Computational Study of Mechanical Support to the Failing Total

Cavopulmonary Connection

Mauro Grigioni

, Giuseppe D’Avenio

, Salvatore Donatiello

and Antonio Amodeo

Technology and Health Department, Biomechanics and Rehabilitative Technologies Unit, Istituto Superiore di Sanità,

Viale Regina Elena 299, 00161 Rome, Italy

Department of Cardiac Surgery, Bambino Gesù Children’s Hospital, Rome, Italy

Keywords: Cardiovascular Surgery, Fontan Circulation, Bioengineering, Computational Fluid Dynamics (CFD).

Abstract: The performance of an axial flow blood pump in an idealized total cavopulmonary connection (TCPC)

model was intravascularly evaluated. This blood pump was inserted within a modified Fontan surgery using

a reinforced Gore-Tex conduit, to be connected to the caval veins with the pulmonary arteries. Two

different computational models were examined (i) the new geometric model without a pump and (ii) with

the pump. Computational fluid dynamics analyses of these models were performed to assess the hydraulic

performance under varying pump’s operating conditions. Numerical simulations indicate that the pump

generates a pressure distribution which could prove to be beneficial for the univentricular patient with

failing Fontan circulation, allowing to provide a possible intervention, at least as bridge to heart

transplantation or as end-stage pump implant.

1 INTRODUCTION

Patients with single ventricle anomaly, even though

this term groups very different pathologies, must

deal with the condition of having only one of the

two ventricles of adequate functional size. Some of

the anomalies described as single ventricle defects

include tricuspid atresia, hypoplastic left heart

syndrome, double inlet left ventricle and other

cardiac defects. The incidence of this heart defect

constitutes about 1-2 % of all congenital heart

defects (Samanek et al., 1999). Whenever there is

only one ventricle capable to pump blood efficiently,

the circulation must be reconfigured to maximize the

efficiency of this single ventricle without

overloading it.

The total cavopulmonary connection (TCPC)

(Giannico et al., 2006) represents one of the most

successful clinical options to obtain a sufficient lung

perfusion in single-ventricle patients. It consists of

the direct connection of the venae cavae to the

pulmonary arteries, avoiding the usual pathway of

blood through the right heart and then to the lungs,

on account of the dysfunctional ventricle. Even

though the survival to 15-20 years after surgery is

superior to 82% (Gersony, 2008), statistics are

nevertheless indicating that the risk of long-term

failure of TCPCs is remarkable, ultimately requiring

transplantation (Cromme-Dijkhuis et al., 1993).

Actually, the absence of the subpulmonary

ventricle in the Fontan patient induces an elevation

of pressure in the systemic venous circulation. The

central venous pressure (CVP) rises to a mean

pressure of about 12 mmHg, or even more in the

most unfavourable cases (it can reach as high as 20

mmHg).

Clearly, an unphysiologically high CVP is poorly

tolerated with time by the patients. In particular, it

has deleterious effects on the liver and the

splanchnic circulation, possibly resulting in protein-

losing enteropathy and plastic bronchitis (Feldt et

al., 1996), in the worst cases. At the liver level, the

elevated CVP may induce complex liver

dysfunction; consequently, the release of

angiogenesis factors is expected, favouring the

occurrence of venovenous anastomosis, pulmonary

venous fistulas, and aortopulmonary collateral

anastomoses (APCA).

Considering that a single ventricle must work

against both systemic and pulmonary compartments,

the ventricle itself faces a significant increase in

total systemic resistance. Hence, the systemic

ventricle undergoes hypertrophy, with elevated end-

diastolic pressure, which diminishes its diastolic

Grigioni, M., D’Avenio, G., Donatiello, S. and Amodeo, A..

Computational Study of Mechanical Support to the Failing Total Cavopulmonary Connection.

In Proceedings of the 3rd International Congress on Cardiovascular Technologies (CARDIOTECHNIX 2015), pages 43-48

ISBN: 978-989-758-160-1

performance (Cheung et al., 2000; Gewillig, 2005).

Also as a consequence of the elevated vascular

resistance, the ventricular preload is limited, which

also impairs diastolic performance.

