a near-smooth
surface, in which there is no direct stem–cement bonding, produces
successful results. The use of rough surfaces, or rough surfaces with a
precoating of cement to obtain stem–cement bonding, has not been
uniformly successful. It appears that shear and tensile stresses at the
interface can cause progressive debonding and interface micromotion (150).
The latter can then generate metal and cement debris, which enter the
joint space, followed by accelerated wear of the UHMWPE liner. Evidence
of stem–cement micromotion in all types of stem is frequently seen in
retrievals and produces fretting wear of both stem and cement (Fig. 100.24).

The stem material for cemented components is important,
and forged cobalt-chrome or stainless steel is usually preferred.
Titanium alloys have been prone to surface wear or even crevice
corrosion and are not in general suitable for cemented application. The
ideal combination for a cemented stem with a modular head is to use
cobalt-chrome alloy for both owing to the minimal corrosion on the
taper connection (31) and reduced scratching of
the femoral head over time. Other head materials are discussed later in
this chapter. Because of its success in elderly patients, and even in
younger patients (12,71,145)
(although studies are variable on this issue), it is unlikely that the
design of the standard cemented stem for primary cases will change
substantially in the foreseeable future. Long-stem cemented components
are used in revisions, but the cement penetration and shear strength
are likely to be inadequate owing to the loss of most of the cancellous
bone from the previous implant (146).

Standard cemented acetabular components consist of a
solid UHMWPE hemisphere, or just greater than a hemisphere, with
grooves on the outer surface for keying to the cement. A metal wire is
usually embedded on the outside to measure the wear relative to the
femoral head on radiographs. The range of motion (ROM) between the
femoral neck and the socket before impingement occurs is important
because it affects the potential for dislocation as well as loosening.
It is notable that, as wear proceeds, the ROM steadily reduces, and
this process can lead to problems for 22 mm heads in the long term. The
factors affecting ROM are:

Overall, the best compromise appears to be a head
diameter in the range of 26 to 28 mm with an elliptical neck of minimum
diameter for strength. In practice, the most widely used diameters are
28 mm and 22 mm. To obtain sufficient plastic thickness the minimum
outer diameters of the acetabular component recommended for a 28 mm
diameter head are 44 mm for plastic and 48 mm for metal-backed plastic.
Below those diameters, a 22 mm head is recommended.

The principles of acetabular fixation (158)
are the same as those for stems, notably intimate contact of the cement
with a rough bone surface and penetration within the trabeculae. There
is still uncertainty regarding removal of the subchondral plate (161) and the size and number of key holes (Fig. 100.25).
Although retention of a subchondral plate is attractive for force
transmission, areas of relatively smooth bone are prone to interface
micromotion, bone resorption, and particle ingress. The loosening
mechanism of cemented acetabular components is that the peripheral
regions become infiltrated with UHMWPE debris, leading to bone
resorption, followed by a creeping process that eventually involves the
entire interface (192). This process has been
conceptualized as a progressive increase in the “effective joint
space.” Mechanical factors are also likely to play a role due to higher
than normal stresses occurring in the trabecular bone (64).
Steady migration of sockets into the acetabular bone is common, and
acetabular component failure usually occurs at a higher rate than for
stems in follow-up examinations after more than 10 years. The use of
metal backing (115), which reduces the
incidence of interface radiolucency in tibial components, has not been
successful in cemented acetabular components. Simple finite element
models indicated a more uniform stress distribution across the
interface, but more complex 3-D models, which included the exact
shapes, bone densities, and forces acting (64),
have shown otherwise. This has highlighted the care needed in
formulating and interpreting finite element models and the conclusions
that may be drawn from them.

In the 1950s and 1960s, thin polished metal
hemispherical shells were interposed between reamed surfaces of the
femoral head and acetabulum in the expectation that new cartilaginous
surfaces would form between the metal and the bone. The tissue
formation, however, was unpredictable, consisting of localized islands
of fibrocartilage and hyaline articular cartilage. There may still be
possibilities for this type of approach if the type of tissue formation
can be controlled to a greater degree. The Austin-Moore and similar
hemiarthroplasties, generally with rectangular sectioned stems, were
moderately successful with regard to fixation and pain relief, although
most of the patients had limited functional requirements. Subsequently,
many such uncemented stems with “satin” or smooth interface surfaces
have been introduced in total hip stems, but have all been subject to
bone resorption, interface micromotion, and migration, leading to
unsatisfactory results.

Successful fixation of uncemented components depends on
achieving tolerable stresses at the implant–bone interface and
minimizing interface micromotion to approximately 50 µm or less over
most of the interface. These conditions depend on the surface of the
stem, the sectional shape, and the overall geometry. The major factors
are

It is often assumed that an uncemented stem will cause
more proximal stress shielding than a cemented stem owing to its larger
cross-sectional area (38,80).
Laboratory tests have shown that this is not necessarily the case
because the stem becomes tightly wedged in the femur, producing high
circumferential tensile stresses as well as up to 50% of the axial
compressive stress component (226). Where
proximal bone ingrowth has been achieved, bone has been well preserved
over time, based on DEXA-scanning data. However, serious proximal bone
loss has occurred under the following conditions:

