RIJKSUNIVERSITEIT GRONINGEN

Strategies for optimisation of paediatric cardiopulmonary

bypass

PROEFSCHRIFT

ter verkrijging van het doctoraat in deMedische Wetenschappen

aan de Rijksuniversiteit Groningenop gezag van de

Rector Magnificus, dr. F. Zwarts,in het openbaar te verdedigen op

woensdag 12 februari 2003om 16.00 uur

door

Filip Maria Jan Jozef De Somergeboren op 10 mei 1960

te Aalst (België)

Promotores: Prof. dr. T. Ebels

Prof. dr. G. Van Nooten

Co-promotor: Prof. dr. P. Verdonck

Beoordelingscommissie: Prof. dr. R.Berger

Prof. dr. H.J.Busscher

Prof. dr. M. Hazekamp

ISBN 90-423-0210-0

Voor Caroline en Casper

Voor mijn ouders

© Copyright Shaker Publishing 2002All rights reserved. No part of this publication may be reproduced, stored in aretrieval system, or transmitted, in any form or by any means, electronic,mechanical, photocopying, recording or otherwise, without the priorpermission of the publishers.

Printed in The Netherlands.

ISBN 90-423-0210-0Shaker Publishing BVSt. Maartenslaan 266221 AX MaastrichtTel.: 043-3500424Fax: 043-3255090http:// www.shaker.nl

Contents

1

Contents

Chapter 1: Introduction 3

Chapter 2: Vascular access for total body perfusion 9

Chapter 3: Circuit design 21

Chapter 4: Oxygenation by artificial lung systems 33

Chapter 5: Systemic inflammatory response 43

Chapter 6: Summary and new prospectives 59

Appendix 1 Evaluation of different paediatric venous cannulas using

gravity drainage and VAVD: an in vitro study

Perfusion, 2002; 17(5): 321 – 326

63

Appendix 2 Hydrodynamical Comparison of Aortic Arch Cannulae

Int. J. Art. Organs, 1998; 21(11): 705 – 713

83

Appendix 3 Comparison of two dissimilar designs of paediatric aortic

cannulae

Int. J. Art. Organs, 2002, 25(9): 867 – 874

115

Appendix 4 D-901 Neonatal oxygenator: a new perspective

Perfusion 1994; 9: 349 – 355

141

Appendix 5 Low extracorporeal priming volumes for infants: a benefit?

Perfusion 1996; 11: 455 – 460

159

Appendix 6 Hydrodynamic characteristics of artificial lungs

ASAIO, 2000; 46(5): 532 – 535

175

Contents

2

Appendix 7 Impact of oxygenator design on hemolysis, shear stress,

white blood cell and platelet count

J. Cardiothor.Vasc. Anesth. 1996; 10: 884 - 889

195

Appendix 8 Can an oxygenator design potentially contribute to air

embolism in CPB. A novel method for the determination of

the air removal capabilities of neonatal oxygenators

Perfusion, 1998; 13: 157 – 163

219

Appendix 9 In vivo evaluation of a phosphorylcholine coated

cardiopulmonary bypass

Journal of Extra-corporeal technology, 1999; 31 (2): 62-67

241

Appendix 10 Phosphorylcholine coating of extracorporeal circuits

provides natural protection against blood activation by the

material surface

European Journal of Cardio-Thoracic Surgery, 2000;

18(5): 602 – 606

261

Appendix 11 Tissue factor as main activator of the coagulation system

during cardiopulmonary bypass

The Journal of Thoracic and Cardiovascular Surgery,

2002; 123: 951 – 958

283

Nederlandse samenvatting 309

Dankwoord 315

Curriculum vitae 317

Chapter 1

3

Chapter 1 Introduction and aim of the thesis

The mortality associated with the repair of congenital heart defects in early life

has decreased considerably over the years. However improved survival has

unmasked a whole spectrum of morbidity associated with the practice of

cardiopulmonary bypass [1].

As a general concept, cardiopulmonary bypass will temporarily bypass heart

and lungs. This is achieved by introducing one or two venous cannulas in the

venae cavae that direct venous return of the patient, by means of plastic

tubing, into a reservoir. This reservoir replaces the compliance of the veins.

From the reservoir blood is pumped through an artificial lung or oxygenator.

The oxygenator heats or cools the blood and maintains physiologic blood

gases. Subsequently the oxygenated blood is guided through an arterial filter

and re-infused by means of an arterial cannula into the aorta. All these

components need to be primed before cardiopulmonary bypass can be

started. Apart of this life support, the circuit is designed to meet specific

surgical needs. Most systems have one or more aspiration lines for the

recuperation of blood losses in the surgical field, the unloading of the left

ventricle and aspiration of blood from additional blood vessels such as a left

superior vena cava or collateral blood vessels. In many institutions the

cardioplegia delivery is also integrated into the cardiopulmonary bypass

circuit.

During conduct of paediatric cardiopulmonary bypass quite drastic changes

occur. Due to haemodilution by priming solutions and cardioplegia, the

haematocrit varies between 20 – 35%. Most operations require a certain

Chapter 1

4

amount of hypothermia. Depending on the specific procedure the actual blood

temperature might vary between 15 and 38° C. As a consequence of these

temperature and haematocrit changes, viscosity will change and thus

influence tissue perfusion. Also blood flows will change depending on the

surgical procedure from circulatory arrest to high flow (up to 150 mL/kg) in the

rewarming phase.

It is often assumed that a paediatric cardiopulmonary bypass circuit is a

miniaturised adult system. This is not correct. In contrast to adults the priming

volume of even the smallest paediatric circuits will equal or exceed the total

blood volume of a baby. At the same time blood of the child will be exposed to

at least four times more foreign surface relative to an adult. The unique

physiology of the neonate and his sometimes aberrant anatomy, leads to

technical limitations and, therefore, makes the design and conduct of a

dedicated paediatric cardiopulmonary bypass complicated.

The combination of a new-born at one hand and open-heart surgery and

cardiopulmonary bypass at the other hand is quite challenging. The new-born

is a fast developing organism with immature organs within which the organic

systems are developing or maturing at different rate. Open-heart surgery and

cardiopulmonary bypass represent an extreme stress to the functioning of

these developing systems. Moreover, the response of those organs to this

stress will be different from what is reported in adults. Children are definitively

more prone to inflammatory response. Also neurological consequences of the

developing brain are different from those observed in the developed or

degenerating brain.

Chapter 1

5

The small size of vascular and cardiac structures not only challenges surgical

skills but also limit the possibilities for obtaining an optimal vascular access

and a bloodless surgical field.

Due to this unique anatomical and physiological environment specially

designed components have been developed. This research and development

is expensive and will often reach the end spectrum of technical know how.

Unfortunately, most of the time some industries are reluctant to invest in the

paediatric domain because of the small numbers compared to the huge

amount of adult cardiac procedures performed yearly.

Further research is also required to investigate the long and short-term

influence of different surgical strategies and techniques for conducting

cardiopulmonary on the different organ systems. Recent research clearly

demonstrates a correlation between conduct of cardiopulmonary bypass and

morbidity [2-6].

However, as pointed out by Jonas and Elliott [1], the consequences of a badly

conducted paediatric cardiopulmonary bypass should not be underestimated

as it may impact several decades. The child’s quality of life is likely to be

markedly diminished. Yet that is only part of the potential disaster. Children

have parents and relatives. Each will be affected by the poor outcome of

cardiopulmonary bypass. One bypass disaster can ruin many lives.

Chapter 1

6

Aim of the thesis

The aim of this thesis is to address different aspects of paediatric

cardiopulmonary bypass in detail and to propose modifications in order to

reduce cardiopulmonary bypass related morbidity and by doing so, improve

patient outcome. We will focus on four major items: (1) vascular access, (2)

mass transfer and fluid dynamics of oxygenators, (3) circuits and (4) whole

body inflammatory reaction.

• The small vascular structures of the new-born demand a better design

description of the geometry and fluid dynamic characteristics of cannulas.

There is not only a need for a better validation of today’s cannulas but also

for research into the relation between the hemodynamic characteristics of

these cannulas and possible damage to blood elements.

• The oxygenator is prone to less optimal flow, due to its tortuous flow path,

its large foreign surface area and the rapid changes in blood velocity

resulting in non-optimal mass transfer and activation of the whole body

inflammatory response. Additionally, most oxygenators have a priming

volume that is too high compared to the total blood volume of a new-born.

There is an urgent need for smaller, more blood compatible oxygenators,

with optimisation of their fluid mechanics and gas exchange in order to fit

the paediatric needs. These needs will include the capability for achieving

subnormal arterial oxygen tensions in cyanotic children without

compromising the high oxygen consumption of children during rewarming.

• Most circuits today have been designed based on empirically derived data.

This results in large volumes in the arterial and venous lines as well as in

Chapter 1

7

the aspiration lines. The use of an arterial line filter is highly recommended

although it is not used in an appropriate way in most institutions.

• Finally, the use and conduct of a paediatric cardiopulmonary bypass will

end in a mild or more pronounced whole body inflammatory reaction. The

strength of this reaction will vary from child to child, the equipment used,

and the conduct of the bypass.

We will propose techniques and strategies to overcome or to reduce these

problems and by doing so to ameliorate the cardiopulmonary bypass related

morbidity.

References

1. RA Jonas, MJ Elliott. Cardiopulmonary bypass in neonates, infants and

young children. Butterworth-Heinemann, Oxford 1994.

2. S Daniel. Review of the multifactorial aspects of BioInCompatibility in CPB.

Perfusion, 1996; 11: 246-255.

3. DT Pearson, RF Carter, MB Hammo, PS Waterhouse. Gaseous micro-

emboli during open heart surgery. In: Towards safer cardiac surgery. Ed.

DB Longmore. Lancaster, MTP Press, 1981: 325-354.

4. JM Pearl, DW Thomas, G Grist, JY Duffy, PB Manning. Hyperoxia for

management of acid-base status during deep hypothermia with circulatory

arrest. Ann Thorac Surg 2000; 70: 751-755.

5. RA Jonas, DC Bellinger, LA Rappaport et al. Relation of pH strategy and

development outcome after hypothermic circulatory arrest. J Thorac

Cardiovasc Surg. 1993; 106: 362-368.

Chapter 1

8

6. T Shin’oka, D Shum-Tim, PC Laussen et al. Effects of oncotic pressure

and haematocrit on outcome after hypothermic circulatory arrest. Ann

Thorac Surg 1998; 65: 155-164.

Chapter 2

9

Chapter 2 Vascular access for total body perfusion

2.1. Introduction

This chapter introduces the limitations and boundary conditions of vascular

access in paediatric cardiopulmonary bypass. The different requirements for

venous and arterial access are reviewed. Finally, the hydrodynamic

characteristics and different evaluation methods are presented and discussed.

Recommendations for an optimal communication between manufacturer and

clinician are given.

2.1.1. Problems related to vascular access

Unsuccessful cannulation may lead to cerebral complications [1-3] A

malpositioned aortic cannula may obstruct cerebral blood flow, or it may

cause a preferential flow into the descending aorta and “steal” blood from the

brain’s circulation [3]. Alternatively, obstruction by the superior vena caval

cannula may decrease cerebral venous drainage and potentially lead to brain

dysfunction [3]. A direct correlation between age and cerebral alterations (low

cerebral blood flow velocity and EEG slowing) caused by malpositioning of the

cannulas has been reported [3].

2.2. Venous access

Cannulation of the venous side of the circulation aims at draining the venous

blood from the central veins or right heart cavities in a laminar flow without

inducing any marked change of the pressure within the large veins. Only then

an adequate forward flow can be established. The entire venous return to the

Chapter 2

10

heart should be able to pass through the chosen cannulas without obstruction

and without damaging the blood vessel [4].

An essential problem of venous drainage is a compliance and geometric

mismatch. Wide, low-resistance, collapsible vessels are connected to smaller,

less compliant, artificial conduits. When suction is applied to the venous

reservoir, flow starts to increase linearly, but once the vessel starts to

collapse, the flow will stagnate. Increase in suction force beyond a critical

level, therefore, cannot increase the amount of venous drainage. Additionally,

high resistance in the drainage tube necessitates higher degrees of suction

than is needed with short, wide tubing. Maintenance of a positive pressure at

the tip of the cannula broadens the range of flow regulation because it

prevents venous collapse [5]. Reduced venous drainage may be due to

reduced venous pressure, inadequate height of the patient above the venous

reservoir, malposition of the venous cannulas or obstruction or excess

resistance of the lines and cannulas. Venodilation or hypovolaemia may

cause inadequate venous pressure.

2.3. Arterial access

Cannulation of the arterial side of the circulation must provide an adequate

forward flow of blood to the patient. The cannula and its placement must not

be non-obstructive and flow must be directed to the distal aorta in order to

perfuse all areas of the body.

The ideal cannula will generate sufficient flow without obstructing or damaging

the blood vessel.

