Вы находитесь на странице: 1из 13

Available online at www.sciencedirect.

com

Medical Engineering & Physics 29 (2007) 1106–1118

Assessment of endoleak significance after endovascular repair of


abdominal aortic aneurysms: A lumped parameter model
B.J.B.M. Wolters a,∗ , M. Emmer a , M.C.M. Rutten a , G.W.H. Schurink b , F.N. van de Vosse a
a Department of Biomedical Engineering, Eindhoven University of Technology, The Netherlands
b Department of Surgery, University Hospital Maastricht, The Netherlands

Received 14 January 2006; received in revised form 24 November 2006; accepted 25 November 2006

Abstract
The outcome of endovascular repair of abdominal aortic aneurysms (AAAs) is greatly compromised by the possible occurrence of endoleak.
Previously, the causes and effects of endoleak on a patient-specific basis have mainly been investigated in experimental studies. In order to
both reconcile and physically substantiate the various experimental findings, a lumped parameter model of an incompletely excluded AAA
was developed. After experimental validation, the model was applied to study the effects on the intrasac pressure of the degree of endoleak,
the degree of stent-graft compliance, and the resistance of a possible outflow tract formed by a branching vessel. It is concluded that the
presence of endoleak leads to elevated intrasac pressure, the mean of which is mainly governed by the outflow tract resistance, while the
pulse pressure is governed by both the endoleak resistance and the stent-graft compliance. Based on the agreement of the current results
with previous findings, it is further concluded that the lumped parameter modelling method provides a useful numerical tool for validating
experimental endoleak studies.
© 2006 IPEM. Published by Elsevier Ltd. All rights reserved.

Keywords: Abdominal aortic aneurysm; Endoleak; Intrasac pressure; Experimental setup; Computational model

1. Introduction Previously, the causes and effects of endoleak originating


from the aortic lumen have mainly been investigated in exper-
The outcome of endovascular aneurysm repair (EVAR) imental studies. In an animal study of saccular aneurysms
of abdominal aortic aneurysms (AAAs) is greatly compro- connected to the aorta by an endoleak channel, Schurink et
mised by the possible occurence of endoleak [1]. In general, al. [7] showed that irrespective of the channel dimensions,
endoleak is defined as persistent flow between the stent- endoleak yields an identical intrasac mean pressure (p̄sac )
graft and the excluded vascular segment [2,3]. All types compared to the systemic mean pressure (p̄sys ) measured
of endoleak yield continued pressurisation of the aneurysm in the aorta. The intrasac pressure pulsatility, however, was
sac [4]. Since high intrasac pressure is associated with inversely related to the endoleak channel diameter, and was
AAA expansion [5], endoleak is assumed responsible for obliterated by thrombosis of the channel. The first finding
increased risk of rupture. In a recent multicenter study, for was confirmed both by Skillern et al. [9] using animal models
both endoleak related to the stent-graft attachment sites (type of an excluded aneurysm with an attachment site failure and
I) and endoleak related to stent-graft failure and disconnec- by Chong et al. [10] using setups of an excluded aneurysm
tion (type III), this effect has indeed been demonstrated [6]. with either an attachment site failure or a partially permeable
On a patient-specific basis, however, the nature and signifi- stent-graft. The second finding also was confirmed by Chong
cance of endoleak remain poorly understood [1]. et al. [10], who reported a decrease of the intrasac pressure
pulsatility for lesser degrees of endoleak. Further, in a study
on thrombosed endoleak channels by Mehta et al. [11], it

was reported that the reduction of the intrasac peak systolic
Corresponding author. Tel.: +31 40 247 5398; fax: +31 40 244 7355.
E-mail address: b.j.b.m.wolters@tue.nl (B.J.B.M. Wolters).
pressure (pssac ) was inversely proportional to the channel

1350-4533/$ – see front matter © 2006 IPEM. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.medengphy.2006.11.008
B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118 1107