Uncemented stems have an important application in
revisions. The major goals in the design of a revision stem are to
maximize axial and torsional stability, and to preserve or enhance the
remaining bone. To achieve these goals, the following features are an
advantage:

Because of the extreme range of conditions of the bone,
as well as patient factors, a graduated implant system is logical,
basing the design decision on a hierarchy of factors (Fig. 100.27).
Modular systems using separate proximal “plugs” and distal stems, or
using a hollow proximal sleeve with a stem that locks within it (as in
the S-ROM [DePuy-J&J, Warsaw, IN] system) (45)
are useful approaches in dealing with these variations in an
off-the-shelf system. A fragile or detached greater trochanter is
problematic and can be held by hooks or wires. Circumferential bone
wiring reduces hoop stresses on surgical insertion but may cause stress
concentrations in future years. The most difficult revision situation
occurs when the diaphysis below the level of the failed stem is thin.
Here, the choice is between impaction allografting (91),
with a risk of migration, or a very long uncemented or cemented stem
affixing in the femoral condyles. The long stem can even be rigidly
attached to a knee replacement in those cases in which both the hip and
knee need revising or replacing.

From a scientific point of view, a strong case can be
made for providing a custom hip for every case, provided the surgery is
suitably precise (14). For a cemented hip,
there must theoretically be an ideal thickness and shape of cement
mantle and an ideal stem length. For an uncemented hip, there must be
an ideal fit and stem length to minimize interface micromotion (109) and produce the ideal combination of interface and bone stresses (110,232).
In all cases, the ideal position of the femoral head can be obtained.
However, the issue is complex, and separate considerations apply to
cemented or uncemented, primary or revision (49,131). Some of the factors that are noted in the argument against the widespread adoption of custom stems are as follows:

However, even accounting for the above-mentioned
problems, there is still a justification for using custom hips in
primary cases and even more so in revisions. It has been shown that
very low rates of loosening, even superior to that obtained with
cemented stems, can be achieved by using fully porous-coated or
HA-coated stems that are fixed tightly into the diaphysis by
appropriate reaming. However, proximal stress shielding is an
undesirable consequence with this system. If the more preferable
proximal stem fixation is required, the fit requirements are more
stringent. In this situation, custom hips have an advantage. There is
considerable empirical data that interface micromotion is the major
cause of pain and bone resorption, and laboratory data have shown that
micromotion can be minimized by implants with a close fit to cortical
bone (46). Hence, in a large series of cases,
custom stems should reduce the incidence of pain and interface
radiolucency compared with an off-the-shelf design with the same design
features (Fig. 100.28). Even if an
off-the-shelf hip is used for the more “normal” anatomies, custom stems
may still be used for abnormal situations. For revision hip
replacements, the variations in the proximal shape after removal of the
original stem, the diameter and curvature of the diaphysis below that
region, and the required location of the femoral head are so great that
a custom approach has a strong justification. Indications for custom
hip replacements are:

In order to design and manufacture a custom hip stem (223), accurate determination of the 3-D canal shape is required (160). The most accurate method is by the use of CT scans (116,184).
A number of commercially available software packages can be used to
contour the inside and outside cortical bone boundaries, from which a
3-D model of the bone can be generated. The internal contouring
requires definition of a suitable Hounsfield number (a value signifying
bone density, where 100 equals water and 1000 equals cortical bone)
distinguishing the boundary between cancellous and cortical bone. For
purposes of reaming and rasping, a value of 500 to 600 has been shown
to be suitable. Another method for determining the shape of the femoral
canal that is more direct and less expensive but is restricted to
femurs of relatively normal shape is to use scaled AP and ML
radiographs (119). The canal outlines are
digitized and then a 3-D computer model of the “average femur” is
numerically distorted so that its canal outline fits that of the
specific femur. This method has been shown to be accurate to better
than 1 mm in the regions where a close stem fit is required.

Given the shape of the femoral canal, special software
is needed to design the stem to a particular scheme. For example, it
can be assumed that distal reaming and proximal rasping are performed
to produce a more geometric shape for ease of manufacture and surgery.
The general principle is such that the major load transmission occurs
proximally (110), which results in the following ideal features:

For economy, convenience, and speed of manufacture of
custom hip replacements, prewritten computer numerically controlled
(CNC) software is required. This software provides instructions to a
milling machine to produce the stem from a bar of preformed material.
The end result is an integrated software package that covers both the
design and manufacturing stages.

One approach for assessing the relative value of a
custom hip system compared with an off-the-shelf system is to assume
that accuracy of fit is the single criterion for comparison, and then
to determine how many off-the-shelf

sizes
would be needed to fit the general population of femurs to a given
accuracy. This requires that the most ideal set of sizes for the
off-the-shelf system be synthesized. The problem can be solved in the
following way:

If such a process is carried out, data such as those shown in Figure 100.29
can be produced. If an accuracy of fit of 0.5 mm or better between the
stem and the bone were required, then a 40-sized system would achieve
this in just under 50% of all cases. However, if a 1 mm accuracy were
satisfactory, a 20-sized system would deal with more than 80% of all
cases. Virtually all cases could be dealt with using a 10-sized system
if only 1.5 or 2 mm accuracy were required.