As de Leval (1998) pointed out, the condition of

the Fontan patient is paradoxical, in that there is

systemic venous hypertension and simultaneously

pulmonary arterial hypotension.

Many attempts have been made to optimize the

Fontan connection, in order to make energy losses as

negligible as possible (e.g., Amodeo et al., 2002).

This notwithstanding, the failure of the Fontan

circulation is an occurrence that must always be

considered, with an increasing probability as a

function of the time elapsed since the operation

(Khairy et al., 2008).

Owing to the chronic shortage of heart donors, in

recent years the possibility of using mechanical

assistance for failing TCPC, either as bridge to

transplantation or destination therapy, has been

addressed in several studies.

In order to prevent ventricle hypertrophy induced

by elevated vascular resistance and limited

ventricular preload, it is natural to consider

leveraging on the VAD technology already available

to design therapeutic solutions for the unloading of

the only functional ventricle in the Fontan patient.

In this viewpoint, different connections and

assist devices have been proposed. Lacour-Gayet

(2009) and others suggested the inclusion of an

axial pump model used for circulatory support in an

extracardiac tube that connects caval veins to

pulmonary arteries, avoiding backflow in the

superior vena cava (SVC).

An assistance device positioned in the SVC was

found to increase the pressure in the PAs

(Santhanakrishnan et al., 2013).

A pump installed in the inferior vena cava (IVC)

is in principle capable of generating a strong

pressure decrease, upstream of the device itself,

possibly causing a collapse of the venous vessel. For

this reasons, this study considers the insertion of a

miniaturized axial pump inside a reinforced

GoreTex conduit which connects the caval veins

district to the pulmonary arteries district, in a

different fashion with respect to the classical Fontan

(TCPC) surgery. This connection, together with

properly set pump operating conditions, was thought

to improve the balance of arterial and venous

pressures, preventing also vessel collapse, thanks to

the surgical geometry and the use of a GoreTex

prosthesis.

The herein selected device is supposed to solve

the principal obstacles for long-term implantation.

Various studies showed how the Jarvik 2000 pump

might be among the major candidate for destination

therapy due to its biomechanical characteristics,

particularly for its very low hemolysis rate (Gibber

et al., 2010).

The goal of the collaboration between the

Biomechanics and Rehabilitative Technologies Unit

of the Technology and Health Dept. of ISS and the

Bambino Gesù Children’s Hospital (BGCH) is to

create a permanent solution to the failing Fontan

connection as both bridge to transplant and

destination therapy, avoiding the necessity of a

subsequent transplant for the patient. Then, we

selected an assist device whose long-term in vivo

performance is well documented by the literature,

representing a good reference model for a realistic

therapeutic intervention. In this paper we study by

CFD the feasibility of a surgical approach based on

both an innovative surgical connection and the use

of an axial pump model similar to the child-size

Jarvik 2000, as a permanent solution to sustain the

failing Fontan circulation.

The final aim of the investigation thus was to

investigate the fluid dynamics of TCPC innovative

connection making use of the mechanical support,

defining the axial pump’s condition for safe

performance (range of flow rates, pump speeds,

pulmonary artery pressures) to be related to the

clinical setting to provide indication for a safe

procedure in mechanical support of failing TCPC.

2 METHODS

This study of circulatory support device in the

Fontan circulation uses a model of TCPC

circulation, designed with an innovative connection

with respect to traditional Fontan, and a model of

cardiac pump resembling a pump already on the

market, widely used as a Ventricular Assist Device

(VAD), the child-size Jarvik 2000. The latter has the

essential characteristics (function, size and form

compatible with the insertion in a cylindrical tube)

that are required for our research, among those most

widely used in Europe.

The assistance device was positioned between

the two caval veins and the pulmonary arteries, as

shown in Fig 1. A second ideal geometric model

characterized by cylindrical ducts of intersection

between the caval veins and pulmonary arteries with

no pump was also created for CFD study of the

connection without mechanical support, in order to

evaluate pressure and flow fields characterizing the

connection used, with and without the pump.