diameter and directly proportional to its length. In studies of generated, incorporating an excluded AAA in the presence
setups of incompletely excluded AAAs it was further shown of both an endoleak channel and an outflow tract. The
that the intrasac pressure due to endoleak is reduced in the model comprised several modules representing segments of
presence of an outflow tract formed by a branching vessel the aorta and large connected vessels (see Fig. 1(a)). Each
[10,12]. Finally, pulsatility of the intrasac pressure was also module was built from resistors, inductors, and/or capacitors,
shown present in fully excluded aneurysms in both studies that respectively accounted for the blood viscous properties,
of experimental setups [10,12,14,15] and animal studies [9]. the blood inertial properties, and the vessel wall or stent-
Despite that experimental studies themselves can provide graft compliance (see Fig. 1(b and c)). The elements were
insight into the consequences of endoleak in specific situa- mutually connected by nodal points. At the branch ends, the
tions, the results of various experiments from literature are modelled vessels were terminated by Windkessel modules
hard to compare. This is mainly due to the fact that of different representing distal vascular beds (see Fig. 1(d)). The AAA
studies using apparently similar setups, varying results have was modelled by a single capacitor, and the endoleak channel
been reported. Using a setup comparable to that of Chong et and the outflow tract were both modelled by single resistors
al. [10], Schurink et al. [16] for various degrees of endoleak (see Fig. 1(a)). All outflow resistors and the capacitors of the
found that p̄sac was decreased compared to p̄sys . The value segmental and Windkessel modules as well as the capacitor
of pssac was decreased as well compared to the systemic peak representing the AAA were connected to a standard refer-
systolic pressure (pssys ), while the intrasac and systemic dias- ence at zero-pressure. For the capacitors of the stent-graft
tolic pressures (pdsac and pssys ) were similar. Again for various modules, the reference pressure was the intrasac pressure
degrees of endoleak, Parodi et al. [12] found that p̄sac was psac .
increased compared to p̄sys . Again, pssac was decreased com- The behaviour of the lumped parameter model was
pared to pssys , but pdsac was higher than pdsys . Using a perforated described by a set of equations that was derived from the
behaviour of the constituting elements. With q = (q1 , q2 )T
stent-graft, Mehta et al. [15] found pdsys and pssac both to be
even higher than pssys . With respect to the effect of a possi- the column of locally defined nodal flows, and p˜ = (p1 , p2 )T
˜
the column of local nodal pressures, the behaviour of the
ble outflow tract, the latter further reported that the intrasac
pressure pulsatility was increased rather than reduced in the resistors, inductors, and capacitors, respectively, could be
case of full exclusion, and that pssac and pdsac were equal rather described by
than reduced compared to systemic pressures in the presence q̇R (t) = R
of endoleak. Finally, Schurink et al. [16] found the intrasac - ṗ(t); q̇L (t) = L
- p(t); q̇C (t) = C- p̈(t),
˜ ˜ ˜ ˜ ˜ ˜
pressure in fully excluded aneurysms to be fully obliterated (1)
rather than being non-zero.
with
In order to both reconcile and physically substantiate    
the various findings of the experimental endoleak studies 1 1 −1 1 1 −1
described in literature, in the current study, a computational R
- =R ; L
- =L ;
−1 1 −1 1
model of an incompletely excluded AAA is developed. After  
experimental validation, the model is applied to study the 1 −1
effects on the intrasac pressure after EVAR of AAA of the C- = C (2)
−1 1
degree of endoleak, the degree of stent-graft compliance, and
the resistance of a possible outflow tract formed by a branch- the element matrices, where R denotes the element resistance,
ing vessel. In addition to elucidating previous observations, L the element inductance, and C the element capacitance. By
the results of this study could help to assess the significance assembling the equations describing the behaviour of the dif-
of endoleak in individual patients using intrasac pressure ferent elements in the aortic tree model, using the property
measurements by means of implantable sensors (e.g., refs. that coinciding local nodes had equal pressures and bal-
[17,18]). Furthermore, the computational model developed anced flows, the set of equations describing the total model
in this study could help to provide boundary conditions for behaviour could be written as:
simulations of the effects of endoleak on the AAA wall stress
using computational methods based on 3D fluid–structure C- p̈(t) + R
- ṗ(t) + L
- p(t) = q̇(t), (3)
interaction (e.g., ref. [19]). ˜ ˜ ˜ ˜
where now p denotes the column of pressures in the globally
˜
numbered nodal points, q denotes the column of globally pre-
scribed external flows, and˜ C, R, and L contain the assembled
- - -
2. Methods element matrices.
The parameter values of the elements constituting the
2.1. Aortic tree model different modules of the aortic tree model were based on
literature. Of the segmental modules, the parameter values
Based on the arterial tree model by Westerhof et al. [20], were directly gathered from Westerhof et al. [20], who for
a lumped parameter model of the human aortic tree was the various segments of the arterial tree in Fig. 1(a) report
1108 B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118

Fig. 1. Layout of the aortic tree model with an excluded AAA in the presence of an endoleak channel and an outflow tract (a), and configurations of the
segmental modules (b), stent-graft modules (c), and Windkessel modules (d). Resistors are indicated by their resistance R, inductors by their inductance L, and
capacitors by their capacitance C. The variables psac and psys denote the intrasac pressure and the systemic pressure measured in the aorta, respectively.

the laminar flow resistance, the fluid inertance, and the wall E the elastic modulus, and h the wall thickness [22]. The
compliance. These properties were assigned to respectively compliance was determined by integrating (4) along the
the resistor, the inductor and the two capacitors depicted in axis of an aneurysm of an assumed sinusoidal shape. The
Fig. 1(b), where C1 and C2 each represent half of the total remaining element parameter values that were considered as
compliance C. Of the Windkessel modules, the parameters free model parameters were the compliance of the stent-graft
were determined such that the module input impedance modules, the resistance of the element representing the
approximated that of the represented vascular bed (e.g., [21], outflow tract, and the resistance of the element representing
see Appendix A). the endoleak channel.
The parameter values of the remaining elements were The actual parameter values are given in Appendix B. To
either determined based on geometrical and physical assump- determine these values, the desired mean and pulse pressures
tions, or considered as free model parameters. Based on the that were applied in the definition of the Windkessel module
assumption that the modelled stent-graft was of the same parameters as described in Appendix A were defined as 100
shape as the excluded section of the abdominal aorta in and 40 mmHg (or 13 and 5.3 kPa), respectively. Further,
healthy state, the resistance and the inductance of the stent- with a mean flow rate at the inlet of 95 ml/s, based on
graft modules were based on the properties of the abdominal a physiologic flow signal as depicted in Fig. 2, the time
aortic segments of the original model by Westerhof et al. [20]. constant τ as defined in the appendix was determined as
The compliance of the element representing the aneurysm 1.36 s. Finally, the aneurysm compliance was determined
was based on the variational area-to-pressure relation of thin- as 0.65 ml/kPa, based on assumed minimal and maximal
walled tubes diameters of 20 and 55 mm, respectively, and a length
of 85 mm, which was 80 percent of the length of the
∂A 2πa3 (1 − ν2 )
= , (4) infrarenal abdominal aorta in the model by Westerhof et al.
∂p Eh [20]. Other properties that were applied to determine the
where A denotes the tube cross-sectional area, p the internal compliance were ν = 0.5, E = 2.7 MPa [23], and h = 2.0 mm
pressure, a the tube radius, ν the wall material Poisson ratio, [24,25].
B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118 1109

input of the tube was connected to a piston pump driven by


a computer-controlled linear motor (ETB32, Parker Hannifin
Corp., Cleveland, OH), drawing fluid from a collecting sys-
tem filled with water. By means of an adjustable outflow
resistance, the output of the tube was also connected to the
collecting system. In addition, the tube was connected to
a Windkessel model consisting of a silicone tube partially
filled with air that was locally constricted by means of an
adjustable clamp to provide a resistance at its inlet. The
aneurysm was excluded from the main circulation by means
of an inserted glass tube. An endoleak was modelled by a
small channel connecting the main tube to the aneurysm sac.
The endoleak channel comprised two small PU tubes that
Fig. 2. Flow rate at the inlet of the physiologic model (after ref. [33]).
were interconnected by a replaceable glass mid-section. By
applying mid-sections of various diameters, the size of the
2.2. Experimental setup channel could be varied.
In both the aneurysm and the setup main tube, the pressure
To validate the lumped parameter modelling method, an response to the flow prescribed at the inlet could be measured
experimental setup of an incompletely excluded aneurysm by means of pressure catheters (PressureWire-4, Radi Medi-
was developed (see Fig. 3(a and b)). The aneurysm was mod- cal Systems AB, Uppsala, Sweden). Further, in both the main
elled as a local dilatation of a poly-urethane (PU) tube. The tube and the endoleak channel, fluid flow rates could be mea-

