J.

Bmmechanm
Vol. 26. No.
Printed in Great Britain

8. pp

929-943.

1993.

1X.21 -9290/ 93
$6.00+.00
1993 Pergamon Press Ltd

NUMERICAL SIMULATION OF LEAFLET FLEXURE IN

BIOPROSTHETIC VALVES MOUNTED ON RIGID AND
EXPANSILE STENTS
S. KRUCINSKI,* I. VESELY,~$ M. A. DOKAINISH* and G. CAMPBELLS~II
*Department of Mechanical Engineering, McMaster University, Hamilton; TDepartment of Electrical
Engineering; $Medical Biophysics, University of Western Ontario; The John P. Robarts Research
Institute; and )/The University Hospital, London, Ontario, Canada
Abstract-Recent
studies suggest that flexural stresses induced during the opening phase may be
responsible for much of the mechanical failures of bioprosthetic
heart valves. Sharp leaflet bending is
promoted by the mounting of valves on rigid stents that do not mimic the systolic expansion of the natural
aortic root. We, therefore,
hypothesized
that flexural stresses could be significantly
reduced
by
incorporating
a flexible or expansile supporting stent into the valve design. Using our own non-linear finite
element code (INDAP) and the pre- and post-processor
modules of a commercial finite element package
(PATRAN), we simulated the opening and closing beha.iour
a trileaflet bovine pericardial
valve. The
leaflets of this valve were assumed to be of uniform thickness. with a non-linear elastic behaviour adapted
from experimentally obtained bending stiffness data. Our simulations have shown that during maximal
systolic valve opening, sharp curvatures are induced in the leaflets near their commissural attachment to the
supporting
stent. These areas of sharp flexure experience compressive stresses of similar magnitude to the
tensile stresses induced in the leaflets during valve closure. By incorporating
a stent with posts that pivot
about their base, such that a 10% expansion at the commissures
is realized, we were able to reduce the
compressive commissural stressing from 250 to 150 kPa. This was a reduction of40%. Conversely, a simple
pliable stent with stent posts that deflect inward and outward under load did not achieve a significant
reduction of compressive
stresses. This numerical analysis, therefore, supports the theory that (i) high
flexural and compressive stresses exist at sites of sharp leaflet bending and may promote bioprosthetic
valve
failure, and (ii) that proper design of the supporting
stent can significantly reduce such flexural stresses.

INTRODUCTION
Bioprosthetic heart valve implants have proven clinically successful over the short term, but their long-term
performance has been disappointing (Bortolotti et al.,

1987: Gallo et al., 1986, 1988; Magilligan et al., 1985;

Teoh ef al., 1989). Bovine pericardial valves, in particufar. have suffered from poor durability, usually
performing significantly worse than porcine xenografts (Bortolotti et al., 1987; Thiene et al., 1986;
Wheatley et ul., 1987). The degenerative failure of
these bioprostheses results mainly from severe calcification and from leaflet tearing at points of their
attachment to the supporting frame or stent. Indeed,
many of the early failures of pericardial valves can be
attributed to design flaws that induced high stresses in
the valve cusps (Trowbridge et al., 1988). While the
calcification problem may be eventually solved
and chemical
intervention
through
biological
(Golomb et al., 1987; Pathak et al., 1990; Vesely et al.,
1991a), mechanical failure caused by faulty design
could be reduced immediately through appropriate
modifications to the stent or cusp geometry.
Until recently, it was generally believed that the
majority of structural failures of bioprosthetic valves

Received in jnal

form

28 December

Address correspondence
to: Ivan
Robarts
Research
Institute,
P.O.
Ontario, Canada N6A 5K8.

1992.

Vesely, The John P.

Box 5015. London,

resulted from tensile stress developed during valve

closure. Numerous studies have suggested, however,
that cuspal degeneration is mediated through flexural
stressing and compressive
buckling (Gabbay er al.,
1988; Thubrikar et al., 1986; Trowbridge and Crofts,
1987; Trowbridge et al., 1988; Vesely and Boughner,
1989; Vesely et al., 1988). Pathologic studies, in par-

ticular, have correlated the sites of cuspal tears to

areas that experience high flexural stresses, likely
caused by deficiencies in valve design (Ishihara et nl.,
1981; Pomar et al., 1984; Stein et al., 1985; Walley and
Keon, 1987; Walley et al., 1987). Design flaws in failed
bioprostheses may be readily apparent in retrospect,
but the prospective evaluation of new bioprosthetic
valve designs is much more difficult. Mechanical testing can assess the performance of leaflet materials and
stent designs (Lee et al., 1984; Pereira er a[., 1990; Song
et al., 1990; Vesely, 1991; Vesely and Boughner, 1989,
1990; Vesely and Noseworthy, 1992; Vesely et al.,
1990a), but cannot predict how the valve will function
as a whole. Mathematical modelling, on the other
hand, has the potential to become a very useful technique in the study and evaluation of new prosthetic
valve designs. Over the years, numerical modelling
has made significant contributions to understanding
the loading of natural and prosthetic heart valves
during the closing phase (Cataloglu et al., 1977; Christie and Medland, 1982; Hamid et al., 1986; Rousseau
et al., 1988). More recently, one group has simulated
leaflet flexure during valve opening in a two-dimensional model of a pericardial valve (Huang et al., 1990)