CARDIOTECHNIX 2015 - International Congress on Cardiovascular Technologies

Thus, two 3D clinical Fontan compartment models

were created, using the computer aided design

(CAD) software SolidWorks (SolidWorks, Concord,

MA, USA), to study the properties of the flow inside

the proposed model:

- a first model with an axial pump, whose design

was inspired by the child-size Jarvik 2000 pump,

- a second model without the pump, in order to

have a basis for comparison.

Figure 1: Model of the TCPC with assist device. At the

left side the anastomosis between SVC and IVC is shown.

At the upper right side the two pulmonary arteries are

shown. A GoreTex conduit connects the caval veins and

the PAs, with the axial pump inserted in the conduit.

In order to study the Fontan fluid dynamics, a

vertically oriented tubular venous compartment of

about 20-cm length was chosen to represent both

IVC and SVC joined to the connected right-left

pulmonary arteries by an extracardiac conduit. The

pump model was positioned in the extracardiac

conduit. The total volume of fluid in the model was

approximately 83 cm

, vs. the 2.1-cm

internal

volume of the pump (priming volume), excluding

the head input. The model was meshed using Gambit

software, with almost 1,800,000 elementary volume

elements, one million of which covered the 2.1-cm

internal volume of the pump.

Ansys Fluent 12.1 software was used for fluid

dynamics simulations. To simulate the rotating part

of the pump and analyze the behavior of the blood

fluid during the stationary phase (when the pump is

operating at full capacity), we used the RNG k-ɛ

model, which has a better capability of studying

rotating flows with respect to the standard k-ɛ model

(Yakhot et al., 1992). This feature of the RNG k-ɛ

model stems from more accurate transport equations

for the turbulent kinetic energy (k) and rate of

dissipation of turbulent kinetic energy (ɛ), with

respect to the standard k-ɛ model.

The following performance conditions were

imposed: pump angular velocities were within the

child-size Jarvik 2000 range (4,000 - 18,000

revolutions per minute [rpm]) and 3 flow rate values

(2, 3, 4 l/min) were considered. The pulmonary

arterial mean pressures was set at 10 mmHg, and

with a constant 40%-60% SVC–IVC flow ratio was

considered, according to the physiological flow

partition seen in children (Fogel et al., 1999).

Vessel walls were modelled as rigid tubes. A

constant viscosity value of 0.0035 kg/m*s and fluid

density of 1,060 kg/m3 were used (Cutnell and

Johnson, 1998).

Animal Study – A preliminary investigation about

feasibility of the presented approach in vivo was

carried out on an animal model. All animals received

humane care in compliance with the “Guide for Care

and Use of Laboratory Animals”. The Bambino

Gesù Children’s Hospital Ethical Committee

approved all conditions for animal surgery and care.

A total of 8 sheep (Western breed, 42-48 Kg) were

considered: 2 for preliminary studies, 4 supported

and 2 non supported. Anesthetic drugs included

ketamine (3 mg/kg), diazepam (0.2 mg/kg) and

atropine (0.02 mg/Kg) and induction was obtained

with propofol 1% (2mg/Kg). Animals were

intubated and ventilated using a Servo 900C

volume-cycled respirator (Siemens®, Danvers, MA)

with 100% oxygen and 1% to 2% isoflurane.

Ventilation parameters were: 10 to 15 breaths/min

with tidal volumes of 10 ml to 15 mL/Kg and 4

cmH2O positive end expiratory pressure.

A 16 gauge femoral arterial line (Intracath®,

Becton Dickinson, Sandy, UT) was placed for

systemic blood pressure monitoring. A 16 gauge

femoral venous line was inserted in a jugular vein.

The heart was exposed trough a median

sternotomy. A fiber optic pulmonary catheter

(Opticath®, Abbott Laboratories, North Chicago,

IL) was placed in the main pulmonary artery. All

pressures and flows were continuously monitored

and recorded. CO, PVR, CI were calculated with

Picco® (Pulsion Medical System) using thermo-

dilution technique.