Fig. 3. Experimental setup of an incompletely excluded abdominal aortic aneurysm: photographic image (a), schematic representation (b), and lumped parameter
model (c). The variables psac and psys denote the intrasac pressure and the systemic pressure measured in the main tube, respectively.
1110 B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118

channel present. The air chamber compliance was determined


based on Boyle’s law. With respect to the endoleak chan-
nel, the different mid-sections were of 50 mm length and
of 0.85 mm (small), 1.6 mm (medium), and 2.4 mm diam-
eter (large). Including the side tubes of 35 mm length and
4.0 mm diameter, the channel resistance values were again
based on (5). Further, the resistance at the main tube outlet
was determined by dividing the mean of the measured main
tube pressure by the mean of the measured systemic flow. The
resistance at the silicone tube inlet was determined experi-
mentally, by subjecting the tube to a pressure difference and
measuring the resulting flow.

Fig. 4. Flow rate at the inlet of the experimental setup: measured in the
experiment (dashed) and prescribed in the simulation (solid). 2.3. Validation and simulations

sured by means of ultrasonic flow meters (TS420, Transonic Using the setup of Fig. 3(a), an experiment was performed
Systems Inc., Ithaca, NY). The piston pump was adjusted in which the pressure response to the prescribed flow at the
such that the prescribed mean flow was 1.2 l/min at 70 bpm, inlet was measured in both the main tube and the modelled
with a flow signal as depicted in Fig. 4. The constriction of the aneurysm. The experiment was performed both in the absence
silicone tube and the outflow resistance were adjusted such of an endoleak channel and in the presence of the small,
that the systemic pressure varied between 10 and 20 kPa. medium, and large sized channels. At the start of the experi-
Based on the experimental setup, an additional lumped ment, the measured pressures were initialised at 0 kPa. During
parameter model was developed (see Fig. 3(c)). Of this the experiment, the pressures were monitored to check for dif-
model, the segmental module resistance and inductance val- ferences between consecutive periods. The actual recordings
ues were based on physical balance laws applied to tubular were only performed after a periodic steady state had been
sections of the setup subjected to fully developed Newtonian reached.
flow (e.g., ref. [22]). In this respect, the parameter R in (1) The experiment was numerically simulated using the
was regarded as the laminar flow resistance, defined by lumped parameter model of Fig. 3(c). After passing the sig-
nal through a low-pass Butterworth filter in order to reduce
8ηl
R= , (5) noise, the measured systemic flow was prescribed as bound-
πa4 ary condition at the inlet of the model (see Fig. 4). At the
with η the fluid viscosity, l the section length, and a the section points connected to the standard reference, a zero-pressure
radius. The parameter L was regarded as the inertance of the was prescribed. After specification of the initial conditions
fluid contained by the section, defined by
p(t = 0) = 0; ṗ(t = 0) = 0; p̈(t = 0) = 0, (7)
ρl ˜ ˜ ˜
L= , (6) ˜ ˜ ˜
πa2 the pressure response of the model was determined by
with ρ the fluid density. The actual parameter values were numerical integration of Eq. (3) using a Newmark differ-
based on the physical properties of water (η = 1.0 mPa s, ence scheme implemented in Matlab (The MathWorks, Inc.,
ρ = 1.0 × 103 kg/m3 ). Further, since the exact wall thickness Natick, MA). The procedure was repeated for the resis-
and mechanical properties of the PU models were unknown, tance values representing the small, medium, and large sized
the compliance values of the modules representing the main endoleak channels. To compensate for possible pressure
tube sections were determined experimentally. By subjecting losses due to bends and sudden changes in flow area that were
the tube to an internal pressure and measuring the additionally not incorporated in the calculation of these resistances, the
stored fluid volume in the absence of an endoleak channel, the values were scaled such that the simulated endoleak channel
main tube total compliance was estimated. From this value, flow rates roughly matched those in the experiment. Simi-
the compliance per unit length was determined, which was larly, the measured value of the main tube compliance was
then used to calculate the compliance per section. The glass adjusted in order to match the simulated systemic pulse pres-
tube and connecting plastic tubes were considered rigid. sure to the systemic pulse pressure that was measured, and
The dimensions of the different tubular sections of the the aneurysm compliance was adjusted to match the simu-
setup upon which the segmental module properties were lated and measured values of the pulse pressure inside the
based along with the properties of the remaining elements aneurysm. The scaling factors are given in Table C.2.
are given in Appendix C. The modelled aneurysm was of Finally, the lumped parameter model of the physiologic
80 mm length and 55 mm maximal diameter. The compli- situation in Fig. 1 was applied to simulate the in vivo aor-
ance of the aneurysm was estimated using an additional tic and intrasac pressures after EVAR of AAA, and to assess
pressurisation experiment of the main tube with an endoleak the effects on these of the degree of endoleak, the degree
B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118 1111