S. KRUCINSKIet al.

930

and in a three dimensional bileaflet valve (Black et al.,

1991). This work has demonstrated the presence of
significant flexural stresses during valve opening.
Our previous research into valve leaflet mechanics
(Vesely et al., 1988, 1990b, 1991b, 1992), and mathematical modelling by others (Hamid et al., 1985,
1986), suggests that the patterns of leaflet stress in the
prosthetic valves can be influenced by stent design.
We have, therefore, hypothesized that if the natural
expansion of the aortic root is duplicated in a bioprosthetic valve, the resultant flexural stresses induced
during valve opening could be significantly reduced.
A numerical modelling technique was, therefore,
developed to test this hypothesis. Specifically, we were
interested in determining to what extent the opening
curvatures and the associated flexural stresses could
be modified with the use of pliable and expansile
mounting frames.

METHODS
In finite element methods, the volume of an analysed structure is subdivided into small but finite
subvolumes called elements. If these elements are of
finite thickness and curved, they can approximate the
shape of the valve very well. In our approach, we have
chosen 20-node isoparametric brick elements for both
the valve leaflets and the supporting stent. The shape
of the valve leaflet was adapted from existing clinically
available pericardial valves, as described by Quijano
(1989). The leaflets were symmetrical, with a top freeedge angle of 17.5 and a bottom angle of 36.5. The
shape of the cusp was spherical, with a flattened
coaptation region (Fig. 1). Coaptation height was
6.4 mm at the commissures, and zero at the centre,
where all three cusps came together. The leaflets were
assumed to be 0.5 mm thick, and the connection
between the stent and the leaflets was assumed to be
perfect, in the sense that nodal displacements at the
attachment surface were identical for the leaflet and
the stent.
Our modelling approach utilized non-linear continuum mechanics to describe the highly non-linear
material characteristics of the valve leaflets, and the
large deformations that occur during valve opening.
We have used true three-dimensional brick elements,
rather than shell elements, for accurately simulating
leaflet stresses (as suggested by Black et al., 1991), and
have paid particular attention to simulating contact
between the cusps during valve opening and closure
to model valve coaptation (Black et al., 1991). We
have also relied heavily on experimentally acquired
data of flexural stiffness to model properly the valve
opening process and realisticatty simulate the magnitude of flexural stresses.
The finite element model was deformed by a timevarying physiological pressure waveform, incrementally applied to both the inflow and outflow surfaces
of the valve leaflets. The deformed shape of the pros-

thetic valve was solved iteratively after each incremental increase in applied pressure. The equilibrium
shape of the valve at time t was characterized by the
principle of virtual displacements referred to the initial unstressed configuration, as given by the relation

+ r fT6(uT- U,)dS.
J sc
In equation (1) Sij and Eij denote components of the
second Piolla-Kirchoff stress tensor and the Lagrange strain tensor, respectively. The f: and fi are
components of the externally applied body and surface force vectors due to the hydrostatic force, and the
6Ui symbolize the ith component of the virtual displacement vector that deforms the mesh. The last two
terms on the right-hand side define an additional
constraint due to the contact condition, which prevents any point in space to be in the interior of more
than one body. The use of such a contact algorithm
enabled realistic simulations of the leaflets in contact
both during opening and closure. Rather than constraining the finite element mesh not to deform past
a given point, as done by Hamid et al. (1986), Christie
and Medland (1982) and Huang et al. (1990), we used
a Lagrangian multipliers approach to simulate contact, as first described by Chaudhary and Bathe
(1985). In this approach, the distance between contacting bodies is measured by a gap function. As any
points of the body come into contact, the value of the
gap function for the points decreases and eventually
reaches zero when contact is established. The Lagrangian multipliers are solved at every node in the vicinity of the coaptation region, but the coaptation region
does not have to be known exactly beforehand. If the
nodes slide, the ratio of the tangential to normal force
is determined by the coefficient of friction and the
nodal forces are adjusted appropriately. A coefficient
of friction of zero was selected to simulate the absence
of friction at the leaflet interface. The material for the
valve leaflets, the pericardium, was assumed to be of
uniform thickness and isotropic. While some investigators have reported that pericardium is mildly anisotropic in its extensibility (Crofts and Trowbridge,
1988; Lee et al., 1989), there is no evidence that
pericardial valves are constructed to accommodate
this mild anisotropy in a systematic way. The material
model also assumed that the pericardium was elastic
in the sense that during a closed deformation cycle the
strain energy is stored and released so that no net
work is done on the body. The strain energy function
of the material can, therefore, be expressed in terms of
invariants of tensor E,
w=

W(II E ,I ~ E V ~ ~ E ),

(2)