The SVC and IVC were sequentially divided

from the right atrium and TCPC was performed by

interposing a 16 mm polytetrafluoroethylene

vascular graft (Gore-Tex®) between the two cavae

and connected to the pulmonary artery through a

second conduit in a T- junction geometry (2 non

supported TCPC).

In the pump-supported group (4 animals), the

child-size Jarvik 2000 axial pump was inserted in

the conduit to the pulmonary artery. Flow rates were

maintained between 2 to 3 L/m in a range of 5.000,

7.000 and 9.000 rpm, at 1st , 2nd and 3rd hour of

support, respectively. The axial child flow pump was

Computational Study of Mechanical Support to the Failing Total Cavopulmonary Connection

positioned as distal as possible from venae cavae to

avoid their collapse. The graft diameters were

chosen to match the sheep’s SVC and IVC

dimensions. The main pulmonary trunk was left in

place to vent the blood from coronary sinus.

Hemodynamic variables were recorded over a period

of 3 hours.

Hourly arterial blood gas measurements were

obtained. In addition, oxygen saturations in PA and

IVC blood, serum lactate levels were measured

before and during operation

3 RESULTS

Fig. 2 reports the pressure values obtained in the

conduit, upstream of the device, at 1 cm downstream

of the IVC-SVC anastomosis. Comparing the curves

relative to the geometry with the pump with the line

representing the connection without assistance, it

can be seen that, for each imposed flow rate (2, 3

and 4 L/min), the pump does not constitute a

resistance to the flow, at a sufficiently high angular

velocity (e.g., from 10000 rpm upwards, for 3

l/min). Moreover, the pump can sustain flows of less

than 2 l/min at speeds below 10000 rpm.

Figure 2: Average intravascular pressure, at 1 cm

downstream of the IVC/SVC connection. The horizontal

line indicates the value of the calculated pressure that

would occur in the geometry without the pump, in the case

of a flow rate of 3 l / min (similar values for other flow

rates). Above the horizontal line the rotational speed is not

usable for blood perfusion, whilst the pump represents an

obstacle for the blood; thus the usable region for the pump

speed in relation to the available venous pressure is

comprised between the horizontal line and a pressure level

close to zero.

These results are encouraging, since the axial

pump is meant to be used in patients with low levels

of physiological flow rate (children at the age of a

few years).

The CFD study enabled also an assessment of the

Wall Shear Stress (WSS) values on the surface of

the different parts of the connection. As apparent

from Figures 3 and 4, a positive correlation was

found between regime conditions (pump angular

velocity and blood flow rate) and magnitude of

WSS.

Figure 3: WSS [dyn/cm

] distribution over the impeller’s

surface (pump angular velocity: 4000 rpm; blood flow

rate: 1 l/min).

Figure 4: WSS [dyn/cm

] distribution over the impeller’s

surface (pump angular velocity: 10000 rpm; blood flow

rate:3 l/min). The color code is the same as in Fig. 3.

It must be underlined, though, that at the highest

regime investigated the maximum WSS value was

6410 dy n/cm

, which is not an excessively high

value for shear stresses in physiological flows with

prosthetic devices. As an example, in Grigioni et al.

(2001) a value of around 5500 dyn/cm

was found

for the maximum Reynolds shear stress associated to

a bileaflet aortic valve available on the EU market.

Animal Study - The group without pump support had

a sudden deterioration of hemodynamic parameters

and they died within one hour. In the pump-

supported (PS) group all animals survived and

cardiopulmonary function was stable. No hemolysis

at different run speeds, neither thrombotic events nor

venous collapse was observed. In the PS group the

central venous pressure values did not increase

CARDIOTECHNIX 2015 - International Congress on Cardiovascular Technologies

during the mechanical assistance period, and were

found to be similar to those in baseline condition.

As for the gas exchange values, arterial pH remained

in the normal range, with slightly alkalosis after 3

hours, as an effect of moderate hypercapnia. A trend

in lactates stability was observed under mechanical

assist, which can be related to optimal hemodynamic

function.

4 DISCUSSION

Heart pumps are available to assist the cardiac

function in case of a pathological state. These

devices are mostly implanted to assist ventricular

function, hence they are called Ventricular Assist

Devices (VAD). They aspirate blood from the

ventricle and inject it into the aorta.