of stent-graft compliance, and the resistance of the outflow Table 1


tract. The boundary conditions, initial conditions, and solu- Measured (meas.) and simulated (sim.) mean systemic and intrasac pressures
in the experimental setup in the absence of an endoleak channel and in the
tion procedure were again as described above, where the flow presence of small, medium, and large-sized channels
prescribed at the inlet was based on a measured physiologic
Endoleak channel Systemic pressure (kPa) Intrasac pressure (kPa)
flow signal from literature (see Fig. 2). Using variations of
the main model, three types of simulation were performed. In Meas. Sim. Meas. Sim.
the first type, the effect of endoleak was ignored by removal None 13.3 13.3 −0.3 0.0
of the element representing the endoleak channel. In the Small 14.3 14.3 13.7 14.2
second type, the effect of the outflow tract was ignored by Medium 14.4 14.4 13.7 14.4
Large 14.0 14.0 13.3 14.0
removal of the outflow tract element. In the last type, the stent-
graft was assumed rigid by removal of the stent-graft module
capacitors. In each type of simulation, results were obtained
for various combinations of the remaining two degrees of
pressure signals (see Fig. 5). In the absence of an endoleak
freedom to be studied. Each of the two parameters was expo-
channel, in both the numerical model and the experiment, psac
nentially varied within a specific interval defined around a
is negligible compared to psys . Further, the measured and sim-
standard value. The lower and upper boundaries of the inter-
ulated systemic signals have equal amplitudes. The simulated
val were defined in proportion to the standard value, at 10
signal, however, is smoother than the measured signal, and
and 1000 percent, respectively. With regard to the stent-graft
the two present a mutual phase difference of about 0.05 times
compliance, the standard value was defined at 10 percent of
the time period. In the presence of an endoleak channel, for
the AAA compliance, which amounted to 0.65 ml/kPa. The
all channel sizes, p̄sac approximates p̄sys . Yet, while the two
standard resistance value of both the endoleak and the outflow
are about equal in the numerical model, in the experiment,
tract was defined at 100 times the highest resistance present
p̄sac is slightly lower than p̄sys (see Table 1). In addition, the
in the original model, which amounted to 6.3 Pa s/ml.
systemic pressure signal amplitude or pulse pressure (psys )
is slightly larger in the numerical model than in the exper-
iment. The endoleak channel flow rates in the two models
3. Results have similar amplitudes. Again, however, the simulated sig-
nals are smoother than the measured signals. Additionally,
The lumped parameter model of the experimental setup minor shape differences are present at the signal maximum
yields good approximations of the experimentally measured and near the end of the time period. Positive flow rates indi-

Fig. 5. Validation of the lumped parameter model: measured (solid) and simulated (dashed) signals in the absence of an endoleak channel (a) and in the presence
of small (b), medium (c), and large-sized (d) channels. (Top) Systemic (thick) and intrasac (thin) pressures. (Bottom) Endoleak channel flow rate.
1112 B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118

Fig. 6. Simulated pressures in the absence of endoleak: intrasac pressure mean (a) and amplitude (b) as a function of both outflow tract resistance and stent-graft
compliance, and systemic (solid) and intrasac (dashed) pressures for the limiting cases A–D, as indicated (c).

cating flow into the modelled aneurysm are present while a different shape from psys , with a generally lower amplitude
psys is higher than psac . During that period, psac is increased. (see Fig. 6, case A).
Outflow from the aneurysm and a decreasing psac are present In the absence of an outflow tract, p̄sac is constant at a value
while psac is higher than psys . The variations of the flow rate of p̄sys for all combinations of endoleak resistance and stent-
signals are generally in phase with the variations of psys . Both graft compliance, while psac again varies as a function of
the flow rate signal amplitude and the intrasac pulse pressure both the studied parameters (see Fig. 7(a and b)). The latter
(psac ) are increased with increasing endoleak channel size. is increased for both increasing stent-graft compliance and
Yet, in all cases, pssac is decreased compared to pssys , and pdsac decreasing endoleak resistance. At minimal endoleak resis-
is higher than pdsys . tance, it approaches a maximal constant value. The limiting
In the physiologic situation, the absence of endoleak yields cases of psac are a signal of almost zero-amplitude at maximal
a mean intrasac pressure p̄sac that equals zero for all combi- endoleak resistance and minimal stent-graft compliance (see
nations of stent-graft compliance and outflow tract resistance, Fig. 7, case D) and signals with an amplitude of almost psys
while the intrasac pressure amplitude psac varies as a func- at minimal endoleak resistance (see Fig. 7, cases A and B).
tion of both these parameters (see Fig. 6(a and b)). The At maximal compliance and maximal resistance, psac has the
latter is increased for increasing stent-graft compliance. At shape of psys with a decreased amplitude (see Fig. 7, case C).
higher compliance values, it is further increased for increas- In case of a rigid stent-graft, both p̄sac and psac are func-
ing outflow tract resistance. The limiting cases of psac that are tions of both the endoleak resistance and the outflow tract
obtained at the boundaries of the parameter value intervals resistance (see Fig. 8(a and b)). The value of p̄sac varies
applied in this study are zero-pressure signals at minimal between approximately zero and p̄sys , and is increased for
stent-graft compliance (see Fig. 6, cases B and D), and a both decreasing endoleak resistance and increasing outflow
signal mimicing psys with a lower amplitude at maximal com- tract resistance. The value of psac varies similarly, although
pliance and maximal outflow tract resistance (see Fig. 6, case the increase with increasing outflow tract resistance is less
C). At maximal compliance and minimal resistance, psac has prominent at higher endoleak resistance. The limiting cases
B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118 1113

Fig. 7. Simulated pressures in the absence of an outflow tract: intrasac pressure mean (a) and amplitude (b) as a function of both endoleak resistance and
stent-graft compliance, and systemic (solid) and intrasac (dashed) pressures for the limiting cases A–D, as indicated (c).