Chapter 2

11

2.4. Cannula characteristics

2.4.1. Design related problems

The choice of the best cannula for a given procedure is not simple. In general,

manufacturers do not mention in their information brochures the internal

diameter of a cannula but only the outer diameter. Depending on the

production process, the wall thickness of comparable cannulas can be quite

different although their respective manufacturers measured identical outer

diameters [6]. Additionally, production tolerances result in important

differences in internal diameter even between cannulas of identical size and

manufactured by the same company. Since the pressure-flow relation highly

depends on the inner diameter and cannulas standard used in paediatric

cardiopulmonary bypass have small diameters, this results in significant

deviations of the mean values given by the manufacturer.

Another difficulty is related to the fact that the pressure-flow characteristic of a

cannula is always measured for water (low viscosity and Newtonian fluid).

Unfortunately, it is difficult to extrapolate water values towards blood (higher

viscosity and non Newtonian fluid) flow conditions.

2.4.2. Available data for clinicians

Manufacturers only report the polynomial regression of the water data of a

certain number of cannulas (Figure 1). Thus, the user has no information

about of the possible variability range. This is demonstrated in Figure 1 where

both the polynomial regression (full line) as given by the manufacturer and the

measured data of ten cannulas (dots) are depicted.

Chapter 2

12

Figure 1: Pressure-flow relationship for two paediatric arterial cannulas

0.0 0.5 1.0 1.5

0

100

200

300

Pres

sure

dro

p [m

mH

g]

DLP 77108

Water flow [L/min]

0.0 0.5 1.0 1.5

0

100

200

300

DLP 75008

2.4.3. Theoretical relationship

For a horizontal straight tube the relation between pressure and flow can be

described by Poiseuilles formula:

QRµLP

=∆ 4

8π

UD

µLP

=∆ 2

32

where µ = dynamic viscosity [N/m².s], L = length [m], R = radius [m], Q =

average flow [m³/s], U = mean velocity [m/s], D = diameter [m].

For cannulas this formula cannot be used since most cannula are not straight

tubes.

2.4.4. Practical characterisation

Several attempts have been described to predict the clinical performance of

cannulas.

Chapter 2

13

(1) Montoya et al. propose a system in which any vascular access device can

be characterised by a single number denoted as “M” which may be

determined from the geometry and/or from simple in vitro pressure-flow

measurements [7-9]. M is defined as log (LDC-4.75) where L represents the

length and DC the characteristic diameter of the cannula. The Dc is also known

as hydraulic diameter for non-circular ducts representing the diameter of a

corresponding circular orifice. The method can be used to choose the best

possible cannula when a given diameter or pressure may not be exceeded

during the procedure.

Unfortunately, the method has some disadvantages. In order to obtain the M-

number on a non-uniform design, such as a cannula, one has to do in vitro

measurements. The M-number also assumes that the flow regimen is

turbulent. However the obtained value is not useable in clinical practice,

especially if it is obtained by water measurements. Water measurements tend

to lie in the turbulent region while the blood flows used during clinical use are

in the laminar region. The latter limits its use in open-heart surgery [10].

(2) Another approach is based on the theory of dynamic similarity [6,11-12].

Flows become identical if the Reynolds number, a measure of the ratio

between inertial and viscous forces, is identical for both fluids [6] in the

experimental set-up (e.g. water) and in the clinical situation (blood).

Re = =UD Q

Dνρ

µπ4

with ρµ

ν =

Where Q = flow [m³/s], ρ = density [kg/m³], μ = dynamic viscosity [N/m² s], D =

diameter [m], ν =kinematic viscosity [m²/s], U = mean velocity [m/s].

Chapter 2

14

For Reblood = Rewater :

water

bloodblood =

νν

waterQQ

The pressures for a given water flow can be transformed to those of blood in

an analogue way by using the Euler number, a measure of the ratio between

pressure and inertial forces:

2

42

16² QPD

UPEu

ρπ

ρ∆

==

Where P = pressure [Pa]

For Eublood = Euwater:

2

=

water

blood

water

bloodwaterblood

UUPP

ρρ so that

2

=

water

blood

water

bloodwaterblood PP

µµ

ρρ

The dimensionless numbers Reynolds and Euler are independent of the fluid

physical properties. This allows converting directly flow rates and pressures.

In order to apply this technique one has to know the rate of the densities and

the rate of the dynamic and kinematic viscosity of both fluids. Since water

tests are performed at room temperature water density is approximately 1000

kg/m³ (998.2019 kg/m³) and water kinematic viscosity 1 10-6 m²/s (1.0038 10-6

m²/s).

If we compare water data with blood at a temperature of 37°C and a

haematocrit of 33.5% we obtain the following pressure and flow conversion

factors presented in Table 1. The factors in table 1 are calculated using the

formulas presented in section 3.1.2.3.

Chapter 2

15

Table 1. Pressure and flow conversion factors

Qblood/Qwater Pblood/Pwater

T = 37°C 2.43 6.21

T = 20°C 3.40 12.19

Flows and pressures measured during water tests are multiplied with these

factors to obtain corresponding blood flows and pressures.

(3) A third method rescales the coefficients of the fitted parabolic equation

between pressure drop (∆P) and flow rate (Q)

waterwaterwaterwaterwater QbQaP +=∆ 2

to blood

bloodbloodbloodbloodblood QbQaP +=∆ 2

For a given awater, bwater and the relationship between pressure and flow one

can determine ablood and bblood as:

waterwater

bloodblood aa

ρρ

=

waterwater

bloodblood b

µµb =

Table 2. Conversion factors for coefficients a and b

ablood/awater bblood/bwater

T = 37°C 1.055 2.56

T = 20°C 1.055 3.59

The factors in Table 2 are derived from Table 1 taking into account awater

blood

ρρ

Chapter 2

16

ratio of 1.03.

In Figure 3 a comparison of both methods (calculation based on dynamic

similarity and the parabolic method) is presented. There is still a deviation

from the measured data but it gives an estimate of what can be expected

under given conditions. The deviation is due to the low accuracy of water

measurements caused by the error range on pressure transducers and flow

meters. These errors are subsequently multiplied with the conversion factors

resulting in even larger deviations. This also explains why the deviation of the

calculated data is smaller at 37°C than at 20°C. Use of water-glycerin

solutions by manufacturers for validation of their cannulas instead of water will

reduce the error.

0.0 0.2 0.4 0.6 0.8 1.0Blood flow [L/min]

0

75

150

225

300

Pres

sure

dro

p [m

mH

g]

Dynamic similarityMeasuredParabolic method

DLP 7700820°C - Hct 33.5%

0.0 0.2 0.4 0.6 0.8 1.00

50

100

150

200

250

300Dynamic similarityMeasuredParabolic method

37°C - Hct 33.5%

Chapter 2

17

2.4.5. Quantification of blood damage

Pressure-flow relationships do not give direct information regarding the

possible damage of blood elements when a given cannula is used. It is not

necessarily the cannula with the highest pressure drop that will generate most

damage. The exerted shear rate and specifically shear stress in combination

with the duration of these forces (residence time) are far more important

factors for blood cell damage [13]. Shear stress equals fluid dynamic viscosity

multiplied by shear rate.

ru

δδµτ = with u the axial velocity component and r the radial variable

or

LRPw2

∆=τ

where τw = shear stress [N/m²], R = radius [m], L = length [m]

As tube length is usually several orders of magnitude greater than radius,

pressure is generally orders of magnitude greater than shear stress [14].

Physiological values of shear stress range from 1 – 50 dynes/cm² 1[14].

Most actual cannulas will easily generate shear stresses of several hundred

dynes/cm² [15], which is far above the trigger values of 75 and 100 dynes/cm²

[14,16] needed to activate white blood cells and platelets, respectively.

1 ²

1²

10mN

cmdyne

=

Chapter 2

18

2.5. Conclusions

Vascular access in neonates and small infants remains a major challenge for

adequate paediatric cardiopulmonary bypass. Small vascular structures,

congenital malformations and technical limitations in the manufacturing of

cannulas give rise to specific problems. A better documentation of the

pressure-flow relationship of a cannula in combination with its shear stress

data will help the clinician in choosing the best cannula for a given procedure.

Thus manufacturers should provide more adequate information regarding the

pressure-flow characteristics and both the inner and outer diameter of their

products.

References

1. FH Kern, PR Hickey. The effects of cardiopulmonary bypass on the brain.

In: Cardiopulmonary bypass in neonates, infants and young children. Eds:

RA Jonas, MJ Elliott. Butterworth-Heinemann, Oxford 1994: 263-281

2. RA Rodriguez, G Cornel, L Semelhago, WM Splinter, NA Weerasena.

Cerebral effects in superior vena caval cannula obstruction: the role of

brain monitoring. Ann Thorac Surg 1997; 64: 1820-1822.

3. RA Rodriguez, G Cornel, WM Splinter, NA Weerasena, CW Reid. Cerebral

vascular effects of aortovenous cannulations for pediatric cardiopulmonary

bypass. Ann Throac Surg 2000; 69: 1229-1235.

4. M Elliott. Canulation for cardiopulmonary bypass for repair of congenital

heart disease. In: Cardiopulmonary bypass in neonates, infants and young

children. Eds: RA Jonas, MJ Elliott. Butterworth-Heinemann, Oxford 1994:

128-140.

Chapter 2

19

5. PM Galletti, GA Brecher. Connection of the vascular system with an

extracorporeal circuit. In: Heart lung bypass; principles and techniques of

extracorporeal circulation. New York: Grune and Stratton; 1962: 171-193.

6. JF Douglas, JM Gaiorek, JA Swaffield, Part III Dimensional Analysis and

Similarity in Fluid Mechanics, 3rd ed., Longman Scientific & Technical,

Harlow, UK; 1985.

7. Delius RE, Montoya JP, Merz SI, McKenzie J, Snedecor S, Bove EL,

Bartlett RH. New method for describing the performance of cardiac

surgery cannulas. Ann Thorac Surg. 1992 Feb;53(2):278-81.

8. Sinard JM, Merz SI, Hatcher MD, Montoya JP, Bartlett RH. Evaluation of

extracorporeal perfusion catheters using a standardized measurement

technique--the M-number. ASAIO Trans. 1991 Apr-Jun;37(2):60-4.

9. Montoya JP, Merz SI, Bartlett RH. A standardized system for describing

flow/pressure relationships in vascular access devices. ASAIO Trans.

1991; 37(1):4-8

10. Kim WG, Park SS. Clinical application of the M-numbers of aortic cannulas

during hypothermic cardiopulmonary bypass in pediatric patients. Artif

Organs. 1999 Apr;23(4):369-72.

11. Uyttersprot N. “Stromingseigenschappen en bloedcompatibiliteit van

kindercanules.” Master of Science in Engineering, Thesis in Dutch, Ghent

University, 1999.

12. Verdonck P, Siller U, De Wachter D, De Somer F. Hydrodynamical

comparison of aortic arch cannulae. Int J Artif Organs, 1998; 21:705-713.

Chapter 2

20

13. LJ Wurzinger, R Opitz, P Blasberg, H Schmid-Schönbein. Platelet and

coagulation parameters following millisecond exposure to laminar shear

stress. Thrombosis and Haemostasis. 1985; 54: 381-386.

14. SM Slack, VT Turitto. Fluid dynamic and hemorheologic considerations.

Cardiovasc Pathol 1993; 2(3): 11S-21S.

15. F De Somer, L Foubert, M Vanackere, D Dujardin, J Delanghe, G Van

Nooten. Impact of oxygenator design on hemolysis, shear stress, white

blood cell and platelet count. J. Cardiothor.Vasc. Anesth. 1996; 10: 884-

889

16. LV McIntire, RR Martin. Mechanical trauma induced PMN leukocyte

dysfunction. In The Rheology of Blood Vessels and Associated Tissues

Eds Gross DR, Hwang NHC.. Alphen aan den Rijn: NATO Advanced

Study Institute Series - E, No 41, Sijthoff & Noordhoff, 1981

Chapter 3

21

Chapter 3 Circuit design

The cardiopulmonary bypass circuit consists basically of venous and arterial

(often including an arterial filter) tubing lines and an oxygenator with

integrated heat exchanger. This chapter deals with the hydrodynamic design

of the tubing and arterial filter. The artificial lung or oxygenator is discussed in

chapter 4.

3.1. Tubing

3.1.1. Priming volume

Once cardiopulmonary bypass is started, the volume in the arterial and

venous line as well as the priming volume of the oxygenator enlarges the total

circulating blood volume of the baby. Additionally, suction and vent lines that

are empty before starting cardiopulmonary bypass, remove an important

amount of blood out of the circulation once in use. Subsequently this blood is

returned into the circulation just before weaning cardiopulmonary bypass. As

a result important and rapid changes in circulating blood volume occur during

cardiopulmonary bypass. Because of this it is important to keep volumes in

the complete extracorporeal circulation as small as possible without

jeopardising flow requirements of the given lines. Its length and diameter

(Table 1) determine the volume of a line

Chapter 3

22

Table 1: Priming volumes for different tubing diameters

Tubing diameter 1

Inch mm

Priming volume per 10 cm of

length (mL)

1/8 3.17 0.792

3/16 4.76 1.781

1/4 6.35 3.167

3/8 9.53 7.126

1/2 12.70 12.668

3.1.2. Dimensions of the tubing

3.1.2.1. Introduction

The dimensions of the venous and arterial lines depend on the desired blood

flow rate and the height difference between table and oxygenator. When

gravity drainage is used a height difference between 30 and 40 cm is

generally accepted [1]. In many institutions sizing of tubing is established in

an empirical way. A more objective way is to decide based on fluid dynamic

parameters [2], thus limiting the dead volume in the aspiration lines to an

absolute minimum. The resulting reduction in priming volume results in less

homologous blood product utilisation [3,4].