By taking an entirely different approach to hip
replacement, namely surface replacement, the problem of fit is greatly
simplified. Such an approach can be classified as a conservative hip,
which can include several forms. A conservative hip is one that
involves fixation to the femur and acetabulum such that its removal for
any reason would allow the placement of standard primary hip components
with little compromise. Early examples of conservative hips are the
Aufranc cup arthroplasty and the various femoral head replacements with
short pegs, such as the Judet (212). From the
1970s to the present, various surface replacements have been attempted
using combinations of metal, ceramic and polyethylene for the
components, and press-fit, cement, porous, and HA for fixation. There
have been several problems with surface replacement to date:

Other possible conservative hip designs are shown in Figure 100.31.
A modification to a standard surface replacement is Townley’s TARA
design. This design removes most of the head, greatly reducing the
trabecular remodeling problem, whereas the short stem improves
stability. Thrust-plate designs (e.g., Wiles, Huggler and Jacob) have
met with some success in clinical follow-up, and further improvements
to this type are possible by using surface textures and coatings for
osseointegration. Advantages of this configuration are that good
initial stability can be obtained to allow for osseointegration, and a
normal modular head can be used. A key issue is that a proportion of
the varus bending is transmitted by the neck collar and by the lateral
plate because this has important remodeling consequences. Variants of
the thrust-plate design use different fixation at the neck resection
level, within the neck itself, and at the lateral cortex. FEA of hip
stems with a lateral flare indicates that, for a high neck cut, the
stem carries only a small bending moment, allowing its length to be
shortened considerably. This provides the possibility of a “proximal
plug” type of conservative hip. In this type, osseointegration can
occur over a large surface area. Preserving part of the femoral neck
for bone support is used in the Mayo conservative hip design (Zimmer,
Inc., Warsaw, IN) and in the Pipino hip. Other conservative hip schemes
include metal shrouds that cover regions of the outer proximal femur.
Such designs could have application for abnormal femoral geometries and
even for some revisions. From the acetabular perspective, those
conservative designs that incorporate a femoral

trunnion to use a standard modular head have an advantage for restoring the required head offset.

Methods used to characterize hip stems and to compare
stems with each other can be divided into laboratory studies and
clinical evaluations.

The general conception of normal knee motion is usually
derived from the measurements of “passive motion” of knee specimens
when the external forces are small (228). The
tibia has either been passively flexed and held at successive flexion
angles, or a small flexion moment has been applied, balanced by a force
in the quadriceps (97,183).
The motion itself has been described in several different ways, which
has led to a number of misunderstandings. The “contact points” are
defined as the common tangent of the femoral and tibial outlines, as
seen on radiographs (although in reality only the bone outlines and not
the cartilage outlines can be seen). These contact points displace
posteriorly, with the lateral contact point moving much more than the
medial (134,183) (Fig. 100.34).
If the actual contact areas were plotted, however, although the areas
would follow the same trend as the “points,” it would be difficult to
define a point directly from the areas because the areas are of
irregular shape and occur on both the cartilage and the menisci (2).
Motion can also be described by defining axes in the femur, the
transverse axis being through the centers of the lateral and medial
posterior femoral condyles, or the epicondylar line. If the motion of
the femoral origin at the center of the femur were plotted as a
function of flexion, it, too, would move posteriorly with flexion,
whereas the transverse x-axis would tilt slightly. This rigid body
motion would not be the same as that determined from the contact
points. For example, in a total knee replacement (TKR) seen in the
sagittal view, it can readily be observed that if the tibial surface is
dished, the contact point motion is much greater than the rigid body
motion.

The passive motion of the knee can now be summarized in rigid body terms, with the axes as defined in Figure 100.34.
As the knee is flexed from 0° to 120°, there is a posterior femoral
displacement and internal tibial rotation about its y-axis. There is
also a valgus rotation of the femur as the lateral femoral condyle
moves down the slope of the tibial plateau. This motion is governed by
the shape of the condylar surfaces, the different mobility of the
menisci on the lateral and medial sides (219), and the locations and stiffnesses of the cruciate and collateral (especially the medial) ligaments (5,41,88,89,183,199,207,241). This “passive motion” of the knee is useful for describing a middle path of knee motion for reference purposes.

Three effects will change this motion path considerably:

In functional activities (7,73,132),
the knee motion occurs within the envelopes of laxity. An infinite
number of motion paths is possible; these paths are likely to vary even
between successive steps (164). The major factors that determine the actual motion path are the external forces and the muscle forces (138).
Whereas the principal force direction across the knee are along the
long axis of the tibia, the combination of the applied forces results
in shear forces and torques (Fig. 100.36).
These force components have been calculated using measured values of
the foot-to-ground force, the limb segment kinematics, and
electromyography (EMG) data of muscle action, input to

a knee model (136,155,156 and 157). In a recent study, telemetry data from an instrumented distal femoral replacement was used (206).
During walking, there are three peak forces of 2 to 3 BW. The shear
forces act in both anterior and posterior directions, with posterior
being predominant, in a direction so as to tense the PCL. The torque is
predominantly internal, the effect being to move the lateral tibial
plateau forward. The forces do not increase dramatically for ascending
and descending stairs, the largest increase of about 15% in the
compressive force occurring in descending stairs (156).
The patellofemoral forces are just less than half BW in walking but
reach approximately 1 BW or more in rising from a chair or climbing
stairs. The axial torques in level walking and on stairs are in the
range of 6 to 8 N·m.