Numerical simulation of leaflet

931

above loads of 0.5 g (Pereira et al., 1990). For a 5 mm

wide specimen 0.5 mm thick, a 0.5 g load corresponds
S, = a(.&),
(3) to a stress of 1.9 kPa. If data below this value are
and are
neglected,
the length of the toe region of the
where S,=(31;,) I2 and E,=2(1;,/3),
stress-strain
curve can be greatly underestimated.
referred to as effective stress and effective strain,
respectively.
The terms 1 is and I;, represent the
Indeed, the measured extensibility of soft tissues vasecond invariants of stress and strain deviators. The
ries between laboratories (Lee et al., 1989; Trowbridge
strain potential function associated with equation (2), and Crofts, 1986; van Noort et al., 1982). The extensiunder multiaxial stress, has the form
bility of the tissue and the general shape of the
- stress-strain
curve at stresses lower than 2 kPa is,
W1E112E.13E)= W1E,&E)+XW3E),
(4a)
therefore, unlikely to be very precise. This lack of
where
good experimental
data may not be important
for
modelling valve closure but it becomes a serious prob@=[a/(a+l)]E:+
(4b)
lem for modelling cusp flexure.
and
Because we wished to simulate leaflet bending usmg
realistic flexural mechanics, we chose to incorporate
direct experimental
measurements
of bending stiffness. These data were previously acquired by us for
i2E=12e-2(1,,-i).
(4c)
glutaraldehyde-treated
tissues (Vesely ef al.. 1988).
The last term in equation (4a) reflects the near incomand were used to calculate an effective elastic modulus
which could be used to augment the gap in the
pressibility
of the material (Cescotto and Fonder,
stress-strain
curves of pericardium at low strains. The
1979). The positive constant 1 represents the bulk
bending stiffness of a 5 mm wide, 0.5 mm thick strip of
modulus of the material. The material parameters
glutaraldehyde-treated
aortic valve tissue was found
appearing in equation (3) were estimated as a=650
and n =4.0, from curves obtained experimentally
(Fig.
to be 12 nNm* (Vesely et al., 1988). Using the simple
2). and modified as explained below.
beam theory as a first approximation,
the effective
elastic modulus of such a material would be 230 kPa.
A realistic evaluation of valve leaflet deformation
The elastic modulus calculated at 5% strain from
requires a relationship
between stress (S) and strain
published data can range from a low of 200 kPa for
(E) tensors that is based on experimental
evidence.
A large volume
of experimental
data on the
Pereira et al. (1990), to a high of 1.0 MPa for van
stress-strain
behaviour
of pericardium
is available
Noort et al. (1982). The use of such stress-strain
from the literature (Pereira et al., 1990; Trowbridge
relationships, in our view, would greatly overestimate
and Crofts, 1986; van Noort et al., 1982), and values
the bending stiffness of the material at strains less than
for parameters a and n can be fitted relatively well.
5% and produce erroneous results in simulating leafExperimental data at very low strains and loads, howlet flexure. Our modified stress-strain
response, based
ever, are imprecise. In fact. most experimentalists
bein part on experimental
data of van Noort er irl.
gin measuring the elastic response of test strips only
(1982), gives an elastic modulus of 325 kPa at 5%
and the stress-strain

relationship

as

1 ,

.
0.8

-.__

9.

Experimental Data
Fitted Function

0.2

0.1

Strain
Fig. 2. The stressstrain curve used for the modelling of the bovine pericardium. A power approximation
fitted to the experimental
data obtained from van Noort et al. (1982). and the shape of the curve was
modified in the extremely low strain portion (below 5% strain) to reflect the very low bending stiffness of the
material.

932

S. KRUCINSKIet nl

strain, and drops exponentially towards zero strain

such that at 2% strain the elastic modulus is only
20.8 kPa. By comparison, the lowest elastic modulus
that can be calculated from published stress-strain
curves is roughly100 kPA at 2% strain (Pereira et al.,
1990). We found that this refined material model,
based on experimentally obtained bending stiffness
data, enabled the valve to open at physiological pressures, while a direct implementation
of uniaxial
stress-strain data produced a valve that did not open
at all. It should be noted that Huangs two-dimensional model (Huang et al., 1990), the only other finite
element simulation that modelled complete valve
opening, also used elastic moduli much lower than
those observed experimentally with uniaxial tensile
tests.
The physiological pressure waveform used for this
simulation was adapted from a Medical Physiology
reference book (Ganong, 1983) (Fig. 3). The peak
pressure gradient was 87 mmHg (11.6 kPa) in diastale, 2.0 mmHg (0.27 kPa) in systole, the pressures
were incremented in 10 ms steps, and the solution was
iterated to convergence prior to each successive step
for 1: cycles.
Three examples of stent designs were simulated to
examine the effects of rigid, pliable and expansile
stents on the distribution of leaflet stresses. A rigid
stent was simulated by setting all nodal displacements
of the stent to zero. This effectively prevented any
leaflet motion along the line of attachment between
the stent and the valve cusps. A pliable stent was
simulated by setting the material of the stent to be
linear and isotropic, with a Youngs modulus of
1.0 GPa and a Poissons ratio of 0.40. This material
model roughly approximated a stent made of highdensity polyethylene. For the third configuration, the
stent posts were prescribed to pivot outward during
systolic valve opening. The bottoms of the posts were