of psac are a zero-pressure signal at minimal outflow tract the effects on the intrasac pressure after EVAR of AAA of
resistance and maximal endoleak resistance (see Fig. 8, case the degree of endoleak, the degree of stent-graft compliance,
D) and a signal with an amplitude of almost psys and a and the outflow tract resistance.
mean of p̄sys at maximal outflow tract resistance and minimal The experimental validation shows that the lumped
endoleak resistance (see Fig. 8, case A). In combination with parameter modelling method is capable of representing the
the maximal outflow tract resistance, the maximal endoleak intrasac pressure variations in an incompletely excluded
resistance yields a signal with a very small amplitude at a AAA. The intrasac mean pressure and pressure oscillations
mean of about half the value of p̄sys (see Fig. 8, case C). At as well as the flow through the endoleak channel can be
minimal outflow tract resistance and minimal endoleak resis- captured. It should be noted that, to obtain these results, the
tance, psac equals approximately half psys (see Fig. 8, case outflow resistance of the numerical model was determined
B). based on the mean systemic pressure and flow measured in the
experiment. Additionally, the endoleak channel resistance,
the main tube compliance and the aneurysm compliance
4. Discussion were scaled in order to match measured results. These adjust-
ments, however, do not undermine the practical value of the
In order to both reconcile and physically substantiate model. The imposed outflow resistance can be considered as
various findings of experimental endoleak studies described merely a boundary condition similar to the prescribed flow at
in literature, a computational model of an incompletely the input. In this respect, the scaling of the endoleak channel
excluded AAA has been developed. Based on the arterial tree resistance, main tube compliance, and aneurysm compli-
model by Westerhof et al. [20], a lumped parameter model ance seems more of an infringement. Still, this poses no
of the human aortic tree has been generated, incorporating problem as long as the model is used in parameter variation
an excluded AAA in the presence of both an endoleak studies.
channel and an outflow tract. After validation by means of The differences found between the measured and sim-
an experimental setup, the model has been applied to study ulated results in the experimental validation may indicate
1114 B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118

Fig. 8. Simulated pressures in the absence of stent-graft compliance: intrasac pressure mean (a) and amplitude (b) as a function of both endoleak and outflow
tract resistance, and systemic (solid) and intrasac (dashed) pressures for the limiting cases A–D, as indicated (c).

that for application of the model in exact simulations rather indicate the need for incorporating a gravitational effect.
than parameter variation studies, some model properties and Since the mean difference is in the order of 0.65 kPa (see
boundary conditions need to be optimised. For instance, the Table 1), while the height difference bridged by the endoleak
need for scaling the endoleak channel resistance may indicate channel is around 6.5 cm, the difference is suspected to be
that the non-linear effects that were neglected in determining due to gravitational forces that need to be overcome during
its value have a considerable effect that, in that case, should the transfer of fluid from the main tube to the aneurysm. Yet
still be incorporated (see, e.g., ref. [26]). Further, since a non- again, in parameter variational studies, the current model
optimal choice of filter parameters may induce amplitude and remains valid.
phase changes to the filtered signal, the model response may The results of the physiologic model show that by means
be improved by optimisation of the Butterworth filter applied of lumped parameter models of the human aortic tree, the
to the inlet flow signal. In the current study, such amplitude effects of endoleak can be simulated in a physiologic context.
changes may be the reason for the shape differences in the It makes the model extendable to simulate experiments
endoleak channel flow rates, and probable, less apparent concerning type II endoleak, in which there is persistent
differences in the systemic and intrasac pressures. The effect flow into the aneurysm sac through collateral branches (e.g.,
on the overall amplitudes may be limited, however, due to refs. [27–29]). In that case, by connecting the branches
the scaling of the model response. Phase changes, on the to other parts of the circulation, the collateral flow could
other hand, cannot have been induced by the filter, since this be generated rather than prescribed as originating from a
was of the zero-phase type. Possibly, phase shifts observed in separate source. The current study, however, focuses on
the results are due to limited spatial discretisation. Since the endoleak originating from the aortic lumen. In this study,
amplitude and phase changes observed in the current study neither of the endoleak resistance, outflow tract resistance,
are small, they do not invalidate the model for application in and stent-graft compliance is based on physiologic mea-
parameter variation studies. Finally, the measured difference surements. The representing parameter values, however,
between p̄sac and p̄sys not present in the simulation, may are varied in such an approach, that the intrasac mean and
B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118 1115