The most obvious form for a TKR is a replacement of the
bearing surfaces using metal and plastic components, positioned so as
to restore the neutral alignment of the knee and achieve the correct
tensions in the ligaments (121). The simplest
form is a compartmental type (a “uni”), which is applicable if the
patellofemoral surfaces do not require replacement (55).
Most often, only the medial side is required, but replacement of both
sides as an alternative to a TKR has been used extensively in some
centers. Some of the issues concerning compartmental knees are the
following:

On the basis that the AP movement of the contact point
on the medial side is small, it appears logical to provide a dished
tibial surface, which would alleviate a number of the above problems.
Improved designs and instrumentation, coupled with small incisions,
present an important opportunity for outpatient surgery in the
appropriate indications.

One-piece tibial and femoral components covering both
compartments are able to provide stable fixation, are self-aligning,
and incorporate the patellofemoral surfaces; for these reasons, such
designs are the most frequently

used.
A fundamental principle of condylar replacement is that the artificial
surfaces are designed such that, in combination with the remaining
joint structures, the laxity and stability characteristics are similar
to those of the normal intact knee (144). In general terms, if all of the ligaments are preserved, the tibial surface requires only shallow dishing (Fig. 100.37).
If the ACL is absent, posterior curvature is required (although in most
designs there is insufficient curvature to compensate for the ACL). If
both cruciate ligaments are absent, posterior and anterior curvatures
are required (85).

The AP shear force is shared between the cruciate
ligaments and, to a lesser extent, other ligaments and soft tissues,
and the reaction forces at the joint surfaces (19). The ACL is more tense in early flexion, and the PCL is more tense in high flexion (183).
The fraction of the shear force carried by the cruciate ligaments
depends on the tibial surface geometry, the location of the “bottom of
the dish,” and the placement of the components. Even modest dishing of
the tibial surface (e.g., sagittal radius, 70 to 80 mm) provides
considerable stability, especially when high compressive forces of 2 to
3 BW are exerted (144,214).
In TKRs in which the PCL is preserved, the fraction of the shear force
carried by the PCL depends on the curvature of the tibial surface, the
curvature of the interfacing femoral surface (which depends on the
angle of flexion), the location of the bottom of the tibial dish, the
tightness of the PCL (which depends on the surgery), and the angle of
flexion (89). In practice, the PCL, especially
the anterior fibers, carry an increasing proportion of the anterior
shear force with flexion (129). The different schemes for condylar TKR design have advantages and disadvantages:

The basic principles of the stability and laxity of
dished condylar surfaces, embodied in the original Total Condylar
(Howmedica, Rutherford, NJ) (213,215,228) and in a number of other later designs, is shown in Figure 100.37. If there is a compressive force V acting down the long axis of the tibia, the femoral component locates at the bottom of the tibial dish. If a shear force S is superimposed, the femoral component displaces anteriorly such that the reaction force at the contact point exactly balances V and S.
The contact point displacement is greater than the rigid body
displacement. A torque applied to the tibia is equilibrated by an
anterior displacement of one side of the knee and a posterior
displacement of the other. The torque Q carried by the surfaces is S × BS, where BS
is the bearing spacing. The total condylar type of design thus provides
both stability and laxity, and has the advantages of moderate
conformity for low wear, simplicity, and ease of surgery.

An interesting variation of a standard total condylar
concept is an evolution of the Freeman and Freeman-Samuelson (Protek)
designs (86). The medial side is almost fully
conforming, whereas the lateral side has low conformity. This has the
potential advantages of achieving physiologic motion and reducing wear
on the more heavily loaded medial side.

In coupled motion, when AP displacement and internal-external rotation are combined, the laxity curves of a TKR are complex (Fig. 100.38).
In neutral rotation, an AP laxity test, in which an anterior shear
force is applied to the femur followed by a posterior shear force,
produces anterior and posterior femoral displacements. As shown in Figure 100.38,
when there is greater conformity anteriorly, the anterior femoral
displacement is less. When rotation is superimposed, one femoral
condyle moves up the posterior curve. Because of the greater anterior
conformity, the movement up this curve is less than that at the

posterior
and, hence, there is a posterior shift in the neutral femoral position.
When the AP laxity test is now carried out, the AP displacement is
reduced. The more rotation that is superimposed, the less is the AP
laxity and the more is the shift in the neutral position. If the femur
is now flexed, the smaller posterior femoral radius is now in contact
with the tibia, and the laxity is reduced. However, as shown in the
geometric drawings of the sagittal femoral outline, the neutral
position of the femur is now shifted anteriorly.
These shifts in the neutral position, which are primarily dependent on
femoral geometry, have an important effect on soft-tissue tensions and
kinematics.

Another important effect on kinematics is friction between the femoral and tibial surfaces (188) (Fig. 100.39).
The coefficient of friction for metal on polyethylene surfaces is 0.05
to 0.1. For a condylar prosthesis implanted in a knee, without
friction, the laxity curve would be strain stiffening (Fig. 100.14)
with a hysteresis loop. However, owing to friction when a compressive
force is acting and the shear force direction is reversed, as occurs
during the gait cycle, there is a “stick” period without motion. In
function, this will result in some erratic behavior of the motion, with
combinations of rolling, sliding, and stick.