Pressure Waveform

150

300
The

450

600

hinged at the rigid base of the stent, and the tops of

the stent posts moved radially outward, away from
the vertical axis of symmetry of the valve. This is
a very practical design that could be duplicated by
affixing the tops of the stent posts to the recipient
aorta. The amount of outward radial movement was
set at lo%, to reflect the expansion observed experimentally (Vesely et al., 1990b). No inward motion of
the stent $osts was permitted in diastole since, in this
simulation, we wanted to examine only the outward
motion of the stent.
RESULTS
Rigid scent

In the closed position, a tensile major principal

stress of 350 kPa was observed in the commissural
region [Fig. 4(a)]. Such stresses are induced at the
commissures due to the tension that is generated in
the valve leaflets in the closed position, and is consistent with all previous mathematical simulations of
closed valves (Black et al., 1991; Christie and Medland, 1982; Hamid et al., 1985; Rousseau et al., 1988).
In the fully open configuration, a compressive major
principal stress of 250 kPa was observed, again in the
commissural region. This compressive stress resulted
from the sharp circumferential bending of the leaflet
at that location [Fig. 4(b)]. Compressive stresses
induced by leaflet flexure are, therefore, of magnitude
comparable to tensile stresses and likely contribute to
material failure through a compressive buckling phenomenon (Vesely et al., 1988).
Pliable stent

In this simulation, the stent posts deflected inward

by 3.6% of the top radius (0.37 mm) during diastole
and outward by 4.9% (0.5 mm) during systole. Because of the inward deflection of the pliable stent
posts in systole, significant compressive stresses were
induced near the central coaptation area [Fig. 5(a)],
This was accompanied by some wrinkling of the leaflet free edge [Fig. 5(b)]. During opening, this stent
design produced compressive stress at the commissural region very similar to that of the rigid stent [Fig.
5(a)]. This simulation has, therefore, demonstrated
that our very simple flexible stent design not only
disturbed the coaptation process but also restrained
the outward radial motion of the valve cusps. Such
a stent design, therefore, offers no appreciable reduction in compressive stresses during the opening phase
of this type of valve.

750

in milliseconds

Fig. 3. Plot of the time-varying pressures simulated on the

ventricular and aortic surfaces of the prosthetic valve.
A maximal pressure gradient of 2 mmHg was assumed during the opening phase of the valve.

Pivoting stent posts

The results of this numerical simulation showed

that an outward movement of the stent posts, equivalent to a 10% radial expansion of the commissural
diameter, considerably reduced the circumferential
leaflet curvatures at the commissures when the valve

Fig 1. Oblique and outflow views of the model pericardial valve used in the finite element simulations. The geometry of the valve was approximated
from published information
on a size 25 A
(24.34 mm base diameter, 20.7 mm top diameter) Carpentier-Edwards
bovine pericardial xenograft. Other than the size and approximate
geometry, no other design features were duplicated.

Open

Fig. 4(a). Stress map of a valve with a rigid stem in the closed and open configurations. A map of the major principal stresses is superimposed on one half of the
undeformed leaflet for the sake of clarity. The right edge of the leaflet is the line of symmetry. Red denotes the compressive stresses that are potentially damaging to
bioprosthetic tissues. Note the high concentration of compressive stress at the region of the commissures in the open configuration. The stent has zero stress sjnce it was
assumed to be perfectly rigid and, therefore, did not deform. The colour bar denotes stresses in kPa.

Closed

Rigid Stent

Rigid Stent

Fig. 4(b). An oblique view of the whole valve in the closed and open configuration.
Note the high reverse
flexing of the leaflets in the open configuration
and the sharp curvatures
near the stent posts.

Open

Fig. 5 (a). Stress map of a valve with a pliable stent in the closed and open configurations. This valve had a stem that deflected only by the action of the leaflets pushing the
posts outward. Note the high concentration of compressive stress near the central coaptation region, due to the inward movement of the stent posts during valve closure.
Also note that such a pliable stent did not significantly reduce the compressive commissural stresses in the open configuration. The deflection of the stem in diastole
created the high tensile and compressive stresses.

Closed

Pliable Stent

Pliable Stent

Fig. 5(b). An oblique view of the whole valve in the closed and open configuration.
Note the sharp
curvatures
near the stent posts that were not eliminated by this type of stent design.

93-l

Open

Fig. 6(a). Stress map of a valve with pivoting stent posts. In this simulation, the stent posts were prescribed to pivot outward about the base such that the separation of the
commissures increased by 10% during systole. Note the decrease in compressive stress at the commissures, compared to the previous two designs. Since the stent was
prescribed to move only during systole, it behaves like a rigid stent in diastole and is, therefore, not stressed.