pulse pressure values cover almost full ranges between et al. [12], Chong et al. [10], and Mehta et al. [15], is consistent
zero and the systemic mean and pulse pressure values, with the current finding that for higher degrees of stent-
respectively. This ensures that physiologic model parameter graft compliance, the systemic pulse pressure is transmitted
values are included in the study. Nevertheless, it should be through the stent-graft to the aneurysmal sac. Also the finding
noted that, especially towards the boundaries of the applied of Schurink et al. [16] that the intrasac pressure is obliterated
parameter value intervals, not all modelled situations will in fully excluded aneurysms is consistent with the current
be realistic. See, e.g., Fig. 8, case B, where the endoleak results, if it supposed that a relatively stiff stent-graft was
resistance and the outflow tract resistance both are at such applied.
low levels, that the systemic pressure cannot be sustained. Findings of previous studies that could not be confirmed
Such limiting behaviour can still be helpful, however, in the may indicate that for the simulation of measurements on
understanding of the general phenomena associated with experimental setups, dedicated model configurations and
endoleak. boundary conditions are necessary. For instance, the find-
By combining the results of the different physiologic ing of Schurink et al. [16] that with endoleak in the absence
situations simulated, it can be concluded that the presence of an outflow tract pssac and p̄sac are decreased compared to
of endoleak leads to elevated intrasac pressure, the mean of pssys and p̄sys , may indicate that the conditions applicable to
which is mainly governed by the outflow tract resistance, the experiment were different from the conditions currently
while the pulse pressure is governed by both the endoleak simulated. Based on the current model, it would be expected
resistance and the stent-graft compliance. In general, the for p̄sac and p̄sys to be equal. In case of a net mean pres-
effect of a higher degree of endoleak is an increased sure difference, the endoleak channel would be subject to a
intrasac pulse pressure (see Figs. 7(b) and 8(b)). The general net mean flow that would not vanish until the two pressures
effect of a higher degree of stent-graft compliance is a had levelled. Similarly, an increased p̄sac compared to p̄sys
more pronounced transmission through the stent-graft to the as found by Parodi et al. [12] could not be confirmed either.
aneurysmal sac of the systemic pulse pressure (see Fig. 6(b)). In the current model, again the resulting pressure difference
In the absence of endoleak, a higher degree of stent-graft would result in a net mean flow that would not vanish until the
compliance leads to an increased intrasac pulse pressure at two values had levelled. Finally, the observation of pdsac and
zero mean pressure (see Fig. 6, case C). In the presence of pssac values both higher than pssys as observed by Mehta et al.
endoleak it has a similar effect, in which the mean is at the [15] remains unconfirmed. Based on the current model, due
level of the systemic mean pressure (see Fig. 7, case C). The to leakage back into the main circulation, all intrasac pres-
general effect of a decreased outflow tract resistance is mainly sures would be expected to level at least somewhere below
a reduction of the intrasac mean pressure in the presence of pssys .
endoleak (see Fig. 8(a)). Additionally, it may result in atten- In the interpretation of the results of this study, it should
uation of the intrasac pulse pressure resulting from either be noted that there may be additional factors affecting
the presence of endoleak (see Fig. 8(b)) or a high degree the aneurysm intrasac pressure that were not studied or
of stent-graft compliance (see Fig. 6(b)). Based on these incorporated in the numerical model. One of these is the
observations, it seems that, in order to minimise the intrasac compliance of the aneurysm itself, which in turn is influ-
pressure, the most favourable situation is one of maximal enced by the aneurym volume and its wall thickness. Both
resistance of a possible endoleak combined with minimal these factors have experimentally been shown to affect
stent-graft compliance and minimal outflow tract resistance. the intrasac pressure increase, in which an increased vol-
Based on the agreement of the current results with previ- ume and a decreased wall thickness each have a damping
ous findings, it can be concluded that the lumped parameter effect [13,30]. Further, it is not unlikely that the compli-
modelling method provides a useful numerical tool for vali- ance is also affected by the presence of thrombus inside the
dating experimental endoleak studies. The finding that, in the aneurysm, by means of either an influence on the amount
presence of endoleak, p̄sac equals p̄sys [7,9,10], is consistent of fluid that is contained, or a contribution to the total stiff-
with the results of the current study in case there is no outflow ness. Additionally, the presence of thrombus may influence
tract. Further, the findings that the intrasac pressure pulsatility the intrasac pressure increase by affecting the resistance
is inversely related to the endoleak channel diameter [7,10] to intrasac fluid flow. Although the presence of throm-
and is obliterated by thrombosis [7] are consistent with the bus itself does not necessarily reduce the pressure acting
current finding that psac is inversely related to the endoleak on the aneurysm wall [8], a resulting decreased intrasac
resistance. This finding also substantiates the findings on the flow may induce complete aneurysm thrombosis, which is
reduction of pssac in case of thrombosed endoleak channels by related to a decrease of both the intrasac pressure pulsatil-
Mehta et al. [11]. Further, the reducing effect of the presence ity and the intrasac mean pressure [7]. Finally, the degree
of an outflow tract on the intrasac pressure as found by Parodi of the intrasac pressure decrease due to the presence of
et al. [12] and Chong et al. [10] is consistent with the current the outflow tract will be dependent on the tract’s refer-
finding that p̄sac and psac are reduced for decreased values ence pressure. In a physiologic situation this will be the
of the outflow tract resistance. Finally, the intrasac pressure pressure in the connected vascular bed, and a high pres-
pulsatility in fully excluded aneurysms as observed by Parodi sure may even lead to type II endoleakage back into the
1116 B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118

aneurysm. Although Chong et al. [10] for a certain degree Table A.1
of flow through the tract observed no influence of the out- Distribution of the systemic flow among the body parts represented by the
Windkessel modules of the physiologic lumped parameter model (based on
flow pressure, this will be different if the tract flow is left to ref. [34])
be influenced by this pressure by means of a constant tract
Body part Fraction of the systemic flow
resistance.
Upper body
Brain right side and right arm 0.19
Brain left side 0.07
5. Conclusion Left arm 0.12
Liver 0.27
The lumped parameter modelling method is capable of Kidneys 0.22
representing the intrasac pressure variations in an incom- Legs 0.13
pletely excluded AAA. By means of lumped parameter
models of the human aortic tree, the effects of endoleak
can be simulated in a physiologic context. It is con- Finally, the parameter C accounts for the local peripheral
cluded that the presence of endoleak leads to elevated compliance. Its value is determined from the time constant τ
intrasac pressure, the mean of which is mainly governed associated with the Rp ,C-circuit, defined by
by the outflow tract resistance, while the pulse pressure is
governed by both the endoleak resistance and the stent- τ = Rp C. (A.5)
graft compliance. Based on the agreement of the current
results with previous findings, it is further concluded that This so-called RC-time indicates at what rate the modelled
the lumped parameter modelling method provides a use- fluid volume stored in the capacitor is released through the
ful numerical tool for validating experimental endoleak resistor. Of each of the different Windkessel modules, the
studies. RC-time is set equal to the RC-time of the total system τ s ,
defined by

τs = Rs Cs , (A.6)
Appendix A. Windkessel module parameters
with Rs again the systemic peripheral resistance as defined
The parameter Rc in Fig. 1(d) accounts for the resistance
by (A.4) and Cs the total systemic compliance. The systemic
of the vessel segment just distal to the connected tree end. Its
RC-time indicates the rate of decay of the systemic pres-
value equals that of the characteristic impedance Zc (see ref.
sure during diastole. Its value is determined by estimating
[31]), defined by
the value of Cs using the pulse pressure method proposed

by Stergiopulos et al. [32]. In this procedure, using a least-
Lc
Zc = , (A.1) squares optimisation technique, Cs is determined such that
Cc the pressure response of an Rs ,Cs -circuit to the flow sig-
with Lc and Cc the inertance and the compliance, respec- nal to be prescribed at the input of the aortic tree model
tively, of the connected tree end last segment. Further, attains both the desired mean pressure p̄ and a desired pulse
Rp accounts for the resistance of the local periphery. Its pressure p.
value is determined as the remainder of Zc subtracted
from the total resistance of the represented vascular bed
Rv : Appendix B. Physiologic model properties