3.1.2.2. Laminar or turbulent flow

Two types of steady flow of real fluids exist: laminar flow and turbulent flow

with a transition zone in between. Different fluid dynamic laws govern the two

types of flow.

1 1 inch = 25.4 mm

Chapter 3

23

In laminar flow, fluid particles move along straight, parallel paths in layers.

Magnitudes of velocities of adjacent layers are not the same. The viscosity of

the fluid is dominant and thus suppresses any tendency for turbulent

conditions due to the inertia of the fluid.

In turbulent flow, fluid particles move in a haphazard fashion in all directions.

The critical velocity is the velocity below which all turbulence is damped out by

the viscosity of the fluid. It is found that a Reynolds number of about 2000

represents the upper limit of laminar steady flow of practical interest. The

Reynolds number is a dimensionless number, representing the ratio of inertia

forces to viscous forces, in circular pipes [2].

νUD Re =

U = mean velocity [m/s], D = diameter [m], ν =kinematic viscosity [m²/s]

with

ρν

µ=

where ρ = density [kg/m³], µ = absolute blood viscosity [N/m² .s]

3.1.2.3. Blood viscosity

Dynamic viscosity of a fluid (µ) is either determined from literature data or

measured in a viscosity meter. Blood viscosity can be described by

exponential formula with:

100)273(

180064.5exp

+

+−=

Tplasmaµ

)31.2exp( Hctµµ plasma=

[ ])1(035.109.1 HctHct −+=ρ

Chapter 3

24

µplasma = plasma viscosity [N/m².s], T = absolute temperature [°C], Hct =

haematocrit [expressed as fraction]

Figure 1: Relationship between haematocrit, temperature and kinematic

blood viscosity

20 22 24 26 28 30 32 34 36 38Blood temperature [°C]

1.5

2.0

2.5

3.0

3.5

4.0

Bloo

d vi

scos

ity [x

10-6

N/m

².s]

Hct 36%Hct 34%Hct 32%Hct 30%Hct 28%Hct 26%Hct 24%Hct 22%Hct 20%

Blood viscosity calculation

Based on these calculations a nomogram can be constructed for a quick

estimate of blood viscosity when haematocrit and temperature are known

(Figure 1).

3.1.2.4. Pressure-flow relationship

In general the pressure drop can be calculated in function of diameter, length,

blood viscosity and height difference between patient and heart-lung machine,

using the equation:

Chapter 3

25

gU

DLfP

2

2

=∆

where f = friction factor, g = gravitational acceleration [m/s²] and

Re64

=f when flow is laminar.

However when the flow regimen is turbulent f is calculated using the

Colebrook equation:

+−=

fDf Re51.2

7.3log21 ε

with ε the roughness parameter.

Besides the Colebrook equation the Blasius formula is valid for smooth pipes

and low Reynold numbers. The friction factor becomes independent of the

roughness of the tube

41

Re316.0−

=f

By using these equations flow diagrams can be calculated for venous and

arterial lines in function of length, diameter, required blood flow, viscosity and

desired pressure drop.

3.1.2.5. Case study

If a baby needs cardiopulmonary bypass support one can calculate what

should be the appropriate diameter for both arterial and venous line. In our

example, the cardiopulmonary bypass circuit has an arterial and venous line

of 150 cm. The surgeon wants for this specific case a haematocrit of 30% and

no hypothermia during cardiopulmonary bypass. The maximum blood flow to

ensure adequate tissue perfusion is 700 mL/min.

Chapter 3

26

From Figure 2 we learn that both 3/16 and 1/4 inch arterial lines generate

laminar flow (shaded zone) for the given conditions. However, the pressure

loss over the arterial line will be approximately 20 mmHg higher if a 3/16 inch

diameter is chosen. This difference is acceptable so a 3/16 inch line gives the

best compromise between priming volume and pressure-flow characteristics.

Figure 2. Flow regimen in paediatric arterial lines

0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9Blood flow [L/min]

0

50

100

150

Pres

sure

dro

p [m

mHg

]

Characteristics of 3/16" and 1/4" arterial lines.

Reynolds < 20003/16 3/16 3/16 3/16

3/163/16

3/163/16

3/163/16

3/16

3/16

3/16

3/16

3/16

3/16

3/16

3/16

3/16

3/16

3/16

1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/4 1/41/4

1/41/4

Length: 150 cmTemperature: 37° CelsiusHaematocrit: 30%

Suppose it is decided to use a 3/16 inch venous line in the above described

case and the height difference between the operating table and the

oxygenator is 35 cm H20. We can determine the limitations of this choice by

using Figure 3. On the right Y-axis we notice that the Reynolds number

(squares), when using a haematocrit of 30% (X-axis) and a blood temperature

of 37°C, is below 2000 for a blood flow of 700 mL/min. The maximum blood

flow we can drain for these conditions (circles) is 770 mL/min (left Y-axis).

Chapter 3

27

This is approximately 10% higher than the maximum flow we anticipate. Thus,

a 3/16 inch venous line is a correct choice for this particular case.

Figure 3. Flow characteristics of a 3/16 inch venous line

20 22 24 26 28 30Hematocrit [%]

0.5

0.6

0.7

0.8

0.9

1.0

Bloo

d flo

w [L

/min

]

Blood flow at 37°CBlood flow at 20°CReynolds number at 37°CReynolds number at 20°C

500

1000

1500

2000

2500

Rey

nold

s nu

mbe

r

Characteristics of a 3/16" venous line

Tubing length: 150 cmHeight difference between oxygenator and patient: 35 cm H2O

It is important to notice that in Figure 2 and 3 paediatric cardiopulmonary

bypass blood flow is laminar up to 1 L/min, this in contrast to adult

cardiopulmonary bypass where blood flow is mostly turbulent. As a

consequence pressure losses will be smaller in paediatric cases and less

energy will be needed for generating a given blood flow.

3.3. Arterial filter

Arterial filters were introduced during the era of bubble oxygenators. Those

elderly generation oxygenators were well known sources of gaseous

Chapter 3

28

microemboli. At the end of the eighties membrane oxygenators became the

standard resulting in almost no gaseous microemboli. The removal of

gaseous microemboli by arterial filters is based on the concept of the bubble

trap and the bubble barrier. The bubble trap concept exploits the tendency of

bubbles to rise in a liquid if given the opportunity. This can be accomplished

by reducing the velocity of the incoming blood so that the natural buoyancy of

the bubbles becomes the dominant force. If an escape path is provided these

bubbles can be eliminated. This technique can remove bubbles of 300 µm or

more in diameter. Gas separation based on the surface tension phenomena

at a wetted screen is employed for the removal of bubbles less than 300 µm.

The mechanism takes advantage of the surface tension of the liquid. In simple

terms the pressure applied across a pore of the filter screen, must be

sufficient to disrupt the surface tension and only then air can be driven

through the pore (Figure 4).

The critical pressure or bubble point pressure, below which no air can pass

the pore, is calculated by the equation:

DΘ

=cos4γP

where P is bubble point pressure [mmHg], γ is the surface tension [dynes/cm],

D is the diameter of the pore [cm], Θ is the wetting angle.

For most filters, Θ approaches 0 and thus cos Θ = 1.

Chapter 3

29

Figure 4 Equilibrium position

P 1

P 2

Pore size [D]

Hydrophylic material offilter screen

Direction of fluid flow

γ surfacetension

circumferenceof pore π D

γ cos Θsurface tensionacts at contact withpore)

Θ

surface ofgas bubble

Θ

wetting angle

For a typical system γ = 50 dynes/cm and D = 40 µm, resulting in a bubble

point pressure of 37 mmHg. The pressure drop over a clean 40 µm screen is

about 3 mmHg at a blood flow of 5 L/min, the wetted screen can act as a

barrier to gas micro-emboli until the bubble point is reached. Any increase in

pressure drop above the bubble point pressure will result in passage of the

bubble, any decrease in pressure drop over the filter screen will the bubble

retract from the pore.

Unfortunately in paediatric cardiopulmonary bypass the gas escape path of

the arterial filter, the vent at the top of the filter, cannot be opened

continuously since this will create an important arterio-venous shunt. As a

consequence the arterial filter in combination with its bypass line will enlarge

the circuit volume and thus the circulating blood volume of the child with

Chapter 3

30

approximately 50 mL. This volume increase represents approximately 25% of

the total circuit volume.

However, the microporous fibres of the membrane can actively remove

gaseous microemboli. When blood enters the oxygenator its velocity will be

reduced, in the same manner as in an arterial filter, due to the larger open

area for blood flow. When gas comes into contact with the microporous fibres

it will be transported through the micropores due to the pressure difference

between the blood and gas side. This process is in function of pressure drop,

contact area and the availability of gas exchange fibres at the entrance of the

oxygenator.

3.4. Conclusions

The use of hydrodynamic formulas for the calculation of tubing length and

diameter allows the surgical team to define the best possible solution for a

given clinical situation based on desired pressure drop and flow pattern.

The use of an arterial line filter is debatable since it is a passive device that

cannot operate with open vent line during paediatric cardiopulmonary bypass.

The exclusion of the arterial filter in combination with an adequate choice of

tubing will result in an important reduction of dead volume and less

haemodilution, leading to a reduced use of homologous blood products.

Chapter 3

31

References

1. JE Brodie, RB Johnson. In The manual of clinical perfusion. Augusta,

Glendale Medical Corporation, 1994, 9-14.

2. P Dierickx, D De Wachter, P Verdonck. Fluid mechanical approach of

extracorporeal circulation. Course notes Institute Biomedical Technology,

Hydraulics laboratory Ghent University, 1998.

3. Elliot M. Minimizing the bypass circuit: a rational step in the development

of pediatric perfusion. Perfusion 1993; 8: 81-86

4. Tyndal M, Berryessa RG, Campbell DN, Clarke DR. Micro-Prime Circuit

Facilitating Minimal Blood use during Infant Perfusion. J. Extra-Corpor.

Technol. 1987, 19: 352-357

Chapter 3

32

Chapter 4

33

Chapter 4 Oxygenation by artificial lung systems

The artificial lung or oxygenator is the most technical part of the

cardiopulmonary bypass circuit. The design objectives of the “ideal”

oxygenator are still the same as in 1962 when Galletti and Brecher [1]

described, the “ideal” oxygenator as one that provided: oxygenation of venous

blood, carbon dioxide elimination, minimum blood trauma, small priming

volume and safety.

Today almost 100% of the oxygenators used are membrane oxygenators [2].

Meaning that a membrane separates the gas and the blood phase. The

majority of devices use a microporous hydrophobic membrane. Beside the

function of gas exchanger most devices incorporate a heat exchanger and a

reservoir. Thus the oxygenator performs all major functions of the natural

lungs except for their endocrine function, which can be suspended for a short

time without major ill effects.

4.1. The venous reservoir

There are two basic types of venous reservoirs: closed and open. The closed

system consists of a PVC bag with an in and outlet and one or more venting

ports for the evacuation of air. Advantages of the closed system are almost no

blood-air interface, small foreign surface area; collapse of the outlet when the

reservoir is suddenly emptied; quick indication of fluid changes and the ability

of volume controlled weaning from cardiopulmonary bypass. Disadvantages

are a more difficult air removal, when air accidentally enters the system, and

the need for an additional cardiotomy reservoir.

Chapter 4

34

The open system is in essence a reservoir open to the atmosphere with

incorporated cardiotomy reservoir. This system is somewhat easier to set-up

than a closed system and avoids the use of an additional cardiotomy

reservoir. When accidentally large amounts of air enter the reservoir, this can

be faster removed than in a closed system. The major disadvantages are the

large foreign surface area, the hold-up of volume in filter and defoamer and

the risk of inadvertently pumping air.

4.2. The heat-exchanger

The working principle of a heat exchanger is based on the principles of

conduction and forced convection. Water is used to control the temperature of

the blood. A common misconception is that the blood side is the determining

factor for performance. The water side is as important because it is desirable

to have high flows and turbulent flow to promote conductance. On the blood

side it is important to maintain laminar flow to minimise blood component

damage, but also to keep the total cross sectional area for blood flow as small

as possible to increase conductance [3].

The material used for the separation between the blood and water flows

should be as thin as possible for the highest conductance, with a very high

thermal conductivity, yet still have the integrity to withstand the expected

water and blood side pressures without failure.

Unfortunately, the most haemocompatible materials used in extracorporeal

blood handling devices have very poor thermal conductivities (k)(Table 1).

There is a trend to use more polymeric heat exchangers since these can be

Chapter 4

35

more easily coated or surface modified compared to metal heat exchangers

[3], what makes them more haemocompatible.