Closed

Pivoting Stent Posts

Pivotiner Stent Posts

Fig. 6(b). An oblique view of the whole valve in the closed and open configuration.
Note the wide leaflet
excursion, and the much more gentle free-edge curvatures of this stent design.

939

Real Valve

Fig. 7. Comparison images of our finite element simulation of a pericardial valve in the open configuration and a pericardial valve open in a pulse duplicator system. Note the similarity of
the flexure pattern at the free edge. The image on the right has been reprinted by kind permission of Sorin Biomedica.

Simulation

Numerical simulation of leaflet

was fully open [Fig. 6(b)]. The amount of compressive stressing of the valve leaflets during opening
decreased as well [Fig. 6(a)]. Compared with a major
principal stress of 250 kPa for a rigid stent, this simulation produced a major principal stress of 150 kPa,
a reduction of 40%. This reduction in compressive
stress, however, was not accompanied
by an increase
in tensile stress at the centre of the cusp free edge, as
would be expected if the radial movement of the stem
posts produced leaflet tension.
DISCUSSION

The relatively good durability of some bioprosthetic valves, and the premature failure of others, is
likely related to design parameters. The early failure
of pericardial valves employing an alignment stitch at
the commissures
is a typical example of a design
feature that proved to be catastrophic
(Walley and
Keon, 1987; Walley et al., 1987). We feel that another
detrimental feature of all stent mounted tissue valves
is the inability of the supporting
stent to expand
radially with the recipient aortic root. The natural
aortic root has clearly been shown to expand during
the ejection phase of the cardiac cycle (Brewer et al.,
1976; Thubrikar et al., 1980; Vesely et al., 1990b), yet
no existing valve mounting frame has addressed this
phenomenon.
Presently. most supporting stents have
pliable stent posts that aim to reduce shock loading
on the leaflets during valve closure. No stent design,
to our knowledge, has attempted to reduce the commissural flexing that damages valve leaflets (Vesely et
al., 1988). and contributes
to leaflet tearing in the
absence of calcification (Ishihara et al., 1981: Pomar et
al.. 1984; Stein et al., 1985). We have, therefore,
simulated the flexure patterns that occur during valve
opening with our three-dimensional
modelling capability, and have examined how varying the design of
the stent can affect leaflet stresses.
This model, like most numerical analyses of bioprosthetic valves. has three primary limitations: (i) it
does not mimic the interaction of fluid flow with tissue
deflection; (ii) it is an elastic model that does not take
into account the viscous nature of the valve tissue; and
(iii) it employs a solid mechanics approximation
of
a composite material that has poorly characterized
constituents.
The principal advantage of this model
over works published
to date, however, is that it
simulates flexure patterns in three-dimensions
during
complete valve opening. The ideal simulation
that
could duplicate the complexity of the leaflet motion
during valve opening would require a coupling of the
fluid flow during the ejection phase to the motion of
the valve leaflets in a mutually dependent way. Such
a coupled simulation. lo our knowledge, has never
been done on heart valves because of the difficulty of
the problem, and the large computational
power required to do so. An uncoupled simulation that would
deflect the leaflets by a prescribed flow field was not

941

possible because sufficient experimental

data are unavailable in the literature, and because such an approach could not simulate stent-leaflet
interactions.
Because of their high pliability, the systolic leaflet
motion is likely affected the most by pressure fluctuations resulting from unsteady fluid flow through the
open valve. Our modelling approach, therefore, cannot simulate the subtle undulations and flutter of the
valve leaflets during systole. Nevertheless, since final
leaflet curvatures are ultimately the factor that generates the pattern of three-dimensional
stressing in the
open configuration,
it is reasonable to simulate the
opening phase in the best way possible without using
a solid-fluid interaction,
as long as the patterns of
leaflet curvature
compare
well with experimental
data. Indeed, the orifice view of our simulated valve
with a rigid stent compares very well with that of
a pericardial valve functioning in a pulse duplicator
(Fig. 7). If the patterns of the free-edge curvature are
similar between the model and the real valve, and the
model utilizes a reasonable material description, it is
safe to assume that the patterns of leaflet stress are
also realistic.
The only major difference between the circumferential curvatures produced by our model and those of
the real pericardial valve is at the centre of the coaptation region. We assumed the zero-stress-state
configuration to be that shown in Fig. 1. In reality, this is
not the case. In the manufacture of the prosthesis, the
pericardial leaflets are actually pushed together and
bent into the shape shown in Fig. 1. The valve is.
therefore, pre-stressed even when it is at zero pressure,
and will open to a more circular shape without the
central peak. This initial pre-stress, however, is limited
to the central portion of the free edge and does not
affect the curvatures induced elsewhere in the leaflets.
Our model also had somewhat more radial bending,
but that did not affect the circumferential
curvatures
at the free edge. This model, therefore, promises to be
a good analytical instrument for evaluating the curvatures and induced flexural stresses in prospective
bioprosthetic
valve designs, and for providing an insight into the mechanical behaviour of biological tissues.
The generated stress maps (Figs 4-6) indicate that
flexural stresses are, indeed, induced near the commissures and within the free margin during valve opening.
The magnitude of these compressive stresses is comparable to those presented by Huang et al. (1990) with
a two-dimensional
simulation, and to those of Black
et al. (1991), who modelled a bicuspid pericardial
valve. We have shown previously that such compressive deformations
damage bioprosthetic
valve material, likely through a compressive
buckling process
(Vesely and Boughner, 1989). This numerical simulation, therefore, further supports the hypothesis that an
important mode of mechanical failure of tissue valves
is flexural damage induced during valve opening
(Vesely and Boughner, 1989; Vesely et al., 1988). In-