Rp = Rv − Zc . (A.2) The segmental module properties of the lumped param-


eter model depicted in Fig. 1 are given in Table B.1. The
The value Rv is computed as a fraction of the systemic
Windkessel module properties are given in Table B.2.
peripheral resistance Rs :
Rs
Rv = , (A.3)
fq Appendix C. Experimental model properties
with fq the fraction of the systemic flow associated with the
The dimensions of the different tubular sections of the
considered body part (see Table A.1). The value of Rs is
experimental setup depicted in Fig. 3 upon which the seg-
computed as a desired systemic mean pressure p̄ divided by
mental module properties of the lumped parameter model
the mean flow q̄ of a flow signal prescribed at the input of the
depicted in Fig. 3(c) were based are given in Table C.1. The
model:
properties of the remaining elements of the model along with
p̄ associated scaling factors as applied in the simulations are
Rs = . (A.4)
q̄ given in Table C.2.
B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118 1117

Table B.1
Resistance (R), inductance (L), and capacitance (C) of the segmental modules of the physiologic lumped parameter model (after ref. [20])
Module Artery R (105 Pa s/m3 ) L (105 Pa s2 /m3 ) C (10−11 m3 /Pa)
1 Aorta ascendens 0.0327 0.309 53.4
2 Aorta ascendens 0.0358 0.322 51.0
3 Arcus aorta 0.0971 0.533 29.6
4 Arcus aorta 0.227 1.14 52.1
5 Aorta thoracalis 0.397 1.74 59.7
6 Aorta thoracalis 1.62 3.82 25.1
7 Aorta thoracalis 2.31 4.18 22.5
8 Aorta abdominalis 2.88 4.72 20.4
9 Aorta abdominalis 1.06 0.716 3.62
10 Stent-graft 4.22 2.86 Variable
11 Stent-graft 4.75 3.62 Variable
12 Aorta abdominalis 1.19 0.904 3.16
13 A. iliaca communis 24.4 14.3 6.88
14 A. iliaca externa 62.8 23.0 3.88
15 A. iliaca externa 27.0 9.94 1.72
16 A. anonyma 2.96 1.76 13.5
17 A. carotis com. sin. 14.4 24.0 7.12
18 A. subclavia 6.35 8.11 5.60
19 A. coelica 2.20 3.30 1.36
20 A. renalis 15.8 53.5 1.67

Table B.2
Characteristic resistance (Rc ), peripheral resistance (Rp ), and capacitance (C) of the Windkessel modules of the physiologic lumped parameter model
Module Body part Rc (108 Pa s/m3 ) Rp (108 Pa s/m3 ) C (10−10 m3 /Pa)
A Brain right side and right arm 0.468 7.39 19.6
B Brain left side 1.42 20.0 7.28
C Left arm 1.06 11.7 12.8
D Liver 1.27 52.0 34.6
E Kidneys 3.08 12.8 14.0
F Legs 2.84 21.6 7.23

Table C.1
Length (l) and radius (r) of different tubular sections of the experimental
setup
Section Material l (mm) r (mm)
1 Plastic 35 12.5 Table C.2
2 PU 25 12.5 Resistance (R) and capacitance (C) of different elements of the experimental
3 PU 25 12.5 lumped parameter model along with associated scaling factors as applied in
4 PU 35 12.5 the simulations
5 PU 35 12.5
Property Value Factor
6 Glass 40 12.0
7 Glass 30 12.0 R (108 Pa s/m3 )
8 Glass 30 12.0 Endoleak channel
9 Glass 30 12.0 Small 39 2.0
10 Glass 40 12.0 Medium 3.2 5.0
11 PU 35 12.5 Large 0.73 10
12 PU 35 12.5
Main tube outlet 7.5 –
13 Plastic 35 12.5
Silicone tube inlet 3.0 –
14 Plastic 40 7.0
15 Plastic 40 7.0 C (10−9 m3 /Pa)
16 Plastic 40 7.0 Main tube (per m) 9.9 0.84
17 Plastic 40 7.0 Aneurysm 2.3 0.57
18 Plastic 40 7.0 Air chamber 3.9 –
1118 B.J.B.M. Wolters et al. / Medical Engineering & Physics 29 (2007) 1106–1118