Table 1: Thermal conductivity of different materials

Material K

(W/m.K)

Stainless steels 15.1

Aluminium 237

Polycarbonate 0.2

Silicone 0.2

Epoxy 0.2

Polyurethane 5

4.3. The gas exchanger

The intrinsic physico-chemical and transport characteristics of the membrane,

the fluid dynamics and the haemocompatibility of the membrane module will

all determine its final mass transfer. As soon as blood gets into contact with

the hydrophobic polymeric surface, the material will adsorb proteins. The

amount of the protein layer and the nature of the proteins that are adsorbed

will depend on the physico-chemical characteristics of the membrane and on

the fluid dynamics in the membrane module. Poor fluid dynamics in the blood

flow channel of the membrane module will affect dramatically its performance

[4] because of:

Chapter 4

36

1. High blood boundary layer resistance to mass transport. This remains

extremely important since the resistance to mass transfer in a microporous

membrane oxygenator is fluid bound.

2. Poor haemocompatibility. High shear rates, eddy formation and stagnation

will favour the occurrence of clotting [5]

3. Large membrane surface. This will is needed for obtaining enough mass

transfer but will on the other hand cause activation of the complement

system [6,7]

In order to obtain the best possible fluid dynamics, most manufacturers use

today extra luminal flow (ELF) designs. In this design blood is flowing outside

regularly spaced hollow fibres. The hollow fibres are delivered knitted together

in a double layer mat. The membrane module is manufactured by wrapping a

double layer hollow fibre mat around a solid core, which is then inserted into a

cylindrical shell. In these modules blood flows through the membrane mesh

while gas flow is fed counter-currently into the hollow fibres. Since flow

through the membrane mesh will be forced to flow partly along and partly

around each hollow fibre secondary flows will be generated. This particular

membrane arrangement induces mixing in every section of the membrane

module to an extent that will depend on the membrane angle with respect to

the main direction of blood flow [4]. The efficient destruction of boundary

layers by this “static mixer” configuration leads to reduced resistance to mass

transfer [8] and yields high transfer rates across the membrane. Aside of the

better mass transfer this design has also lower pressure drops at the blood

side and no sharp edges in the blood flow path resulting in a better

haemocompatibility.

Chapter 4

37

The introduction of this ELF design in paediatric oxygenators has resulted in

an important reduction of the total priming volume of the paediatric

cardiopulmonary bypass (Figure 1).

Figure 1. Evolution of priming volume in the University Hospital in Gent

Nevertheless the priming volume and the total amount of foreign material

remains large in even the smallest circuits (Figure 2). Each millilitre of blood in

such a small paediatric cardiopulmonary bypass is still exposed to more than

three times the amount of foreign surface compared to an adult circuit. This

will have a major impact on the inflammatory response [7].

Figure 2. Relationship between blood volume and foreign surface

0

5000

10000

15000

20000

25000

Adult (70 Kg) Baby (5 Kg)

Blo

od v

olum

e (m

L) -

Sur

face

are

a (c

m²)

0

24

68

10

1214

16

cm²/

mL

bloo

d

Blood volume Surface Ratio

050

100150200250300350400450500

Pri

min

g vo

lum

e [m

L]

1990 1991 1993 1994 1995 2002

Chapter 4

38

4.4. Fluid dynamics and shear stress

As pointed out when describing the gas exchanger module, fluid dynamics is

an important item for obtaining optimal mass transfer. However an oxygenator

does consist of different components which must be connected. At the same

time, blood has to be evenly distributed over the heat exchanger and through

the membrane mesh by manifolds. As a result blood velocity will change when

blood passes through the oxygenator and this may result in zones of stasis,

eddy formation and or high shear. The average shear stress at the wall in a

membrane oxygenator can be calculated by starting with the general

macroscopic force balance for flow in a tube [9]. However, tube flow does not

accurately represent the complex flow in an ELF oxygenator. Flow through an

oxygenator can be considered as flow through a porous medium. According to

Bird [10] the shear stress in each oxygenator was calculated by considering

the flow equivalent to the flow in a packed column governed by:

LPRh∆

=τ

where: ∆P = pressure drop [N/m²], L= blood path length [cm], Rh = hydraulic

radius [cm]

)()(

)6/25(AeP

LQRhε

µ∆

=

where: ε = porosity of membrane area that fills that cross section

Q = volumetric pump flow [L/min]

µ = dynamic fluid viscosity [N/m².s]

Ae = cross sectional area for flow [m²]

25/6 = experimental derived factor.

Chapter 4

39

Mockros proposed a different formula for calculating shear in an oxygenator

[11,12].

21

∆

=VPQµ

τ

where: V = volume oxygenator [L]

The average shear for two different neonatal oxygenators calculated by both

formulas are given in table 2.

Table 2: Characteristics of two neonatal oxygenators

Parameter DidecoD901

PolystanSafe Micro

Membrane Surface area, m² 0.34 0.33Heat Exchanger Area, m² 0.02 0.05Void volume 0.58 0.48Total priming volume, cm³ 60 52Blood Pressure Drop @ 0.8 lpm (mmHg) 95 51Blood Pressure Drop @ 0.6 lpm (mmHg) 65 35Blood Pressure Drop @ 0.4 lpm (mmHg) 40 21Blood Pressure Drop @ 0.2 lpm (mmHg) 20 9Blood Path Length oxygenator, cm 30 15.3Average Cross sectional area for flow, cm² 12 7.62τ oxygenator (Ben Brian) [dynes/cm²]1 18 17τ oxygenator (Mockros) [dynes/cm²]1 25 20τ membrane compartment (Ben Brian)[dynes/cm²]1

31 19

τ membrane compartment (Mockros)[dynes/cm²]1

7 5

Although the calculated values are comparable with those in blood vessels

(see chapter 2), these values are average values and do not exclude that at

certain points in the design shear stress is above the critical level of 75 – 100

dynes/cm² needed to activate white blood cells and platelets.

1 ²

1²

10mN

cmdyne

=

Chapter 4

40

Every extracorporeal device will have a flow window with “ideal shear”. If

shear is to high platelets and blood elements will be damaged but when shear

is too low platelets will be more easily adsorbed by the material. As explained

earlier not only the magnitude of shear stress is important but also the

exposure time to this absolute value. It is well known that high shear for a

short time period is better tolerated than average shear during a long

exposure time [13]. In order to define spots with high or very low shear stress

in a design computational fluid dynamics are used [14-16].

4.5. Conclusions

Major improvements in oxygenator design has led to a large reduction in

foreign surface area, better haemocompatibility and enhanced mass transfer.

Although fluid dynamics have improved more work should be done to locate

risk zones at micro level. Computational fluid dynamics might offer the tool for

obtaining this goal. Finally this may lead to the ideal paediatric oxygenator

that will combine optimal fluid dynamics and thus mass transfer with a small

priming volume and foreign surface area.

References

1. PM Galletti, GA Brecher. Bubble oxygenation and membrane oxygenation.

In: Heart lung bypass; principles and techniques of extracorporeal

circulation. New York: Grune and Stratton; 1962: 108-120.

Chapter 4

41

2. Giovanni Cecere, Robert Groom, Richard Forest, Reed Quinn, Jeremy

Morton. A 10-year review of pediatric perfusion practice in North America.

Perfusion 2002; 17: 83-89.

3. RL Rigatti, R Stewart. Heat exchange in extracorporeal systems. In:

Cardiopulmonary bypass Principles and techniques of extracorporeal

circulation. Ed. CT Mora. New York: Springer Verlag; 1995: 247-256.

4. G Catapano, A Wodetzki, U Baurmeister. Blood flow outside regularly

spaced hollow fibers: the future concept of membrane devices. The Int J

Artif Organs 1992; 15: 327-330.

5. HL Goldsmith. The effects of flow and fluid mechanical stress on red cells

and platelets. Trans ASAIO 1974; 20: 21-26.

6. A Mahiout, H Meinhold, M Kessel, H Schulze, U Baurmeister. Dialyzer

membranes: effects of surface area and chemical modification of cellulose

on complement and platelet activation. Artif Organs 1987: 11: 149-154.

7. J Sonntag, I Dähnert, B Stiller, R Hetzer, PE Lange. Complement and

contact activation during cardiovascular operations in infants. Ann Thorac

Surg 1998; 65: 525-531.

8. WJ Dorson, KG Larsen. Secondary flows in membrane oxygenators. In

Mechanical devices for cardiopulmonary assistance. Eds. RH Bartlett, PA

Drinker, PM Galletti Adv. Cardiol., vol 6, pp 17-39 Karger, Basel 1971.

9. BF Brian. Comparative analysis of shear stress and pressure drop in

membrane oxygenators. White paper. Cobe Laboratories, Inc. 1995.

10. RB Bird, WE Stewart, EN Lightfoot. In: Transport phenomena. John Wiley

& Sons, NY, 1960.

Chapter 4

42

11. JM Ramstack, L Zuckerman, LF Mockros. Shear induced activation of

platelets. J Biomech 1979; 12: 113-125.

12. M Bluestein, LF Mockros. Hemolytic effects of energy dissipation in flowing

blood. Med Biol Eng 1969; 7: 1-6.

13. VT Turitto, CL Hall. Mechanical factors affecting hemostasis and

thrombosis. Thromb Res 1998; 15: S25-31.

14. MS Goodin, EJ Thor, WS Haworth. Use of computational fluid dynamics in

the design of the Avecor Affinity oxygenator. Perfusion 1994; 9: 217-222.

15. PW Dierickx, F De Somer, DS De Wachter, G Van Nooten, PR Verdonck.

Hydrodynamic characteristics of artificial lungs. ASAIO Journal, 2000;

46(5): 532-535.

16. Peter W Dierickx, Dirk S De Wachter, Filip De Somer, Guido Van Nooten,

Pascal R Verdonck. Mass Transfer Characteristics of Artificial Lungs.

ASAIO Journal, 2001; 47(6): 628-633.

Chapter 5

43

Chapter 5 Systemic inflammatory response

At the moment cardiac surgery starts; the baby is aggressed by many factors.

This agression by both surgery and cardiopulmonary bypass results in an

inflammatory response. There is little doubt that this inflammatory response is

responsible for a proportion of the mortality and morbidity associated with

cardiac surgery. Certain organs and tissues are at higher risk of developing

deranged function after the perfusion and in the postoperative period. At the

greatest risk are the formed elements in the blood, the platelet and white cell,

resulting in clotting problems and abnormal organ and tissue functions. In

particular the pulmonary system, heart and myocardium, kidney and

splanchnic bed, and the brain and cerebral circulation are specifically affected

and thus contribute to early postoperative morbidity and mortality [1]. Small

babies are even more at risk due to the larger volume and foreign surface

area of the extracorporeal circuit in combination with the immaturity of many

organs systems and the large amount of blood that after contact with tissue is

returned into the systemic circulation.

The bio-incompatibility of cardiopulmonary bypass is multifactorial (Figure 1)

and can be divided in two major groups: material independent and material

dependent [2].

Chapter 5

44

Figure 1.Bioincompatibility of paediatric cardiopulmonary bypass is

multifactorial

MaterialsMaterialsrelatedrelated

CircuitCircuitrelatedrelatedSurgerySurgery

relatedrelated

PatientPatientGeneticGeneticrelatedrelated

TempTemp..

CardioCardio--plegiaplegia

TissueTissueFactorFactor

DrugsDrugs

ShedShedandand/or/or

SuctionedSuctionedBloodBlood. Air. Air

SterilitySterility

OpenOpenvsvs

ClosedClosed

RollerRollervsvs

CentrifCentrif..

PulsePulsevsvs

NonNon

StasisStasisPointsPoints

EmboliEmboliShearShearStressStress

DebrisDebrisSurfaceSurfaceAreaArea(s)(s)

CanuCanu

--lationlation

CoagCoag

Pathophysiology and bioincompatibility of CPB

5.1. Material dependent

Under normal conditions, when blood is in a blood vessel with intact

endothelium, no activation of blood proteins or elements will occur. However,

the moment blood leaves this protected environment and comes into contact

with damaged endothelium, other tissue or artificial surfaces several cascades

of reactions will start. At the same time the shear stresses that work beneficial

when applied on endothelium by releasing mediators such as nitric oxide, will

now in absence of the endothelium activate blood elements. Aspects that

contribute to this activation cascade are the surface characteristics [3], the

sterilisation method and the chemical composition of the surface of the

polymer. It is important to notice that there can be major chemical differences

between the bulk material and the surface.

Chapter 5

45

5.2. Blood interactions with polymers

5.2.1. Protein adsorption and complement activation

As soon as blood comes in contact with the hydrophobic polymer surfaces of

the cardiopulmonary bypass the latter will be almost immediately covered with

proteins. The formation of this protein layer is followed by the adherence of

platelets. In addition to fibrinogen, γ-globulin preadsorbed to artificial surfaces

enhances the platelet release reaction in vitro. In contrast, serum albumin

passivates the surface towards platelet adhesion [4]. Glycosyl transferase

reactions involving incomplete terminal oligosaccharide units were postulated

as mediators for these platelet-protein interactions. These groups are present

in fibrinogen, γ-globulin and many other glycoproteins in plasma, but are

absent in albumin [5]. The highest concentration of fibrinogen on the material

is realised after 15 minutes [5]. Fibrinogen adsorption has been used as a

measure of thrombogenicity of materials. Aside from its role in the fibrin

formation it will bind blood platelets via their surface glycoproteins IIb/IIIa and

Gib [6]. However the platelets do not seem to interact with the material directly

but through the adsorbed protein layer.