942

S. KRUCINSKIet al.

deed, if one examines the patterns of tears on explanted pericardial heterografts and porcine xenografts (Gabbay et al., 1988; Ishihara et al., 1981; Stein et
al, 1985; Wheatley et al., 1987), there is a high correlation between the sites of tears and the location of
stress concentrations produced by our simulation.
These simulations have also demonstrated that
valves with stent posts that pivot about their attachment to the base can significantly reduce leaflet
tlexure and the associated leaflet stresses. While additional iterative studies need to be performed to determine the optimal amount of expansion that can reduce compressive stresses the most, a 10% radial
expansion at the commissures appears to be adequate.
Conversely, a simple, pliable stent support that relies
on the compressive forces generated within the valve
cusps to deflect the stent posts outward does not
provide a sufficient amount of stress reduction. While
it can be argued that a stent constructed from a more
pliable material would have deflected more and could
have performed better in systole, a more pliable stent
would have collapsed under diastolic loads. In our
simulation, the stent posts deflected inward by
0.37 mm, increasing the free-edge angle from 17.5 to
19. A greater stent post flexibility would have deformed the leaflets to such an extent that leaflet coaptation during diastole would be compromised. With
this engineering exercise, we have shown that one type
of stent design can significantly reduce commissural
stressing, while another equally plausible design cannot. It is, therefore, clear that detailed mathematical
simulations of preliminary valve designs can provide
valuable contributions to the development of new
bioprosthetic valves. Perhaps, the best stent would be
one that functions in harmony with the patients aortic root. Ideally, the tops of the stent posts should be
fastened to the recipient aortic root such that the
expansion of the aorta during ventricular contraction
will pull the tops of the stent posts outward with it.
Such motion will enable the reduction of leaflet stresses during valve opening and may reduce the incidence of leaflet tearing at the flexure points.
Because of the inherent simplicity of pericardial
tissue relative to aortic valves leaflets, we chose to
investigate the pericardial valve first. Such an approach, however, does not need to be limited to
pericardial valves. Our primary interest lies in the
porcine aortic valve xenograft, with its more physiological construction. The mechanism that we propose to
reduce flexural stresses at the commissures already
exists in natural aortic valves. The commissures of
aortic valves naturally move outward during systolic
valve opening as the root expands (Brewer et al., 1976;
Thubrikar et al., 1980; Vesely et al., 1990b). It is,
therefore, logical to assume that the mechanism of
stress reduction proposed for the pericardial valve can
be readily applied to porcine aortic valve xenografts
as well. It should be noted, however, that because of
the much greater complexity of the natural aortic
valve (Vesely and Noseworthy, 1992) many of the

assumptions applicable to pericardial valves, such as

material isotropy and geometrical symmetry, cannot
be applied. The concept of pivoting stent posts, however, is the rational application of a mechanism that
exists in natural aortic valves. If such a stent design
can significantly reduce flexural stresses in pericardial
valves, it is likely to perform equally well in porcine
xenografts.

Acknowledgement-This research was supported in part by

a Strategic Grant from the Natural Sciences and Engineering
Research Council of Canada, and by the Heart and Stroke
Foundation of Ontario. Dr Vesely is a Research Scholar of
the Foundation. The authors are grateful to Dr D. R. Boughner, from the University of Western Ontario, for his advice
and discussion during the course of this work.
REFERENCES

Black, M. M., Howard, I. C., Huang, X. and Patterson, E. A.

(1991) A three-dimensional analysis of a bioprosthetic
heartvalve. J. Biomechanics 24, 793-801.
_
Bortolotti. U.. Milano. A.. Thiene. G.. Guerra. F.. Mazzucco.
A., Valente; M., Talenti, E. and Gallucci, V. (1987) Early
mechanical failures of the Hancock pericardial xenograft.
J. 7horacic Cardiovasc. Surg. 94, 200-207.

Brewer, R. J., Deck, J. D. and Copohe, B. (1976) The dynamic

aortic root: its role in aortic valve function. J. Thoracic
Cardiovasc. Surg. 72, 413.