Based on the above-mentioned factors, it can be seen
that the geometric parameters of a TKR play an important role in
determining function and durability. The geometric considerations can
be broadly categorized as being in the sagittal plane or the frontal
plane.

In the total condylar type of knee prosthesis, stability
is provided by the articular surfaces and the retained ligaments. The
motion of the femorotibial contact point is hence determined by these
factors, in response to the external forces and the muscle forces. As
indicated previously, designs with shallow tibial surfaces and
preservation of both cruciate ligaments have the theoretical capability
of reproducing normal kinematics. Fluoroscopic techniques have been
used to record relative tibial and femoral knee motion during a range
of activities. The method involves sequential high-resolution
fluoroscopic images obtained in the sagittal plane. If the 3-D geometry
of the femoral and tibial components is known and input to the
computer, then the computer image can be moved into a 3-D orientation
so as to match the radiographic image. In this way, the relative 3-D
position of the femur on the tibia can be determined throughout the
activity. The accuracy of such methods is now better than 1 mm and 1°.
Interestingly, this method was originally developed for the military to
identify high-flying aircraft.

Fluoroscopic studies of TKR patients (13,68,204) have shown the following:

There is thus a rationale for designing condylar knees
that control relative femorotibial motion in a more predictable way, at
least during part of the flexion range. One scheme for controlling the
kinematics is by the use of intercondylar cams.

Two cam types have been used since the 1970s. The first
is embodied in the Kinematic (now the Kinemax) Stabiliser (Howmedica,
Rutherford, NJ) (78) (Fig. 100.41),
which consists of cam surfaces that are in contact throughout flexion
and produce a progressively posterior movement of the femorotibial
contact point with increasing knee flexion. The second type is used in
the Posterior Stabilised Knee (Zimmer, Warsaw, IN), in which the cams
interact only after about 70° flexion (233), thereafter producing a large posterior displacement of the contact point. The aims of these particular designs were to:

Important mechanical criteria concerning the cam design are as follows:

However, if cams or guide surfaces, in combination with
the condylar bearing surfaces themselves, are to provide ideal knee
kinematics, the following criteria can be added:

The rationale is to provide optimal muscle lever arms
and soft-tissue tensions to restore optimal function as closely as
possible. By using special computer programs (225), the designs shown in Figure 100.41 and other types of intercondylar guide surfaces with different patterns of motion control can be synthesized (Fig. 100.42). Another example is the Superstabiliser-CCK type (137,191,213).
(Superstabilizer is a product of Howmedica, Rutherford, NJ; CCK is a
constrained condylar knee manufactured by Zimmer, Warsaw, IN.) This is
designed primarily to carry varus-valgus moments, and hence, the
computer program is used to maximize the area and strength of the
plastic tibial post. In addition, it can be seen that the interaction
between the femoral and tibial guide surfaces produces anterior femoral
displacement toward extension and posterior displacement toward full
flexion.

Another useful configuration occurs when the femoral
guide surface is an eccentric circle and the tibial guide surface is a
saddle (225). This can provide AP motion control in different parts of the flexion range (Fig. 100.42).
By defining and varying geometric parameters of the above-mentioned
types, different AP control patterns can be achieved. In another design
using intercondylar guide surfaces (66), the
saddle moved the contact from anterior to posterior during flexion, the
second part of flexion using a ball-in-socket. In another design (76), the saddle shape was such as to induce pure rolling during flexion and extension.

Other designs have been introduced in which anterior
femorotibial contact in early flexion and posterior contact in late
flexion have been achieved using separate pairs of bearing surfaces.
The TRAC (Biomet, Inc., Warsaw, IN) uses intercondylar guide surfaces,
shown at the lower left of Figure 100.42, to transfer the contact from the first pair of bearing surfaces to the second. The Bi-Articular

(Kyocera, Kyoto, Japan) uses conventional total condylar bearing
surfaces until about 90° of flexion. On further flexion, intercondylar
surfaces of spherical shape, which are posteriorly located, interact.
The goals are maximum flexion and freedom of internal-external rotation.

The term “meniscal bearing” is used for a plastic bearing used on one side of the joint (92)
(usually medial). A “mobile bearing” is a plastic bearing that includes
both the medial and lateral compartments. A “mobile bearing knee” has
come to mean any unicompartmental or total knee replacement
incorporating a meniscal bearing or a mobile bearing (37). The purposes of a mobile bearing knee (Fig. 100.43) are to minimize the wear of the plastic and to allow relatively unrestrained motion to occur between the femur and the tibia.

In concept, the components model the scheme of the
natural knee such that nonconforming main bearing surfaces, which allow
freedom of motion, have menisci interposed to increase the contact area
(2,219). The
deformability of the menisci allows for changing femorotibial geometry
during flexion while still maintaining a large contact area. The
constraint of a mobile bearing knee compared with a natural knee is
less on the medial side and more on the lateral side.

There are two schemes for the femorotibial bearing surfaces:

Until more data are available from knee simulator
testing, it is not possible to assert which scheme will produce the
least wear and plastic damage over the long term. However, surface
wear, pitting, and delamination wear must all be considered in any
comparison.