In high flow rate conditions it seems that the platelet response is a major

determinant of blood incompatibility with artificial surfaces [2]. At this point it is

important to put in perspective the effects of shear stresses near the wall of

the hydrophobic polymers since this will contribute to leukocyte and platelet

activation and in exceptional situations red blood cell lysis.

Although complement is activated to a large degree via the alternative

pathway, it is only to a minimal extent, in adult surgery, linked to the foreign

Chapter 5

46

materials. Other pathways must be playing a role in the complement

activation such as factor XIIa, kallikrein and tissue factor [7]. Likewise, C3a

and C5a anaphylatoxins may appear to be reduced in plasma where in reality

they are adsorbed by the protein layers and thus are measured in lower

amounts [8].

5.2.3. Contact activation

The intrinsic coagulation cascade as well as the fibrinolysis system are both

initiated by the contact activation phase. Four proteins are activated during

the contact phase: factor XII, high molecular weight kininogen (HMWK),

prekallikrein and factor XI [9]. Adsorption of factor XII in presence of

prekallikrein and HMWK produces active proteases, factors XIIa and XIIf [10].

In a feedback loop, factor XIIa cleaves prekallikrein to produce kallikrein and

HMWK to produce bradykinin, a short acting vasodilator. Factor XIIa in the

presence of kallikrein and HMWK also activates factor XI to factor XIa

activates the intrinsic coagulation cascade, which proceeds through factor IX

to activate factor X and form thrombin [10]. Electrical charge (cationic or

negatively charged surface) and the hydrophobicity of the artificial surface can

also promote this initial contact activation with foreign material. The contact

activation phase, as seen previously by factor XII and kallikrein, will also

directly activate the complement system and initiate the plasminogen/plasmin

formation. Contact activation may be more prominent at low flow than high

flow conditions.

Interestingly, recent research [11-12] shows a much lower activation of the

intrinsic pathway but on the other hand the activation pathway with KK and

FXII on leukocytes may be more that what has been shown so, far.

Chapter 5

47

5.3. Material independent

Other factors that influence the degree of inflammatory response do not

depend on the material but are equally or more important for the initiation of

an inflammatory response. A very aggressive activator is the cardiotomy

suction. Especially in paediatric surgery the amount of blood recuperated by

the cardiotomy reservoir can be quite large due to additional blood vessels

(e.g. left vena cava superior), flow through collateral vessels etc. This

aspirated blood is contaminated with tissue factor, tissue and fat fragments,

free plasma haemoglobin, thrombin, tissue plasminogen activator and fibrin

degradation products. All these elements in combination with the turbulent

flow and the blood-air mixing in the aspiration lines will activate, through blood

platelets and leukocytes, both coagulation and complement cascades. At the

same time the aspirated fat emboli are an important source of cerebral

embolisation [13] which, unfortunately cannot be prevented by the use of

venous or arterial filters [14-15]. Important is also the presence of high

amounts of S100BB in aspirated blood originating from fat, muscle and

marrow in the mediastinal blood [16]. Since it had always been postulated that

S100BB was a specific marker for brain damage and that the elevated plasma

levels found after cardiopulmonary bypass were caused by damage of the

brain.

A second factor is flow dynamics and fluid mechanical stresses (See also

chapters 2 & 4). Especially stasis and eddy formation has an important impact

on protein adsorption and thus on the formation of thrombi. Also shear stress

is an important activator of primarily platelets and leukocytes. The magnitude

and duration of shear stress will dependent from component to component

Chapter 5

48

and the blood flow characteristics in a given cardiopulmonary circuit, but will

always be present to some extent. A high value with a short duration will be

found in arterial cannulas while different magnitudes of shear but with longer

duration are found in oxygenators and reservoirs [17,18]. Shear stress

induced platelet activation is mediated by von Willebrand factor binding to

platelet membrane receptors GPIb and GPIIb/IIIa [2]. Shear stress as small as

100 dynes/cm² will induce platelet and leukocyte activation [19].

A third factor is related to the use of homologous blood products and

haemodilution. The risks of homologous blood transfusion such as

immunobiological disorders [20] and transmission of infections are well

documented [21]. Because of their young age infections caused to the use of

homologous blood products should be avoided in every extent. Open-heart

surgery without the use of homologous blood products is commonly

performed in adults, but still difficult in small children because priming volume

of the cardiopulmonary bypass circuit results in extreme haemodilution [22].

A fourth factor is related to the use of drugs. Best documented is the

activation of the classical pathway of the complement system by the heparin-

protamine complex. This will lead to monocyte and neutrophil activation [2,

23-24].

Finally also conduct of cardiopulmonary bypass as well as the genetic

footprint of the child will play a role. The use of open or closed system [25],

the oxygen tension used during cardiopulmonary bypass [2, 26-27], the

cooling protocol [27, 28] and haemoglobin content when using deep

hypothermic circulatory arrest [29] have all been put forward as variables that

can influence inflammatory response. Of course every child is unique in his

Chapter 5

49

genetic footprint and this can interact in the way they biologically will react on

the damage caused by the surgery and cardiopulmonary bypass. The

haptoglobin phenotype for example will determine the capacity for binding free

plasma haemoglobin [30] and might have an impact on the immune response

[31]. While the different platelet PLA allelic frequency have been associated

with a predisposition for increased thrombogenicity [32], increased release of

IL8 and TNF after cardiopulmonary bypass [33] and more pronounced

neurocognitive decline after cardiopulmonary bypass [34]. Beside these

genetic factors also the pathology might influence the activation of the

different cascade. The higher incidence of fibrinolysis in cyanotic children is a

perfect example of the latter.

Inflammatory response to cardiopulmonary bypass is considerably more

complex than it seemed a decade ago. In children the analysis of

inflammatory response is even more complex due to different response of

neonates and children to cardiopulmonary bypass [35]. Nevertheless, it is

possible, based on our present knowledge to attenuate inflammatory

response. The large foreign surface area of the paediatric cardiopulmonary

bypass circuit, almost 4 times more than an adult circuit, remains an important

issue [5]. Changing this surface into a more blood compatible surface looks

promising. The aims of such a re-engineering should be elimination or

reduction of [2]:

1. Plasma protein adsorption in order to reduce cellular activation

2. Coagulation activation

3. Complement activation

4. Leukocyte activation

Chapter 5

50

While at the same time the physical properties of the various bulk polymers

are preserved.

Different approaches have been published in order to achieve these goals.

Best known is heparin coating of polymers. In adults non-uniform results have

been published over the years [36]. This might be related to the fact that in

most clinical studies aspirated blood, recognised as one of the most injurious

components [37], is still re-used. In paediatric open heart surgery this aspect

will even gain in importance due to the larger amounts of aspirated blood.

Nevertheless, lower inflammatory response is reported with heparin coated

paediatric cardiopulmonary bypass [38-41], although not for all markers [42].

Also the use of phosphorylcholine coating was reported to be beneficial [43].

The attractive idea of combining surface amelioration with separation of

aspirated blood for further reduction of the inflammatory cascade has not

been realised yet due to technical limitations.

More controversial is the use of ultrafiltration for removal of inflammatory

mediators [44-45] especially when compared to cardiopulmonary bypass

circuits with a low priming volume and reduced foreign surface area.

A last method to control inflammatory response is by pharmacological

interaction. Aprotinin has been reported to attenuate cellular and humoral

response to cardiopulmonary bypass both in adult [46] and paediatric [47-49]

populations. Also the use of some inhibitors [50] looks promising, but larger

study cohorts are necessary to confirm these data.

Chapter 5

51

5.4. Conclusion

The inflammatory response to cardiopulmonary bypass is considerably more

complex than it seemed a decade ago. The acute phase response to trauma

may be an integral part of this process. Our expanding knowledge of

inflammatory mediators will allow a better understanding of cardiopulmonary

related morbidity and may hopefully lead to improvement of biocompatibility of

cardiopulmonary bypass resulting in less injurious systemic responses and

diminished organ and tissue damage.

References

1. D Royston. Systemic inflammatory responses to surgery with

cardiopulmonary bypass. Perfusion, 1996; 11: 177-189.

2. S Daniel. Review of the multifactorial aspects of BioInCompatibility in CPB.

Perfusion, 1996; 11: 246-255.

3. NP Ziats, DA Pankowsky, BP Tierney et al. Absorption of Hagemann

factor and other human plasma proteins to biomedical polymers. J Lab

Clin Med 1990; 116: 687-696.

4. V Videm, E Fosse, JL Svennig. Platelet preservation during coronary

bypass surgery with bubble and membrane oxygenators: effect of albumin

priming. Perfusion 1993;8: 409-415.

5. BR Young, LK Lambrecht, SL Cooper. Plasma Proteins: Their Role in

Initiating Platelet and Fibrin Deposition on Biomaterials. In Cooper SL,

Peppas NA, eds. Biomaterials: Interfacial Phenomena and Applications..

Washington DC: Advances in Chemistry Series 199, American Chemical

Society, 1982; 317-350

Chapter 5

52

6. BR Young, LK Lambrecht, RM Albrecht et al. Platelet protein interactions

at blood-polymer interfaces in the canine test model. Trans ASAIO, 1983;

29: 442-446.

7. YJ Gu, MA Mariani, PW Boonstra, JG Grandjean, W van Oeveren.

Complement activation in coronary bypass grafting patients without

cardiopulmonary bypass. The role of tissue injury by surgical incision.

Chest 1999; 116: 892-898.

8. H Nishida, S Aomi, Y Tomizawa et al. Comparative study of

biocompatibility between the open and closed circuit in cardiopulmonary

bypass. Artificial Organs 1999; 23: 547-551.

9. RW Colman. Surface mediated defense reactions. The plasma contact

activation system. J Clin Invest 1984; 73: 1249.

10. LH Edmunds, N Stenach. Blood-surface interface. In: GP Gravlee, RF

Davis, M Kurusz, JR Utley eds. Cardiopulmonary bypass. Principles and

practice Philadelphia: Lippincott Williams & Wilkins, 2000: 150-166.

11. W Gil. Inflammo-coagulatory response, extrinsic pathway thrombin

generation and a new theory of activated clotting time interpretation.

Perfusion 2001;16(1):27-35.

12. C Baufreton, JL de Brux. Les traitements de surface en circulation

extracorporelle. RBM 1999;21 suppl 1: 20-25.

13. RF Brooker, WR Brown, DM Moody, et al. Cardiotomy Suction: A Major

Source of Brain Lipid Emboli During Cardiopulmonary Bypass. Ann Thorac

Surg 1998;65:1651-55.

14. DA Stump. Emboli: Their source and significance in neurological outcome.

25th Anniversary NeSECC Journal 2001; 26: 15-19

Chapter 5

53

15. EH Kincaid, TJ Jones, DA Stump et al. Processing scavenged blood with a

cell saver reduces cerebral lipid microembolization. In: Annual meeting of

the Southern Thoracic Association; Puerto Rico, 1999.

16. RE Anderson, LO Hansson, O Nilsson, J Liska, G Settergren, J Vaage.

Increase in serum S100A-1B and S100BB during cardiac surgery arises

from extracerebral sources. Ann Thorac Surg 2001; 71: 1512-1517.

17. F De Somer, L Foubert, M Vanackere, D Dujardin, J Delanghe, G Van

Nooten. Impact of oxygenator design on hemolysis, shear stress, white

blood cell and platelet count. J. Cardiothor.Vasc. Anesth. 1996; 10: 884-

889

18. YJ Gu, PW Boonstra, R Graaff, AA Rijnsburger, H Mungroop, W van

Oeveren. Pressure drop, shear stress, and activation of leucocytes during

cardiopulmonary bypass: A comparison between hollow fiber and flat

sheet membrane oxygenators. Artificial Organs 2000; 24: 43-48.

19. JD Hellums, RA Hardwick. Response of Platelets to Shear Stress - a

Review. In The Rheology of Blood Vessels and Associated Tissues Eds

Gross DR, Hwang NHC.. Alphen aan den Rijn: NATO Advanced Study

Institute Series - E, No 41, Sijthoff & Noordhoff, 1981

20. A Salama, EC Mueller. Delayed hemolytic transfusion reactions. Evidence

for complement activation involving allogeneic and autologous red cells.

Transfusion 1984; 24: 188-193.

21. JW Rasenack, HJ Schlayer, F Hettler, T Peters, AS Preisler, W Gerok.

Hepatitis B virus infection without immunological markers after open-heart

surgery. Lancet 1995; 345: 355-357.