Cataloglu, A., Clark, R. E. and Goulds, P. L. (1977) Refined

stress analysis of human aortic heart valves. J. Biomechanics 10, 153-158.
Cescotto, S. and Fonder, G. (1979) A finite element approach
for large strains of nearly incompressible rubber-like
materials. J. Solids Struct. 15, 589-605.
Chaudhary, H. B. and Bathe, K. J. (1985) A solution method
for static and dynamic analysis of the three-dimensional
contact problems with friction. Comput. Struct. 24,
855-873.
Christie, G. W. and Medland, I. C. (1982) A non-linear finite
stress analysis of bioprosthetic heart valves. In Finite Elements in Biomechanics (Edited by Gallagher, R. H.). Wiley,
New York.
Crofts, C. E. and Trowbridge, E. A. (1988) The tensile
strength of natural and chemically modified bovine
pericardium. J. biomed. Mater. Res. 22, 89-98.
Gabbay, S., Kadam, P., Factor, S. and Cheung, T. K. (1988)
Do heart valve bioprostheses degenerate for metabolic or
mechanical reasons? J. Thoracic Cardiovasc. Surg. 95,
208-2 15.

Gallo, I., Nistal, E., Arbe, E. and Artinano, E. (1988) Comparative study of primary tissue failure between porcine
(Hancock and Carpentier-Edwards) and bovine pericardial (Ionescu-Shiley) bioprostheses in the aortic position
at five- to nine-year follow-up. Am. J. Cardiol. 61,812-g 16.
Gallo, I., Nistal, F. and Artinano, E. (1986) Six to ten-year
follow-up on patients with the Hancock cardiac bioprostheses. J. Thoracic Cardiovasc. Surg. 92, 14-20.
Ganong, W. F. (1983) Review of Medical
Physiology. Lange
Medical Publications, Los Altos, California.
Golomb, G., Dixon,M., Smith, M. S. and Schoen, F. J., et al.
(1987) Controlled-release drug delivery of diphosphonates
to inhibit bioprosthetic heart valve calcification: release
rate modulation with silicone matrices via drug solubility.
J. Pharm. Sci. 76,271-276.
Hamid, M. S., Sabbah, H. N. and Stein, P. D. (1985) Finite
element evaluation of stresses on closed leaflets of bioprosthetic heart valves with flexible stents. Finite Elements
Anal. Des. 1, 213-225.

Numerical

simulation

Hamid, M. S.. Sabbah, H. N. and Stein, P. D. (1986) Influence of stem height upon stresses on the cusps of closed
bioprosthetic
valves. J. Biomechanics 19, 759-769.
Huang, X., Black. M. M., Howard, I. C. and Patterson, E. A.
(1990) A two dimensional finite element analysis of a bioprosthetic heart valve. J. Biomechanics 23, 753-762.
Ishihara, T., Ferrans. V. J., Boyce, S. W., Jones, M. and
Roberts, W. C. (1981) Structure and classification of cuspal
tears and perforations
in porcine bioprosthetic
cardiac
valves implanted in patients. Am. J. Cardiol. 48, 665-678.
Lee, J. M., Boughner, D. W. and Courtman,
D. W. (1984)
The glutaraldehyde-stabilized
porcine aortic valve xenograft. II. Etfect of fixation with or without pressure on the
tensile viscoelastic
properties
of the leaflet material.
J/. hiomed. Mater. Res. 18. 79-98.
Lee, J. M., Haberer, S. A. and Boughner. D. R. (1989) The
bovine pericardial
xenograft: I. Effect of fixation in aldehydes without
constraint
on the tensile viscoelastic
properties of bovine pericardium.
J. hiomed. Mater. Res.
23, 457475.
Magilligan,
D. J.. Lewis. J. W., Tilley, B. and Peterson, E.
(1985) The porcine bioprosthetic
valve. Twelve years later.
J. Thoracic Cardiovasc. Stag. 89, 4999507.
Noort. R. van. Yates, S. P., Martin, T. R. P., Barker, A. T.
and Blact. M. M. (1982) A study of the effects of glutaraldehyde and formaldehyde
on the mechanical behaviour of
bovine pericardium.
Biomaterials 3, 21-26.
Pathak, Y. V.. Boyd. J., Levy, R. J. and Schoen. F. J. (1990)
Prevention
of calcification
of glutaraldehyde
pretreated
bovine pericardium
through controlled release polymeric
implants: studies of Fe3+, A13+, protamine
sulphate and
levamisole. Biomaterials 11, 718-723.
Pereira. C. A.. Lee. J. M. and Haberer, S. A. (i990) Effects of
alternative
crosslinking
methods on the low strain rate
viscoelastic properties of bovine pericardial bioprosthetic
material. J. hiomed. Mater. Res. 24, 345361.
Pomar. J. L., Bosch, X., Chaitman,
B. R., Pelletier, C. and
Grodin. C. M. (1984) Late tears in leaflets of porcine
bioprostheses
in adults. .4nn. Thoracic Surg. 37, 78883.
Quijano, R. C. (1989) Engineering and design considerations
of the Carpentier-Edwards
pericardial valve. In Proceedings of the Carpentier-Edwards Pericardial Bioprostheses
Mini-Symposium. pp. 2740. Luzern, Switzerland.
Rousseau. E. P. M., van Steenhoven, A. A. and Janssen, J. D.
(1988) A mechanical analysis of the closed Hancock heart
valve prosthesis. J. Biomerhanics 21, 545-562.
Song, T., Vesely, I. and Boughner, D. R. (1990) ElTects of
dynamic fixation on the shear behaviour of porcine xenograft valves. Biomaterials II, 191-196.
Stein, P. D., Kemp. S. R., Riddle, J. M., Lee, M. W., Lewis, J.
W. and Magilligan, D. J. (1985) Relation of calcification to
torn leaflets of spontaneously
degenerated
porcine bioprosthetic valves. Ann. Thoracic Surg. 40, 175-180.
Teoh, K. H., Ivanov, J.. Weisel, R. D., Darcel, I. C. and
Rakowski, H. (1989) Survival and bioprosthetic
valve failure. Ten-year follow-up. Circulation 80 (Suppl. I). 18-115.
Thiene. G., Laborde. F., Valvente. M., Bical, 0.. Talenti, E.,
Bortolotti.
U. and Gallix, P. (1986) Experimental
evaluation of porcine-valved
conduits processed with a calcium-retarding
agent (Tb). J. Thoracic Cardiooasc. Stag.
91.215-224.