Concerning the design scheme of the meniscal or mobile bearings, there are four fundamentally different arrangements (Fig. 100.43 and Fig. 100.44):

Additional configurations of the above-mentioned schemes
are possible to achieve roll-back with flexion and even roll-forward
with extension. A posterior stabilized arrangement can be used, but
this would require partially conforming femorotibial bearing surfaces.
Guide surfaces, such as those described earlier, can interact between
the femoral component and the metal tibial baseplate. This is
applicable to either partially or fully conforming bearing geometry.

Comparisons can be made between the fixed and mobile bearing knee types:

A final point is that (at the time of writing) there are
no reliable data on the wear rates, and the types of particles
generated as a function of contact area and contact pressure. Hence,
comparisons between different designs must be determined by simulator
tests rather than automatically assuming that larger contact areas are
an advantage, as has been commonly done in the last decade.

Over the years, numerous different linked designs have
been introduced, which reflects more the opportunity for inventive
design rather than extensive need. The largest numbers continue to be
used in Europe, whereas in the United States, the tendency is to use
the most conservative design whenever possible, even if substantial
soft-tissue balancing is required. The characteristics of a linked
design are:

The most conservative type of linked TKR is the intercondylar hinge (Fig. 100.45). The linkage is housed in the intercondylar region, preserving the condyles on each side (9). Versions of this type include the Attenborough (9),
the St. Georg Endo (Waldemar-Link, Germany) model, the Sheehan
(Howmedica, Rutherford, NJ), the Rotaflex (JJ Orthopaedics, Warsaw,
IN), and the PFC S-ROM

(J&J
Orthopaedics, Warsaw, IN). The bearing surfaces can be extended to the
full width of the knee, an advantage for reducing varus-valgus bending
moments on the linkage. The patellofemoral resurfacing has been absent
in some designs but ideally should be included. Advantages of the
intercondylar design are:

Disadvantages of the intercondylar design are:

The least conservative type of linked TKR, in terms of
bone resection, is the fixed hinge or rotating hinge. Placement is
achieved by resection of about 25 mm from the distal femoral condyles
and 10 mm from the upper tibia. An axle is then used to connect the
femoral and tibial components, usually with plastic bushings to act as
the bearing. The total thickness is determined by the required
dimensions of the axle and bushings, and the ideal placement of the
axle, which is close to the epicondylar line. A lower, or more
posterior, axle location will reduce bone resection but result in
abnormal tracking of the patella. It should be noted that the most
important requirement in the placement of hinged replacements is that
the patella locates at the correct level on the femoral flange with the
knee in extension.

The fixed hinge is the simpler design and can be used
for patients of low demand who would not overstress the fixation and
for whom flexion-extension motion is sufficient. Examples are the
Guépar (Benoist-Gerard), the St. Georg (Link), the Blauth (Allopro),
and the original Stanmore. The rotating hinge, however, results in a
more “natural” feel to the patient and can be more durable in the long
term. Examples are the Kinematic and Kinemax Rotating Hinges
(Osteonics-Howmedica, Rutherford, N.J.), the Finn (Biomet, Warsaw, IN),
the Noiles (J&J Orthopaedics, Warsaw, IN), and the SMILES (Stanmore
Implants Worldwide). More bone resection, however, is required to
accommodate the extra bearing surface. The rotation can be achieved by
a flat polished metal surface pivoted on a flat plastic surface, or by
a convex metal surface in a dished plastic surface. The convex metal
surface in a dished plastic surface is preferable because it provides a
“soft” limit to rotation, reducing the possibility of instability or
patellofemoral subluxation. Some medium- to long-term follow-up studies
of linked designs have shown durability comparable to that associated
with condylar replacements (32). That fact,
coupled with the relative ease of surgery and the immediate stability,
makes linked designs an attractive option for selected patients.

The function of the patellofemoral joint in the normal knee has been studied from many different aspects (3,4,7,103,111,117,163,200). The shape of the patella prosthesis surface can be either a dome, a rounded cone, or a gaussian curve (Fig. 100.46). The dome has been widely used in total condylar type designs but has several drawbacks:

The cone (Fig. 100.46) has
increased plastic thickness at the sides and larger contact areas in
the form of two “lines” rather than “points.” Finally, the gaussian
shape further increases both the thickness and the contact area. In all
of the designs, a medial offset to the peak is an advantage and
eliminates the need to use a smaller size of symmetric design and
medialize it. A rotating platform design produces the largest area of
contact and potentially the least wear and deformation, at the expense
of some extra thickness due to the metal backing and a second bearing
surface. An additional factor in area of contact is the shape of the
intercondylar cutout on the femoral component, as shown in Figure 100.46.