Chapter 5

54

22. LA Chambers, DM Cohen, JT Davis. Transfusion patterns in pediatric

open heart surgery. Transfusion 1996; 36: 150-154.

23. NC Cavarocchi, HV Schaff, TA Orszulak, HA Homburger, WA Schnell, JR

Pluth. Evidence for complement activation by protamine-heparin

interaction after cardiopulmonary bypass. Surgery 1985; 98(3): 525-531

24. S Ashraf, Y Tian, D Cowan et al. “Low-dose” aprotinin modifies

hemostasis but not pro-inflammatory cytokine release. Ann Thorac Surg

1997; 63: 68-73.

25. H Nishida, S Aomi, Y Tomizawa et al. Comparative study of

biocompatibility between the open and closed circuit in cardiopulmonary

bypass. Artificial Organs 1999; 23: 547-551.

26. DT Pearson, RF Carter, MB Hammo, PS Waterhouse. Gaseous micro-

emboli during open heart surgery. In: Towards safer cardiac surgery. Ed.

DB Longmore. Lancaster, MTP Press, 1981: 325-354.

27. JM Pearl, DW Thomas, G Grist, JY Duffy, PB Manning. Hyperoxia for

management of acid-base status during deep hypothermia with circulatory

arrest. Ann Thorac Surg 2000; 70: 751-755.

28. RA Jonas, DC Bellinger, LA Rappaport et al. Relation of pH strategy and

development outcome after hypothermic circulatory arrest. J Thorac

Cardiovasc Surg. 1993; 106: 362-368.

29. T Shin’oka, D Shum-Tim, PC Laussen et al. Effects of oncotic pressure

and haematocrit on outcome after hypothermic circulatory arrest. Ann

Thorac Surg 1998; 65: 155-164.

30. J Delanghe, K Allcock, M Langlois, L Claeys, M De Buyzere. Fast

determination of haptoglobin phenotype and calculation of hemoglobin

Chapter 5

55

binding capacity using high pressure gel permeation chromatography. Clin

Chim Acta 2000; 291: 43-51.

31. M Langlois, JR Delanghe. Biological and clinical significance of

haptoglobin polymorphism in humans. Clin Chem 1996; 42: 1589-1600.

32. EJ Weiss, PF Bray, M Tayback et al. A polumorphism of a platelet

glycoprotein receptor as an inherited risk factor for coronary thrombosis. N

Eng J Med 1996; 334: 1090-1094.

33. N Drabe, G Zünd, J Grünenfelder et al. Genetic predisposition in patients

undergoing cardiopulmonary bypass surgery is associated with an

increase of inflammatory cytokines. Eur J Cardiothorac Surg 2001; 20:

609-613.

34. JP Mathew, CS Rinder, JG Howe et al. Platelet PlA2 polymorphism

enhances risk of neurocognitive decline after cardiopulmonary bypass.

Ann Thorac Surg 2001; 71: 663-666

35. SS Ashraf, Y Tian, S Zacharrias, D Cowan, P Martin, K Watterson. Effects

of cardiopulmonary bypass on neonatal and paediatric inflammatory

profiles. Eur J Cardiothorac Surg 1997; 12: 862-868.

36. HP Wendel, G Ziemer. Coating-techniques to improve the

hemocompatibility of artificial devices used for extracorporeal circulation.

Eur J of Cardiothoracic Surg 1999;16:342-50.

37. de Haan J, Boonstra PW, Monnink SHJ, Ebels T, van Oeveren W.

Retransfusion of Suctioned Blood During Cardiopulmonary Bypass Impairs

Hemostasis. Ann Thorac Surg 1995;59: 901-7.

Chapter 5

56

38. HH Schreurs, MJ Wijers, J Gu et al. Heparin-coated bypass circuits:

effects on inflammatory response in paediatric cardiac operations. Ann

Thorac Surg 1998; 66: 166-171.

39. EA Grossi, K Kallenbach, S Chau et al. Impact of heparin bonding on

pediatric cardiopulmonary bypass: a prospective randomized study. Ann

Thorac Surg 2000; 70: 191-196.

40. T Ozawa, K Yoshihara, N Koyama, Y Watanabe, N Shiono, Y Takanashi.

Clinical efficacy of heparin-bonded bypass circuits related to cytokine

responses in children. Ann Thorac Surg 2000; 69: 584-590.

41. K Miyaji, RL Hannan, J Ojito, JP Jacobs, JA White, RP Burke. Heparin-

coated cardiopulmonary bypass circuit: clinical effects in pediatric cardiac

surgery. J Card Surg 2000; 15: 194-198.

42. SB Horton, WW Butt, RJ Mullaly et al. IL-6 and IL-8 levels after

cardiopulmonary bypass are not affected by surface coating. Ann Thorac

Surg 1999; 68: 1751-1755.

43. F. De Somer, K. François, W. van Oeveren et al. Phosphorylcholine

coating of extracorporeal circuits provides natural protection against blood

activation by the material surface. Eur J Cardiothorac Surg 2000; 18(5):

602-606.

44. RM Ungerleiter. Effects of cardiopulmonary bypass and use of modified

ultrafiltration. Ann Thorac Surg 1998; 65: S35-39.

45. MS Chew, I Brandslund, V Brix-Christensen et al. Tissue injury and the

inflammatory response to pediatric cardiac surgery with cardiopulmonary

bypass. Anesthesiology 2001; 94: 745-753.

Chapter 5

57

46. D Royston. Preventing the inflammatory response to open-heart surgery:

the role of aprotinin and other protease inhibitors. Int J Cardiol 1996; 53:

S11-S37.

47. H Mössinger, W Dietrich. Activation of hemostasis during cardiopulmonary

bypass and pediatric aprotinin dosage. Ann Thorac Surg 1998; 65: S45-

51.

48. J Boldt. Endothelial related coagulation in pediatric surgery. Ann Thorac

Surg 1998; 65: S56-59.

49. CF Wippermann, FX Schmid, B Eberle et al. Reduced inotropic support

after aprotinin therapy during pediatric cardiac operations. Ann Thorac

Surg 1999; 67: 173-176.

50. B Stiller, J Sonntag, I Dähnert et al. Capillary leak syndrome in children

who undergo cardiopulmonary bypass: clinical outcome in comparison

with complement activation and C1 inhibitor. Intensive Care Med 2001; 27:

193-200

Chapter 5

58

.

Chapter 6

59

Chapter 6 Summary and new prospectives

Since the first use of a heart-lung machine for total cardiopulmonary bypass

on April 5, 1951 major changes took place. While the first two patients did not

survive, today cardiopulmonary bypass related mortality is almost nihil. The

huge circuits with a bubble or film oxygenator requiring several litres of prime

have been replaced with small membrane oxygenators and circuits that only

require a few hundred millilitres of prime.

As more procedures were done the knowledge and long term follow-up of

patients with congenital heart disease increased. Based on these new

insights, more and more children are operated in the first days or weeks of

their life since this seems to have a major impact on long term survival.

However, this approach confronts the clinician with a lot of technical

limitations when he has to place a neonate of 2 Kg on cardiopulmonary

bypass. The often still immature organs demand for further research in order

to keep bypass related damage to an absolute minimum.

A first major problem is that of vascular access. The small blood vessels of

the child need to be cannulated without obstructing blood flow or damaging

the vessel wall. What is the best design for obtaining this goal? How can one

be sure that all organs are perfused, that the native heart will not be

challenged by an additional afterload and that total venous return is directed

towards the cardiopulmonary bypass? Appendix 1 focuses on the limitations

and advantages of vacuum assisted venous return (VAVD) in small babies.

VAVD makes it possible to enhanced venous return with approximately 10%

mainly due to a larger pressure difference. Additionally, VAVD also allows to

Chapter 6

60

use smaller cannulas resulting in less obstruction of the blood vessel and less

damage to the blood vessel wall. The combination of smaller cannulas with

VAVD might result in a larger operating field for the surgeon with less back

flow.

For arterial re-infusion the design of a cannula is of main importance.

Appendix 2 explains how arterial cannula design will affect jet formation while

Appendix 3 points to the limitations of existing paediatric arterial cannulas.

Large differences for their pressure flow characteristics were found based on

deviations in internal diameter and design.

In paediatric cardiopulmonary bypass the oxygenator remains a problem

because of his priming volume, large foreign surface area and not always

optimal fluid dynamics. These problems are partly due to the fact that most if

not all paediatric oxygenators are “downscaled” adult oxygenators and not

specifically adapted for neonatal procedures. Appendix 4 represents the

clinical benefits of an oxygenator specially designed for neonatal

cardiopulmonary bypass. The use of such a neonatal oxygenator makes it

possible to construct much smaller circuits resulting in less haemodilution.

Appendix 5 gives the clinical impact on blood products when using a neonatal

oxygenator in combination with a small circuit. The fluid dynamics in an

oxygenator are important for achieving optimal mass transfer and

haemocompatibility. Appendix 6 presents a new technique for the comparison

of the pressure flow relationship in oxygenators with a different design. This

approach makes it possible to make more objective decisions when

Chapter 6

61

comparing different products. The impact of the new ELF membrane

oxygenators on blood elements was studied in appendix 7.

One can question the use of an arterial filter in a paediatric circuit as it will

enlarge total priming volume without adding any additional safety. Appendix 8

suggests that the hollow fibre stack of the membrane compartment might be

an acceptable alternative since it will act as a depth filter and it is able to

remove gaseous emboli. This alternative will reduce priming volume without

jeopardising safety.

Control of the inflammatory response is a major goal for the paediatric team.

One approach coating all artificial surfaces with a coating that biomimicks the

outer layer of the cell membrane leads to a reduction in complement

activation and a better platelet preservation. This is reported in a dog model in

appendix 9 and confirmed in the clinical setting in appendix 10.

Unfortunately this coating does inhibit the inflammatory response completely

and this might be explained by the findings of appendix 11 that blood coming

from structures not covered with endothelium such as the pericardium and

pleural cavities does activate the coagulation system. By doing so it will also

activate the complement system and promote capillary leak.

Clinical implications and possible future directions

More and more new-borns with congenital heart disease are operated within

the first days or weeks of life. As a result body weight can be very low and the

anatomical structures will be small. Institution of cardiopulmonary bypass

Chapter 6

62

under such conditions asks for dedicated cannulas with minimal deviation of

the inner diameter. In order to achieve optimal venous drainage and arterial

re-infusion under all circumstances, more designs and diameters should be

developed. Pressure-flow diagrams based on viscous solutions such as

water-glycerine should accompany these new designs as well as existing

designs.

Vacuum assisted venous return in combination with dedicated venous

cannulas will further reduce the total priming volume of the cardiopulmonary

circuit and more importantly also reduce the “dead volume” in aspiration lines.

As a result blood will be exposed to a lower amount of foreign material and

less haemodilution of coagulation proteins and blood elements will occur. Due

to the lower haemodilution less homologous products are needed and

exposure to multiple blood donors can be avoided.

The treatment of all foreign material with a biocompatible coating will reduce

inflammatory response.

Future developments should focus on

1. Membrane technology: microporous versus diffusive

2. Surface treatment of all foreign surface

3. Integration of components and miniaturisation of the cardiopulmonary

bypass for further reduction of priming volume and foreign surface

4. Fluid mechanics of the complete cardiopulmonary bypass circuit combined

with extensive modelling of the fluid mechanics in each component

5. Cannulas in combination with the physical and biological aspects of

vascular access in general

6. Selective blood treatment for activated blood

Appendix 1

63

Evaluation of different paediatric venous cannulas usinggravity drainage and VAVD: an in vitro study

F. De Somer, D. De Wachter, PR Verdonck, G. Van Nooten, T. Ebels

Perfusion, 2002; 17(5): 321-326

Appendix 1

64

Abstract

Six different commercially available paediatric venous cannulas together with

a special constructed cannula were tested in vitro for their pressure-flow

relationship. With the cannulas placed in an open reservoir, flow increased

with larger diameters and higher pressures. At a pressure of 30 cm H2O flows

were 219 ± 20 mL/min, 285 ± 13 mL/min, 422 ± 11 mL/min, 728 ± 4 mL/min,

for the 12 Fr, 13.2, 14 Fr and 16 Fr, respectively. No differences were founds

between angled and straight cannulas.

When the cannulas were tested in a latex model simulating the right atrium

and venae cavae, the highest flow obtained by gravity was 164 mL/min using

an angled 14 Fr cannula. When vacuum was applied to augment venous

return a maximum flow of 179 mL/min was measured using an angled 14 Fr

cannula.

Collapse can occur when the pressure difference becomes too high in the test

system. This is important since most children are selectively cannulated in

both major veins. Monitoring of the intravascular pressure might help to

prevent collapse. A larger diameter venous cannula does not always produce

the highest flow when placed in a vein. This is most obvious when

augmenting venous return. The design of the cannula tip in combination with

VAVD can affect the venous return.

Appendix 1

65

Introduction

Vascular access remains an important aspect of cardiopulmonary bypass

(CPB) in paediatric cardiac surgery. Bi-caval cannulation with straight or

angled cannulas, using gravity siphon drainage, is most often used. In the last

decade major improvements have been made to decrease the extracorporeal

blood volume [1], as a result of which the volume of the tubing becomes more

important. Once the diameter required for a calculated flow be chosen, the

only way further to decrease this volume is to shorten the length of the lines.