Apart from ventricle support, other uses of

circulatory assistance have been proposed. The use

of a pump for circulatory support proposed in this

paper aims to solve the serious circulatory problems

that quite often occur in patients previously operated

on with Fontan surgery. The simulations herein

presented show that it is possible to adapt an already

available commercial VAD for extracardiac

circulatory assistance. The original VAD connection

has been modified to insert the device in the

proposed connection. The CFD analysis allowed us

to determine the range of rotational speed that

should be imposed to avoid veins collapse. Fig. 2

can be used to gauge the upper limits of impeller

rotational velocity, in order to prevent venous

collapse. Considering an output pressure of 10

mmHg in the pulmonary arteries, the upper limits for

2 l/min, 3 l/min and 4 l/min are 10000, 12000 and

14000 rpm, respectively. When the flow is 2 l/min,

Fig. 2 shows that at 10000-rpm rotational speed the

pressure is -0.87 mmHg. This value is very close to

the limit value of -0.5 mmHg reported by Riemer et

al (2005) as a trigger for vein instability/collapse

upstream of the pump. The problem in this case can

be solved in two ways. The first one consists of

reducing the rotational speed of the impeller; in the

second, the length of the pipe connection can be

increased, hence, at equal flow, a greater power loss

will be obtained and the problem of excessive

negative pressure can be minimized.

In the present study, we considered a constant

rotational velocity of the pump. This was done for

two main reasons: 1) the axial pump we referred to,

the child-size Jarvik 2000, functions most of the

time at constant angular velocity. Actually, the pump

is restarted periodically, to mitigate the risk of

stationary flow zones inside the pump, but the

device is essentially a constant-speed pump. 2)

Information about pulsatility in assist devices is still

too scarce to implement a rational strategy of time-

varying pump velocity for the failing Fontan. This

notwithstanding, pulsatile assist devices in the future

might be certainly an interesting option, taking also

into account that pulsatility could reduce the risk of

venous collapse, upstream of the device.

In order to minimize the possibility of venous

collapse related to pump functionment, we tested the

insertion of the device inside a reinforced tube (a

GoreTex reinforced prosthesis). This solution avoids

the collapse of the vein (very likely in extreme

conditions) and prevents physiological or electronic

random changes that could lead to a temporary low

pressure with consequent collapse of the walls. The

CFD simulations made on the model without the

pump allowed us to calculate the pressures generated

by the new connection, using the same boundary

conditions for the model with the pump. To analyze

the data provided by simulation with the pump

inserted, we considered the pressure calculated in

the model without the pump as the reference

pressure, at a flow rate of 3 l/min (Fig. 2). A first

comparison between the pathlines (data not shown)

of the model without the pump and the one with the

pump showed how the device provides an

improvement in blood flow in the immediate vicinity

of the anastomosis. The presence of the pump

caused a reduction of recirculation region on the

wall of the cava that is opposite to the junction

between caval veins and conduit; this effect was due

to the presence of a suction force generated by the

pump which linearizes the flow. In the vessel

downstream of the pump, spiral flow trajectories

could be seen, caused by the torque generated by the

impeller on blood flowing through the device.

Before reaching the pulmonary arteries, the pathlines

showed fairly linear trajectories, which demonstrate

the effectiveness of the flow straighteners of the

pump. Wall Shear Stresses are relevant if an end-

stage implant is thought to be provided, thus WSS

calculated values allowed us to verify that,

predictably, the maximum values were correlated to

pump speed and blood flow rate. In any case, the

WSS values did not reach excessively high levels,

confirming that the proposed study is a feasible

approach to the treatment of the failing Fontan

circulation as destination therapy.

The favourable role of the pump-assisted Fontan

circulation, besides the results of the in-silico study,

has been also confirmed by a preliminary animal

study. Hence, we are confident that the growing

Computational Study of Mechanical Support to the Failing Total Cavopulmonary Connection

number of the Fontan patients with impaired

function of the single ventricle will be offered in the

future the possibility to avoid or defer as much as

possible heart transplantation, by means of suitably

designed mechanical assistance to circulation.