Regarding stability, there is a perception that the dome
is more forgiving in alignment and that the other types are susceptible
to tilting and loading on the corners. In the front view, it can be
seen that, owing to the angle between the lines of action of the
patellar ligament and the quadriceps (the Q-angle), there will be
resultant tensile and lateral shear forces on the patella itself (Fig. 100.47). The actual direction of the forces depends on the relative forces in the different parts of the quadriceps (3,4,103,117).
In addition, there will be a compressive force between the patella
(whether the natural patella or a plastic replacement) and the patellar
flange, which increases with the flexion angle up to about 100° flexion
(4). A 3-D force analysis can be carried out, based on the sectional view shown in Figure 100.47. If the resultant compressive force is central, equal resultants act on the lateral and medial facets (Fig. 100.48).
However, in general, the lateral reaction force will be larger than the
medial owing to the laterally directed component of the quadriceps and
patellar tendon forces. As the Q-angle increases, the lateral force
will increase and the medial force will decrease. However, as long as
both forces are positive, the patella will be stable. At the point
where the medial force reaches zero, the patella will become unstable
and will be subject to tilting and subluxation. This is calculated to
occur at a Q-angle of approximately 12° (117).
Except in cases of extreme valgus, this is unlikely, and hence the cone
and gaussian patellar components are likely to be stable.

The methods used to attach TKR components to the bone
include cementation, press-fit fixation, porous-coating, and fixation
with various macroscopic surface finishes or mesh. Any of the
uncemented modes can include HA or other bioactive coating. The goals
are to minimize interface micromotion (17,220), avoid interface bone resorption, and maintain intimate bone contact (170).
An alternative goal, potentially achievable with press-fit fixation, is
to obtain a stable interface with a thin fibrous tissue layer or
velour-type material (154,224).
In cementing at surgery, if the cement is applied in a viscous state
with little pressure, the penetration will only be on the projecting
trabecular tips, allowing distraction to occur when local tensile
stresses are exerted (Fig. 100.49). In
contrast, penetration of 2 to 4 mm, such that transverse trabeculae are
engaged, results in small interface micromotion and an interface that
can remain stable for long time periods. The most uniform penetration
over the entire surface is achieved with cement applied by multiple
applications of a gun nozzle.

Fracture of the cement mantle has not appeared as a
problem in metal-backed components but has occurred with all-plastic
components when there has been deformation due to poor or uneven bony
support or bone overload from malalignment (178).
Components with coatings, for which long-term fixation relies on
ingrowth or apposition of new bone, require initial fixation that is
sufficiently rigid for at least 2 to 3 months. Screws have been found
to be an advantage for tibial components, whereas for femoral
components (197), a tight AP fit has been
adequate. However, the use of screws carries the risk of screw-tray
fretting and of providing a track for wear debris particles, which
produce osteolytic cavities. Press-fit fixation requires suitable
macroscopic features and pegs to maintain a sufficiently low level of
interface micromotion (221). In many cases, a
stable fibrous tissue interface above a new subchondral layer has been
formed, but so far, press-fit fixation has shown a higher incidence of
loosening and pain than cementation.

Fixation also depends on the design of the component and the loading on the bearing surfaces (105,220).
These factors, in turn, directly affect the stress distribution over
the entire interface as well as the stresses in the bone itself (Fig. 100.50).
In the sagittal plane, if the contact is less than approximately 25%
from the edge, there will be excessive bone stresses beneath the load
and tensile stresses opposite to the load. The actual pressure
distribution at the interface reflects the foundation stiffness of the
trabecular bone, whereas the magnitude of the stresses depends on the
surface area of the component. The bone stiffness is modified in
arthritic conditions, usually by increased and decreased stiffness on
opposite compartments. CT studies show that, after realignment and
insertion of a TKR, whereas the strong bone reduces in strength toward
normal, the weak side gains in strength more slowly (118).
Except in conditions of extreme edge loading, metal-backed components
produce a more normal pressure distribution at the interface. In
all-plastic components, there is a high pressure region at the
interface beneath the contact area. Nevertheless, there is clinical
evidence that plastic components with sufficient thickness (e.g., 10
mm) and a strong central peg are viable in patients with low activity
levels and when there are no major bone defects or regions of
osteoporosis (178). Regarding overall pressure
distribution, central posts reduce the maximum interface stresses for
offset loading, such as varus (179), as well as being effective against shear forces (Fig. 100.51).
Design features located toward the periphery, such as pegs or keels,
especially when they are embedded in hard cancellous bone, are
effective against internal-external rotation (221).

An invaluable method for the measurement of component fixation in vivo has been RSA (52,105,166,186,196).
One millimeter tantalum beads are implanted in the component and in the
adjacent bone at surgery. Biplanar radiography at specific time
intervals, with mathematical analysis, produce the component–bone
movements or migrations over time. Continuous migrators after 1 to 2
years are predictive of future clinical loosening.

Cases with bone loss, including revisions, need special consideration (Fig. 100.52). Augments, such as spacers and wedges, are useful for replacing bone defects and for accurately reproducing the joint line (35).
These augments can be screwed or cemented to the main components. For
larger defects, space fillers made from metal or plastic are an
alternative to bone grafting. Stems are useful for bypassing cavitary
defects, for protecting against fractures, for traversing fracture
sites or serious cortical defects, or for stabilizing allografts (179).
When the TKR design carries varus-valgus moments, intramedullary stems
are needed. In older patients, cemented stems are usually preferable,
but uncemented fluted stems should be used if there is sufficient
cortical thickness, in order to reduce the stress-shielding of the
cancellous bone near the joint (209).
Empirically, suitable stem lengths for Superstabiliser-CCK types of TKR
are 100 to 120 mm, and 120 to 150 mm for rotating and fixed hinges.
Revision of a failed cemented stem requires a new stem at least 50 mm
longer. In all cases, attention is needed to prevent the stem tip from
impacting the cortical wall, which frequently produces osteolysis,
penetration, and even bone fracture.