of leaflet

943

Thubrikar,
M. J., Aouad, J. and Nolan, S. P. (1986) Patterns
of calcific deposits in operatively excised stenotic or purely
regurgitant
aortic valves and their relation to mechanical
stress. Am. J. Cardio!. 54, 304-308.
Thubrikar,
M., Nolan, S. P., Bosher, L. P. and Deck, J. D.
(1980) The cyclic changes and structure of the base of the
aortic valve. Am. Heart J. 99, 217-224.
Trowbridge,
E. A. and Crofts, C. E. (1986) The standardisation of gauge length: its influence of the relative extensibility of natural
and chemically
modified
pericardium.
J. Biomechanics 19, 1023-1033.
Trowbridge, E. A. and Crofts, C. E. (1987) Pericardial heterograft valves: an assessment
of leaflet stresses and their
implications
for heart valve design. J. biomed. Engng 9,
345-355.
Trowbridge, E. A., Lawford, P. V., Crofts, C. E. and Roberts,
K. M. (1988) Pericardial heterografts: why do these valves
fail? J. Thoracic Surg. 95, 577-585.
Vesely, I. (1991) Analysis
of the Medtronic
INTACT
bioprosthetic
valve: effects of zero pressure fixation.
.I. Thoracic Cardiovasc. Surg. 101, 90-99.
Vesely, I. and Boughner, D. R. (1989) Analysis of the bending
behaviour
of porcine xenograft
leaflets and of natural
aortic valve material: bending stiffnes, neutral axis and
shear measurements.
J. Biomechanics 22, 655-671.
Vesely, I., Boughner, D. R. and Song, T. (1988) Tissue buckling as a mechanism
of bioprosthetic
valve failure. Ann.
Thoracic Surg. 46, 302-308.
Vesely, I., Gonzalez-Lavin,
L., Graf, D. and Boughner, D. R.
(1990a) Mechanical
testing
of cryopreserved
aortic
allografts: comparison
with xenografts and fresh tissue.
J. Thoracic Cardiovasc. Surg. 99, il9-123.
Veselv. I.. Krucinski. S. and Camnbell. G. (1992) Micromechanics and mathematical
mbdelling:
an inside look
atbioprosthetic
valve function. J. Cardiac Surg. 7, 85-95.
Vesely, I., Menkis, A. and Campbell,
G. (1990b) A computerized system for video analysis of the aortic valve.
IEEE Trans. biomed. Engng 37, 925-929.
Vesely, I., Menkis, A. H., Rutt, B. and Campbell, G. (1991b)
Aortic valve/root interactions
in porcine hearts: implications for bioprosthetic
valve sizing. J. Cardiac Surg. 6,
482489.
Vesely, I. and Noseworthy,
R. (1992) Micromechanics
of
the fibrosa and ventricularis
of aortic valve leaflets.
J. Biomechanics 25, 101-113.
Vesely, I., Noseworthy,
R. and Wilson, G. (1991a) Development of a hybrid xenograft/autograft
aortic valve bioprosthesis. Can. J. Card& 7; (Sup& A), 83A.
Walley. V. M. and Keon, W. J. (1987) Patterns of failure in
Ionescu-Shiley
bovine pericardial
bioprosthetic
valves.
J. Thoracic Cardiovasc. Surg. 93, 925-933.
Wallev, V. M.. Bedard. P.. Brais. M. and Keon. W. J. (1987)
Valie
failure
caused
by cusp tears in low-profile
Ionescu-Shiley
bovine pericardial
bioprosthetic
valves.
J. Thoracic Cardioeasc. Surg. 93, 583-586.
Wheatley, D. J., Fisher, J., Reece, I. J., Spyt, T. and Breeze. P.
(1987) Primary tissue failure in pericardial
heart valves.
J. Thoracic Cardiovasc. Surg. 94, 367-374.