This can be achieved by using active vacuum augmentation of the venous

return, which allows the user to place the oxygenator closer to the patient [2].

Vacuum assisted venous drainage gained renewed interest since the start of

minimally invasive cardiac procedures [3,4], and for reduction of priming

volume in paediatric circuits [5-8]. Veins are compliance vessels and will

collapse at negative pressures between minus 5 – 10 mmHg [2]. When using

gravity drainage, the pressure in the vein(s) will be more or less constant

during the procedure. However, when vacuum is applied as driving pressure

much lower pressures can be achieved. As a result the veins can collapse

and instead of an increase a reduction in flow, due to partial obstruction of the

open area for flow by the vein, will result.

In this study we investigate the influence of vein collapse, cannula diameter

and exerted negative pressure on the venous return in vitro.

Appendix 1

66

Methods

Single stage paediatric venous cannulas (Medtronic, Brussels, Belgium) in

three sizes (12, 14 and 16 French, wall thickness 0.025”) and two

configurations (straight (DLP 661xx) and right angled (DLP 675xx)) were

tested, together with a special constructed cannula. The latter consists of a

plastic helix with a diameter of 15 Fr mounted on 24 cm of 1/8 inch tubing.

This 1/8 inch tubing had an inner diameter of 9.6 Fr and an outer diameter of

13.2 Fr (Figure 1). Due to the use of PVC tubing as connection between the

tip and the venous line, the wall thickness of the specially constructed cannula

is much thicker than that of the commercially available cannulas. This cannula

was only available in straight configuration and was used to investigate the

potential benefit of a design less prone to obstruction in case of collapse of

the vein. In the text this cannula will be referred to as 13.2 Fr.

A first group of measurements validated all cannulas for their pressure flow

relationship. The test fluid was a 30% glycerine solution with a kinematic

viscosity of 2.5 mm2/s, which is similar to blood. The test cannula is placed

horizontally in a reservoir, while the level in the reservoir is kept constant by

means of an overflow. The flow rate through the cannula is regulated by the

height of the collecting chamber, which could be placed as low as 30 cm

below the cannula. Pressure is measured at the tip and the end of the cannula

by means of a differential pressure transmitter (Fuji Electric, Erlangen,

Germany). The flow rate is obtained gravimetrically by a timed fluid mass

collection (Figure 2A).

Appendix 1

67

For a second experiment a model of the right atrium including both caval

veins was constructed in latex. The dimensions of the model were based on

the echocardiographic measurements of the right atrium and caval veins in 10

babies. The average weight of the 10 children was 5.5 ± 0.7 kg. The average

diameter of the superior and inferior caval vein was 4.9 ± 0.4 mm and 5.3 ±

0.4 mm, respectively. For ease of construction both veins in the model had a

diameter of 5 mm. This results in a cannula-vein diameter ratio of 0.79 for the

12 Fr, 0.87 for the 13.2 Fr, 0.92 for the 14 Fr and 1.05 for the 16 Fr cannula.

For the measurements only the inferior caval vein was cannulated and a

purse string was used to prevent back flow into the right atrium. If both caval

veins had been cannulated, validation of the exact flow in each of the two

cannulae would have been difficult. The compliance of both vessels was

designed in such a way that they could collapse at a pressure of

approximately minus 10 - 15 mmHg. Pressures were recorded at the tip and

the end of the cannula. Flow was measured by a Transonic flow meter

(Transonic®, Ithaka,NY, USA). In a first approach gravity drainage was

applied with a height difference of 30 cm H2O. In a second approach a VAVD

controller (Polystan AS, Vaerlose, Denmark) was used to assist venous return

(Figure 2B), with the reservoir fluid level situated 18 cm below the model.

Appendix 1

68

Results (Table 1)

Pressure – flow relationship

The maximum flow obtained at 30 cm H2O with the straight cannulas was 219

± 20 mL/min, 285 ± 13 mL/min, 422 ± 11 mL/min, 728 ± 4 mL/min, for the 12

Fr, 13.2, 14 Fr and 16 Fr, respectively (Figure 3). For the angled canulae

flows of 216 ± 13 mL/min, 454 ± 7 mL/min and 727 ± 35 mL/min were

obtained at 30 cm H2O for the 12 Fr, 14 Fr and 16 Fr cannulas (Figure 3).

First experiment: gravity drainage

With the straight cannulas, the maximum flow before collapse occurred was

136 mL/min, 142 mL/min, 142 mL/min and 149 mL/min for the 12 Fr, 13.2 Fr,

14 Fr, and 16 Fr, respectively. With the angled cannulas flows of 131 mL/min,

164 mL/min and 151 mL/min were obtained with the 12 Fr, 14 Fr and 16 Fr

cannula (Figure 4). The mean pressure at which collapse of the vessel

occurred was 9.9 ± 1.2 mmHg.

Second experiment: VAVD

With the straight cannulas, the maximum flow before collapse occurred was

155 mL/min, 163 mL/min, 129 mL/min and 143 mL/min for the 12 Fr, 13.2 Fr,

14 Fr and 16 Fr, respectively. With the angled cannulas flows of 156 mL/min,

179 mL/min and 165 mL/min were obtained with the 12 Fr, 14 Fr and 16 Fr

cannula (Figure 5). The mean pressure at which collapse of the vessel

occurred was 12.4 ± 1.1 mmHg.

Appendix 1

69

Discussion

Very early in the development of cardiopulmonary bypass techniques for

obtaining maximal venous return were investigated [2]. Several variables have

been put forward as important, cannula/vein ratio, design of the cannula,

cannula position, characteristics of the connecting system and techniques for

augmenting venous return.

Today 18% of all paediatric cardiac surgery is performed in the first month of

life, while over 50 % is performed in children less than 1 year [3]. As a

consequence of the small vascular structures an optimal venous return is

mandatory for a bloodless surgical field. In general the resistance of cannulas

is low in babies because of the relatively large cannula/vein ratio compared to

adults. In spite of the surgical trend, it is difficult to find venous cannulas with

an external diameter smaller than 12 Fr.

In our series not surprisingly flow in a reservoir increased with larger

diameters and higher pressures. Our pressure-flow results support previous

findings showing no correlation between filling pressure and maximum flow

when the cannula is placed in an open reservoir [3]. However, with the

cannula placed in the model, filling pressure and cannula size does influence

flow rates.

When gravity drainage was applied only 5 % less flow was obtained with a

straight 12 Fr cannula compared to the larger cannulas. However, the

combination of an angled 12 Fr cannula with gravity drainage generated less

flow compared to the larger diameter cannulas. The difference between

straight and angled cannulas might be explained by the fact that a straight

Appendix 1

70

cannula tip will be easier pushed towards the wall of the vessel, especially

with the large cannula vein diameter ratios, than an angled cannula when

traction or manipulation is exerted. The small differences in flow found

between the three cannulas might be due to the fact that the cannula vein

diameter ratio was exceeding in every case 0.5. This has been shown to

compromise flow [3]. Based on the pressure flow characteristics in the open

reservoir one probably would have chosen a 14 Fr cannula whereas the 12 Fr

performed almost as good in our experimental model. The 13.2 Fr did not

improve return compared to the 12 and 14 Fr cannulas.

Augmenting venous return by vacuum assist resulted in higher flows

compared to gravity drainage. The difference is most pronounced in the 12 Fr

and 14 Fr angled cannulas as well as with the 13.2 Fr. Due to the fact that the

16 Fr cannula has a diameter equal or even somewhat larger then the vein

diameter (cannula vein ratio: 1.05), the vein will become the limiting factor. In

our experimental model the smallest cannulas, being the 12 Fr and 13.2 Fr,

therefore had the best performance. This might be explained by the fact that

with a small cannula the greater pressure gradient will be between the

reservoir and the tip of the cannula instead of between the tip of the cannula

and the patient’s venous system. This might prevent severe ‘fluttering’ of the

walls of the IVC around the end of the venous cannula [3,9]. The helix design

resulted in excellent flow rate most probably related to its large open area for

flow.

Monitoring of the vein pressure is of major importance for preventing

‘fluttering’ and collapse of the vein, which will result in flow reduction and

haemolysis due to high shear stress at the cannula entrance [10].

Appendix 1

71

Unfortunately, in most articles the authors only report the pressure measured

on top of the venous reservoir [3-8,10]. However this pressure will be the sum

of the pressure at the tip of the cannula, the hydrostatic pressure and the

pressure loss over the tubing and cannula. The latter makes it very difficult to

compare results of different studies since most authors do not mention tubing

length and hydrostatic pressure. The flow regimen in a piece of 3/16 or ¼ inch

will be laminar with the blood flows, blood temperatures and haematocrit

conditions used during CPB on a baby of 5 kg. Using the Hagen-Poiseuille

equation (See appendix) one can calculate the contribution of the tubing in the

pressure difference measured on top of the venous reservoir. When a baby of

5 kg, with a haematocrit of 25% and a blood temperature of 25°C, is perfused

with a flow of 120 mL/kg, the pressure difference for each meter of 3/16 or ¼

inch tubing will be 15 mmHg and 5 mmHg, respectively. When the oxygenator

is not at the same level of the right atrium we also have to add the hydrostatic

pressure head to the total pressure difference. This is approximately 8 mmHg

for each 10 cm height difference.

In summary, we found that collapse can occur when the venous pressure

becomes too low in the test system. This is important since most children are

selectively cannulated in both major veins. Measuring vacuum at the top of

the venous reservoir is not a good indicator of the pressure in the vein.

Monitoring of the intravenous pressure might help to prevent collapse of the

vein. A larger diameter venous cannula does not always produce the highest

flow when placed in a vein. This is most obvious when augmenting venous

return. The design of the cannula tip in combination with VAVD can affect the

Appendix 1

72

venous return. Development of smaller cannulas with tips adapted for the use

of VAVD should be stimulated.

Limitations of the study

Although major efforts have been taken to mimic the anatomical and

physiological situation, it is impossible to simulate all surgical events and their

impact on venous return. The absence of vascular tonus in the model might

also influence our results. For those reasons interpretation of the results must

be done with caution.

Acknowledgements

The authors received sample cannulas free of charge from Medtronic

(Medtronic, Brussels, Belgium) for testing purposes. Polystan (Polystan,

Oelegem, Belgium) kindly offered the vacuum controller for the duration of the

experiment. The authors thank Mrs. Oancea for her technical assistance.

Appendix 1

73

Appendix

By using these equations pressure drop can be calculated for a venous line in

function of length, diameter, required blood flow, viscosity and desired

The Reynolds number, which is dimensionless, represents the ratio of inertia

forces to viscous forces and is calculated by:

ReV d⋅ν

=

V = velocity [m/s], d = diameter [m], ν = kinematic viscosity [m²/s]

The kinematic viscosity, ν for blood is calculated according to following formula:

νη

ρ=

ρ = density [kg/m³], η = absolute blood viscosity [Pa.s]

ηplasma

exp 5.64−1800

T 273+( )+

1000=

η ηplasmaexp 2.31 Hct⋅( )⋅=

ρ 1.09 Hct⋅ 1.035 1 Hct−( )⋅+[ ] 10⋅=

If the Reynolds number is below 2000 flow is considered to be laminar. For

laminar flow, pressure drop can be calculated in function of diameter, length,

blood viscosity and height difference between the patient and the heart-lung

machine, using the Hagen-Poiseuille equation:

∆P128 η⋅ L⋅ V⋅

π D4⋅ρ g⋅ H⋅+=

Where L = lenght [m], Q = blood flow [m³/s], D = diameter [m], ∆P = pressure drop [Pa], H = height [m], g = gravity constant [m/s²]

Appendix 1

74

pressure drop. If we have a venous line of 1 meter, a haematocrit of 25% and

a temperature of 25°C, we would obtain following values:

Blood flow[mL/min]

Reynoldsnumber3/16 inch

∆P[mmHg]3/16 inch

Reynoldsnumber1/4 inch

∆P[mmHg]¼ inch

0 0 0.0 0 0.0100 176 2.6 234 0.8200 352 5.3 469 1.7300 527 7.9 703 2.5400 703 10.5 937 3.3500 879 13.2 1172 4.2600 1055 15.8 1406 5.0700 1230 18.4 1641 5.8800 1406 21.1 1875 6.7900 1582 23.7 2109 7.51000 1758 26.3 2344 8.3

Appendix 1

75

References

1. De Somer F, Foubert L, Poelaert J, Dujardin D, Van Nooten G, François K.

Low extracorporeal priming volumes for infants: a benefit? Perfusion 1996;

11: 455-460.

2. Galetti PM, Brecher GA. Connection of the vascular system with an

extracorporeal circuit. Heart-lung bypass, principles and techniques of

extracorporeal circulation. New York: Grune & Stratton, 1962: 171-93.

3. Kurusz M, Deyo DJ, Sholar AD, Tao W, Zwischenberger JB. Laboratory

testing of femoral venous cannulae: effect of size, position and negative

pressure on flow. Perfusion 1999; 14: 379-387.