The basic terminology of lubrication and wear was
described earlier in the discussion of tribology. In this section, wear
mechanisms applied to TKR are discussed. These mechanisms have been
deduced from retrieval studies, principally from components that have
failed due to instability or loosening (18,27,28,29,60,61,63,139,234). Examples of wear in a range of retrieved TKR components are shown in Figure 100.53. Deformation frequently occurs in dome-shaped patellar components when they are used in anatomically shaped grooves (Fig. 100.54).
For partially conforming condylar replacements, on the application of
load, a contact patch is produced on each condyle, the shape depending
on the geometry of the tibial and femoral surfaces (151,189,190). Contacts approximating circular are called point contacts,
whereas when the surfaces are almost conforming in the frontal plane
but only partially conforming in the sagittal plane, a cigar shape
called a line contact is formed (Fig. 100.10). The contact area varies considerably, from about 150 mm² for moderate- to high-conformity knees in early flexion, down to 30 mm² for low conformity knees in flexion (Table 100.4).
The corresponding maximum compressive pressures are 10 to 50 mPa at the
centers of the contact areas, the mean pressure being the maximum
divided by 1.5.

The yield stress of UHMWPE is about 15 mPa (135,151,176), above which permanent deformation will occur under the center of the contact area (182).
In knees of low conformity, as the femur slides over the plastic
surface, a deformed groove up to 0.3 mm deep occurs in the first

few
months of use, and thereafter the groove is steadily deepened by wear.
In knees of moderate to high conformity, the deformed area is
relatively small, with a depth of less than 0.1 mm. Within the contact
area, there are a multitude of local contact points at a micron level,
depending on the microroughness of both the metal and plastic surfaces.
There are several mechanisms whereby small particles are released. Some
of these mechanisms are fatigue processes requiring numerous cycles of
sliding. Multidirectional sliding is also more damaging than sliding,
which is predominantly limited to the same direction. The wear
processes also depend on whether the kinematics involve rolling,
tractive rolling, or sliding (23,28). These wear processes (27,60,61) are as follows:

Scanning electron microscopy of the surfaces of retrieved plastic components illustrates the wear mechanisms (Fig. 100.55).
Ripples perpendicular to the sliding direction are typical. Fibrils are
seen to be stretched from the crests of the ripples, while there is
also evidence of granule formation. Shearing of the surface layer
results in fibrils or thin flakes. Fluid samples, after processing, can

be filtered to show the resulting particles. They are categorized into granules, fibrils, and flakes (47,48).
The types and sizes of particles are indicative of the wear mechanisms
that were taking place. In general, smaller particles produce more
expression of cytokines from cells.

The above-mentioned phenomena are all related to the
surface of the plastic. However, other wear mechanisms are more
appropriately called “damage mechanisms.” These mechanisms involve more
general breakdown of the surface region of the material, including
failure that starts entirely subsurface (205).
These damage mechanisms, and other phenomena that involve cracks at and
under the surface, are dependent on the component geometry, the sliding
conditions, and the type of UHMWPE (81,189,190). To illustrate this point, a typical contact situation is shown in Figure 100.56.
Around the periphery of the contact area, radial tensile stresses occur
of magnitude 0.13 times the maximum compressive stress (124).
As sliding and rolling take place, at a particular location on the
plastic surface, the stresses cycle between tensile and compressive.
This is equivalent to the stresses in a bar of material subject to
repeated bending in opposite directions. As a result, cracks will be
produced on the surface, which will progress along the surface and into
the material, but because the tensile stresses reduce with depth, the
crack depth is limited. However, when the cracks are not perpendicular
to the surface and adjacent cracks coalesce, pits are formed with
typical dimensions of 0.5 to 2 mm in width and depth. Such pits are
commonly seen on retrieved components (18,63,139), from low to high conformities (27), including mobile bearings.

A more severe form of damage occurs when cracks form and
propagate beneath the surface. This is termed “delamination.” A typical
crack depth is 1 mm, and the initial area affected can be several
millimeters across, seen as a white patch (18,27,29).
These cracks are associated with regions of maximum shear stress, which
occur beneath the center of a contact area at a depth of approximately
0.25 times the contact width, usually 1 to 2 mm. The direction of the
maximum shear stress is at 45° to the surface (Fig. 100.56).
Beneath the periphery of the contact area in the cross-sectional plane,
there are further peaks of shear stress of lesser magnitude oriented at
90° to the surface. The stress directions are opposite at the leading
and trailing edges of the contact area. Hence, as the femoral component
slides across the plastic, particular points beneath the plastic
surface experience shear stresses that change in direction, producing a
fatigue situation that leads to crack formation. Once the cracks reach
the surface, the consequent disruption of the surface leads to rapid
delamination of adjacent areas, eventually covering the entire contact
region. Examples are shown in Figure 100.53 and in sectional view in Figure 100.55.

Physical tests, like computer models, are
simplifications of reality that embody sufficient and appropriate
characteristics of the actual situation to address the question being
asked. In some cases, simple tests suffice; in other cases, a complex
test is required.