References abdominal aneurysm: implications on pressure and imaging—an in


vitro study. J Vasc Surg 1998;28(2):234–41.
[1] Veith EJ, Baum RA, Ohki T, Amor M, Adiseshiah M, Blankensteijn [17] Ellozy SH, Carroccio A, Lookstein RA, Minor ME, Sheahan CM,
JD, et al. Nature and significance of endoleaks and endotension: sum- Juta J, et al. First experience in human beings with a permanently
mary of opinions expressed at an international conference. J Vasc Surg implantable intrasac pressure transducer for monitoring endovascular
2002;35(5):1029–35. repair of abdominal aortic aneurysms. J Vasc Surg 2004;40(3):405–12.
[2] White GH, Yu W, May J, Chaufour X, Stephen MS. Endoleak as a [18] Milner R, Verhagen HJM, Prinssen M, Blankensteijn JD. Noninvasive
complication of endoluminal grafting of abdominal aortic aneurysms: intrasac pressure measurement and the influence of type 2 and type 3
classification, incidence, diagnosis, and management. J Endovasc Surg endoleaks in an animal model of abdominal aortic aneurysm. Vascular
1997;4(2):152–68. 2004;12(2):99–105.
[3] White GH, May J, Waugh RC, Chaufour X, Yu W. Type III and type [19] Wolters BJBM, Rutten MCM, Schurink GWH, Kose U, de Hart
IV endoleak: toward a complete definition of blood flow in the sac after J, van de Vosse FN. A patient-specific computational model of
endoluminal AAA repair. J Endovasc Surg 1998;5(4):305–9. fluid–structure interaction in abdominal aortic aneurysms. Med Eng
[4] Baum RA, Carpenter JP, Cope C, Golden MA, Velazquez OC, Neschis Phys 2005;27(10):871–83.
DG, et al. Aneurysm sac pressure measurements after endovascu- [20] Westerhof N, Bosman E, de Vries CJ, Noordergraaf A. Analog studies
lar repair of abdominal aortic aneurysms. J Vasc Surg 2001;33(1): of the human systemic arterial tree. J Biomech 1969;2:121–43.
32–41. [21] Stergiopulos N, Westerhof BE, Westerhof N. Total arterial inertance as
[5] Dias NV, Ivancev K, Malina M, Resch T, Lindblad B, Sonesson the fourth element of the Windkessel model. Am J Physiol 1999;276(1
B. Intra-aneurysm sac pressure measurements after endovascular Pt 2):81–8.
aneurysm repair: differences between shrinking, unchanged, and [22] Milnor WR. Hemodynamics. 2nd ed. Baltimore: Williams and Wilkins;
expanding aneurysms with and without endoleaks. J Vasc Surg 1989.
2004;39(6):1229–35. [23] Di Martino ES, Guadagni G, Fumero A, Ballerini G, Spirito R, Biglioli
[6] van Marrewijk C, Buth J, Harris PL, Norgren L, Nevelsteen A, Wyatt P, et al. Fluid–structure interaction within realistic three-dimensional
MG. Significance of endoleaks after endovascular repair of abdom- models of the aneurysmatic aorta as a guidance to assess the risk of
inal aortic aneurysms: The EU-ROSTAR experience. J Vasc Surg rupture of the aneurysm. Med Eng Phys 2001;23(9):647–55.
2002;35(3):461–73. [24] Raghavan ML, Vorp DA. Wall stress distribution on three-
[7] Schurink GW, Aarts NJ, Van Baalen JM, Kool LJ, Van Bockel JH. dimensionally reconstructed models of human abdominal aortic
Experimental study of the influence of endoleak size on pressure in aneurysm. J Vasc Surg 2000;31(4):760–9.
the aneurysm sac and the consequences of thrombosis. Br J Surg [25] Thubrikar MJ, Labrosse M, Robicsek E, Al-Soudi J, Fowler B. Mechan-
2000;87(1):71–8. ical properties of abdominal aortic aneurysm wall. J Med Eng Technol
[8] Schurink GW, van Baalen JM, Visser MJ, van Bockel JH. Thrombus 2001;25(4):133–42.
within an aortic aneurysm does not reduce pressure on the aneurysmal [26] Idelchik IE. Handbook of hydraulic resistance. Boca Raton, FL: CRC
wall. J Vasc Surg 2000;31(3):501–6. Press, Inc.; 1994.
[9] Skillern CS, Stevens SL, Piercy KT, Donnell RL, Freeman MB, Gold- [27] Dayal R, Mousa A, Bernheim J, Hollenbeck S, Henderson P, Prince M,
man MH. Endotension in an experimental aneurysm model. J Vasc Surg et al. Characterization of retrograde collateral (type II) endoleak using
2002;36(4):814–7. a new canine model. J Vasc Surg 2004;40(5):985–94.
[10] Chong CK, How TV, Gilling-Smith GL, Harris PL. Modeling endoleaks [28] Diaz S, Uzieblo MR, Desai KM, Talcott MR, Bae KT, Geraghty PJ, et
and collateral reperfusion following endovascular AAA exclusion. J al. Type II endoleak in porcine model of abdominal aortic aneurysm. J
Endovasc Ther 2003;10(3):424–32. Vasc Surg 2004;40(2):339–44.
[11] Mehta M, Ohki T, Veith FJ, Lipsitz EC. All sealed endoleaks are not [29] Mousa A, Dayal R, Bernheim L, Henderson P, Hollenbeck S, Trocciola
the same: a treatment strategy based on an ex vivo analysis. Eur J Vasc S, et al. A canine model to study the significance and hemodynamics
Endovasc Surg 2001;21(6):541–4. of type II endoleaks. J Surg Res 2005;123(2):275–83.
[12] Parodi JC, Berguer R, Ferreira LM, La Mura R, Schermerhorn ML. [30] Gawenda M, Knez P, Winter S, Jaschke G, Wassmer G, Schmitz-
Intra-aneurysmal pressure after incomplete endovascular exclusion. J Rixen T, et al. Endotension is influenced by wall compliance in a latex
Vasc Surg 2001;34(5):909–14. aneurysm model. Eur J Vasc Endovasc Surg 2004;27(1):45–50.
[13] Gawenda M, Winter S, Jaschke G, Wassmer G, Brunkwall J. Endoten- [31] Westerhof N, Elzinga G, Sipkema P. An artificial arterial system for
sion is influenced by aneurysm volume: experimental findings. J pumping hearts. J Appl Physiol 1971;31(5):776–81.
Endovasc Ther 2003;10(6):1091–6. [32] Stergiopulos N, Meister JJ, Westerhof N. Simple and accurate way for
[14] Gawenda M, Jaschke G, Winter S, Wassmer G, Brunkwall J. Endoten- estimating total and segmental arterial compliance: the pulse pressure
sion as a result of pressure transmission through the graft following method. Ann Biomed Eng 1994;22(4):392–7.
endovascular aneurysm repair—an in vitro study. Eur J Vasc Endovasc [33] Olufsen MS, Peskin CS, Kim WY, Pedersen EM, Nadim A, Larsen
Surg 2003;26(5):501–5. J. Numerical simulation and experimental validation of blood flow
[15] Mehta M, Veith FJ, Ohki T, Lipsitz EC, Cayne NS, Darling III RC. in arteries with structured-tree outflow conditions. Ann Biomed Eng
Significance of endotension, endoleak, and aneurysm pulsatility after 2000;28(11):1281–99.
endovascular repair. J Vasc Surg 2003;37(4):842–6. [34] McArdle WD, Katch EL, Katch VL. Exercise physiology: energy, nutri-
[16] Schurink GWH, Aarts NJM, Wilde J, van Baalen JM, Chuter TAM, tion, and human performance. 4th ed. Baltimore: Williams and Wilkins;
Schultze Kool LJ, et al. Endoleakage after stent-graft treatment of 1996.

Вам также может понравиться