REFERENCES

Amodeo A., Grigioni M., Oppido G., Daniele C.,

D'Avenio G., Pedrizzetti G., Giannico S., Filippelli S.,

Di Donato R. M., 2002. The beneficial vortex and best

spatial arrangement in total extracardiac

cavopulmonary connection. J Thorac Cardiovasc Surg;

124(3):471-8.

Cheung Y. F., Penny D. J., Redington A. N., 2000. Serial

assessment of left ventricular diastolic function after

Fontan procedure. Heart;83:420–4.

Cromme-Dijkhuis A. H., Hess J., Hahlen K., et al., 1993.

Specific sequelae after Fontan operation at mid and

long-term follow-up J Thorac Cardiovasc Surg.

106:1126-1132.

Cutnell, John & Johnson, Kenneth, 1998. Physics, Fourth

Edition. Wiley: 308.

de Leval M. R., 1998. The Fontan circulation: what have

we learned? What to expect? Pediatr Cardiol; 19:316 –

20.

Feldt R. H., Driscoll D. J., Offord K. P. et al., 1996.

Protein-losing enteropathy after the Fontan operation.

J Thorac Cardiovasc Surg; 112:672– 80.

Fogel, M. A., P. M. Weinberg, J. Rychik, A. Hubbard, M.

Jacobs, T. L. Spray, and J. Haselgrove, 1999. Caval

contribution to flow in the branch pulmonary arteries

of Fontan patients with a novel application of

magnetic resonance presaturation pulse. Circulation

99:1215–1221.

Gersony, W. M., 2008, Fontan Operation After 3 Decades:

What We Have Learned. Circulation.; 117:13-15.

Gewillig, M., 2005. The Fontan circulation. Heart;

91:839–46.

Giannico, S., Hammad F, Amodeo A, Michielon G,

Drago F, Turchetta A, Di Donato R, Sanders SP, 2006.

Clinical outcome of 193 extracardiac Fontan patients:

the first 15 years. J Am Coll Cardiol. 47:2065-73.

Gibber, M., Wu Z. J., Chang, W., Bianchi G., Hu, J.,

Garcia J., Jarvik R., Griffith, B. P., 2010. In Vivo

Experience of the Child-Size Pediatric Jarvik 2000

Heart: Update. ASAIO Journal; 56(4):369-376.

Grigioni, M.1., Daniele C., D'Avenio G., Barbaro V.,

2001. The influence of the leaflets' curvature on the

flow field in two bileaflet prosthetic heart valves. J

Biomech. 34(5):613-21.

Khairy, P., Fernandes S. M., Mayer J. E. Jr. et al., 2008.

Long-term survival, modes of death, and predictors of

mortality in patients with Fontan surgery. Circulation;

117:85–92.

Lacour-Gayet F. G., Lanning C. J., Stoica S., Wang R.,

Rech B. A., Goldberg S. and Shandas R., 2009. An

Artificial Right Ventricle for Failing Fontan: In Vitro

and Computational Study. The Annals of Thoracic

Surgery. 88:170-176.

Riemer R. K., Amir G., Reichenbach S. H., Reinhartz O.,

2005. Mechanical support of total cavopulmonary

connection with an axial flow pump. J. Thorac

Cardiovasc Surg 130:351-4.

Samanek M., Voriskova M., 1999. Congenital heart

disease among 815,569 children born between 1980

and 1990 and their 15-year survival: a prospective

Bohemia survival study. Pediatr Cardiol 20:411–417.

Santhanakrishnan A., Maher K. O., Tang E., Khiabani R.

H., Johnson J., Yoganathan A. P., 2013.

Hemodynamic effects of implanting a unidirectional

valve in the inferior vena cava of the Fontan

circulation pathway: an in vitro investigation. Am J

Physiol Heart Circ Physiol. Nov 15;305(10):H1538-

47.

Yakhot et al., 1992. Development of turbulence models

for shear flows by a double expansion technique,

Physics of Fluids A; 4(7):1510-1520.

CARDIOTECHNIX 2015 - International Congress on Cardiovascular Technologies