4. Münster K, Andersen U, Mikkelsen J, Petterson G. Vacuum assisted

venous drainage. Perfusion 1999; 14: 419-423.

5. Lau CL, Posther KE, Stephenson GR et al. Mini-circuit cardiopulmonary

bypass with vacuum assisted venous drainage: feasibility of an

asanguineous prime in the neonate. Perfusion 1999; 14: 389-396.

6. Darling E, Kaemmer D, Lawson S, Smigla G, Collins K, Shearer I, Jaggers

J. Experimental use of an ultra-low prime neonatal cardiopulmonary

bypass circuit utilizing vacuum assisted venous drainage. JECT 1998; 30:

184-189.

7. Ahlberg K, Sistino JJ, Nemoto S. Hematological effects of a low-prime

neonatal cardiopulmonary bypass circuit utilizing vacuum-assisted venous

return in the porcine model. JECT 1999; 31; 195-201.

Appendix 1

76

8. R Berryessa, R Wiencek, J Jacobson, D Hollingshead, K Farmer, G Cahill.

Vacuum-assisted venous return in pediatric cardiopulmonary bypass.

Perfusion 2000; 15: 63-67.

9. Kirklin JW, Barratt-Boyes BG. Hypothermia, circulatory arrest, and

cardiopulmonary bypass. Cardiac Surgery, 2nd edn. New York: Churchill

Livingstone, 1993: 76.

10. Pedersen TH, Videm V, Svennevig JL et al. Extracorporeal membrane

oxygenation using a centrifugal pump and a servo regulator to prevent

negative inlet pressure. Ann Thorac Surg 1997; 63: 1333-39.

Appendix 1

77

Figure 1: Special constructed helix cannula.

Appendix 1

78

Figure 2: Experimental set-up

A: Set-up for pressure-flow relationship

B: Set-up for gravity and VAVD

Reservoir 1

Reservoir 2

Pump

TransonicFlowmeter

Reservoir 1

Reservoir 2

Pump

TransonicFlowmeter

Purse string

Cannula

Right atrium

Vacuumcontroller

Venousreservoir

Pressure 1

Pressure 2Differential pressuretransmitter

Reservoir 1

Reservoir 2

DPT

Weight balancePump

Cannula

TransonicFlowmeter

Pressure 2Pressure 1

Differerentialpressuretransmitter

Reservoir 1

Reservoir 2

DPT

Weight balancePump

Cannula

TransonicFlowmeter

Pressure 2Pressure 1

Differerentialpressuretransmitter

Switch between gravityand vacuumheight difference

mmHg

latex model

Appendix 1

79

Table 1. Performance of the cannulas in the reservoir and in the inferior caval

vein when using gravity drainage or VAVD.

Cannula typeReservoir

Gravity VAVD

Flow at minus 30cm H20 (mL/min)

Maximum flow before collapse of thevein (mL/min)

12 Fr straight 219 ± 20 136 15513.2 Fr straight 285 ± 13 142 16314 Fr straight 422 ± 11 142 12916 Fr straight 728 ± 4 149 143

12 Fr angled 216 ± 13 131 15614 Fr angled 454 ± 7 164 17916 Fr angled 727 ± 35 151 165

Mean collapsepressure (mmHg)

9.9 ± 1.2 12 ± 1.1

Appendix 1

80

Figure 3: Pressure – flow relationship.

100 300 500 700Flow [mL/min]

0

5

10

15

20

25

30

Pres

sure

at t

ip [m

mH

g]

Straight cannulae

12 French14 French16 French13.2 French

100 300 500 700Flow [mL/min]

0

5

10

15

20

25

30

Pres

sure

at t

ip [m

mH

g]

12 French14 French16 French

Angled cannulae

Appendix 1

81

Figure 4:

25 50 75 100 125 150 175 200Flow [mL/min]

0

5

10

15

Pres

sure

dro

p [m

mH

g]

12 French14 French16 French13.2 French

Gravity straight cannulae

25 50 75 100 125 150 175 200Flow [mL/min]

0

5

10

15

Pres

sure

dro

p [m

mH

g]

12 French14 French16 French

Gravity angled cannulae

Appendix 1

82

Figure 5.

25 50 75 100 125 150 175 200Flow [mL/min]

0

5

10

15

20

25

30

Pres

sure

dro

p [m

mH

g]

12 French14 French16 French13.2 French

VAVD straight cannulae

25 50 75 100 125 150 175 200Flow [mL/min]

0

5

10

15

20

25

30

Pres

sure

dro

p [m

mH

g]

12 French14 French16 French

VAVD angled cannulae

Appendix 2

83

Hydrodynamical Comparison of Aortic Arch Cannulae

P.R. Verdonck, U. Siller, D. De Wachter, F. De Somer, G. Van Nooten

Int. J. Art. Organs, 1998; 21(11): 705 - 713.

Appendix 2

84

Abstract

The high velocity of blood flow exiting aortic arch cannulae may erode

atherosclerotic material from the aortic intima causing non-cardiac

complications such as stroke, multiple organ failure and death. Five 24 Fr

cannulae from the Sarns product line (straigth open tip, angled open tip with

and without round side holes, straight and angled closed tip with four

rectangular, lateral side holes) and a flexible cannula used at the University

Hospital of Gent (straigth open tip) are compared in an in-vitro steady flow

setup, to study the spatial velocity distribution inside the jet. The setup

consists of an ultrasound Doppler velocimeter, mounted opposite to the

cannula tip in an outflow reservoir. An elevated supply tank supplies steady

flow of 1.3 L/min of water. Exit forces at various distances from the tip are

calculated by integrating the assessed velocity profiles. The pressure drop

across the cannula tip is measured using fluid filled pressure transducers. The

four sidehole design provide the lowest exit velocity (0.85 vs 1.08 m/s) and

forces per jet (0.03 vs 0.15-0.20 N). The round sideholes are useless as less

than 1 % of the flow is directed through them. Furthermore, the use of angled

tip cannulae is suggested because the force exerted on the aortic wall

decreases the more the angle of incidence of the jet deviates from 90°.

Pressure drop is the lowest for the 4 side hole design and highest for the open

tip and increases when an angled tip is used.

Keywords

aortic cannula, in vitro hydrodynamics, sandblasting effect.

Appendix 2

85

Introduction

Atherosclerotic disease of the ascending and transverse aortic arch is an

important risk factor for stroke associated with use of cardiopulmonary bypass

(CPB) [1,2]. Detachment of atherosclerotic material from the aortic intima can

be caused by external manipulation (such as cannulation and clamping) and

internal disruption. Tissue erosion in the aortic arch is caused by the high-

velocity jet emerging from an aortic cannula during CPB termed the

“sandblasting effect” [3]. The high speed jet is caused by the relatively small

cross section of the cannula tip which is around 8 mm in outer diameter for a

cannula used on adults with average blood flows of 4 to 6 l/min.

During the last decade lots of effort have been performed to design better

cannulae. Already in 1986 the use of a long aortic arch cannula with its tip

extending beyond the origins of the arch vessels was suggested because it

could avoid the hazard of stroke by directing the high-velocity blood flow down

the ascending aorta and away from the cerebral arteries [3]. Muehrche et al.

made a different approach by designing a new arterial cannula with four side

holes specifically to reduce the velocity of blood flow and the exit force in

order to decrease the sandblasting effect and to produce a more gentle high-

flow perfusion [4]. Nevertheless because it is difficult to verify the position of

the four jets inside the patient’s aorta one or more jets might still hit calcified

material.

Besides the above mentioned efforts cannula design needs to be improved

both hydro- and hemodynamically to reduce the rate of perioperative

problems. Influencing factors are multiple: pressure drop, flow rate, jet

Appendix 2

86

velocity, geometry of cannula, tip position in the aorta, shear stress, exit force

and operation time. To improve the performance of cannulae on the long term

it is necessary to evaluate the relationship between the hydrodynamic

parameters in an experimental setup.

Appendix 2

87

Materials and Methods

1. Tested cannulae

Five models out of the six cannulae tested in this study are selected from the

3M-Sarns product line and one is a self-made cannula, named “Gent Hospital”

in this paper, used at the University Hospital of Gent. The cannulae

distinguish each other by their geometries, dimensions and materials.

The geometry describes the general shape of a cannula. Most models are a

composition of three components: a connector, a tube and a tip. On one

model of the tested group the distinction between the tube and the tip is not

possible because they consist of the same piece (“Gent Hospital”). The

entrance of the tip is defined as the first deviation from the general tube

design. Figure 1 shows a schematic picture of a cannula. For all models the

connector is a standard 3/8 inch (0.95 mm) Polycarbonate connector

commonly used in the clinical practice.

Tips are straight or angled fitted with or without side holes. The shape of

these side holes is either round or rectangular. All tips except for one (“Gent

Hospital”) of the tested samples are conic to gently decrease the diameter of

the tube towards the end of the tip.

Important dimensions to characterize a cannula are the diameter and length

of the tip and the tube. The outer diameter of the tip is measured in French

(circumference in millimeters). All samples in the test are 24 French (8 mm)

cannulae.

In all cases PVC is used as material for both the tip and the tube. Due to

different additives (softeners) the stiffness of the tested examples at room

Appendix 2

88

temperature varies from soft (easy pliable) to stiff (not flexible at all). An

overview of all tested models is summarized in Table 1 in which GOA

represents the geometric orifice area calculated as π r2, with r the diameter of

the tip, augmented with the area of side holes if present. The flow through a

tube with inelastic walls depends on the velocity of the fluid and the effective

outflow area EOA which does not necessarily equal the GOA. For the side

hole cannulae the EOA will be determined experimentally from velocity flow

measurements.

2. Experimental setup

The setup consists of three parts:

- a system of reservoirs, tubing and a centrifugal pump to provide a constant

flow;

- an ultrasound Doppler velocimeter to measure the velocities of the jet at

various distances away from the tip and to visualize the contour of the jet;

- a data acquisition system to assess the pressure drop over the tube and the

tip as a function of the flow.

Figure 2 gives a schematic overview of the experimental setup.

The water is raised by a centrifugal pump from a tank to the upper reservoir

where it enters a vertical tube of 1250 mm in height. The water column in the

tube provides a constant bottom pressure of 95 mmHg because it is fitted with

an overflow that leads any surplus water back to the tank. An array of two

valves is attached to the outflow at the bottom of the reservoir to allow an

accurate adjustment of the flow. The connection to the cannula is made by the

same 3/8 inch tube that is used on extracorporeal circulation systems in the

Appendix 2

89

clinical practice. It is attached to a luer lockport that permits the introduction of

two fluid lines into the cannula.

The cannula is inserted into the outflow reservoir either through an opening in

the wall (straight tip) or from the top (angled tip). Inside of the container it is

fastened with a horizontally and vertically adjustable clamp. The variable

support is necessary to position the tip of the cannula exactly opposite to the

ultrasound probe. In order to have the option to measure the jet from the side

and from below two extra openings for the ultrasound probe are intended at

the bottom and on the sidewall respectively. The Plexiglas window on one

side of the container makes it possible to see the tip of the cannula.

A cylindrical reservoir is chosen to keep disturbing reflections of the ultrasonic

waves from sharp edges low. The outflow reservoir offers an inner diameter of

200 mm and measures 590 mm in height.

The water level inside of the container must stay constant to apply a positive

back pressure on the tip of the cannula. This is realized by an overflow which

is also connected to the tank by a plastic tube. By pulling the overflow pipe out

of its socket the reservoir can be emptied quickly.

The pressure is measured at two different positions inside the tested cannula

with fluid lines connected to piezoresistive transducers. The pressure in the

reservoir at the level of the tip can be computed by measuring the height of

the water column above the center of the cannula. Knowing the pressure in

three points (reservoir, connector and the beginning of the tip) for a given flow

makes it possible to calculate the pressure drop across the tip ∆ptip, the tube

∆ptube including the connector ∆pconnector and the total length of the cannula:

∆pTube = pConnector - pTip

Appendix 2

90

∆pTip = pTip - pReservoir

∆pCannula = pConnector - pReservoir

The output of the pressure readings takes place numerically and graphically

on the screen of the system with an accuracy of ± 0.5 mmHg within a range of

± 50 mmHg. The data acquisition software is developed at the Hydraulics

Laboratory of the University of Gent.

A clamp-on ultrasound flow probe (Transonic 3/8" Transonic Systems, Ihaca,

New York) attached to the connection tube between the upper reservoir with

the cannula is used to measure the average flow. To ensure that the flow is

fully developed at the position of the flow probe (even for laminar flow

conditions) the sensor is placed at a distance of one meter away from the

origin of the tube.

The velocities inside the jet are measured with (PWD) Pulsed Wave Doppler

echography (Vingmed CFM 800). All measurements are performed in a

detailed way and only high velocities (set arbitrarily above 0.9 m/s for all open

tip and round side hole cannulae, and to 0.6 m/s for all rectangular side hole

cannulae) are studied. The sampling volume is moved along a scan line

running parallel to the symmetry line of the cannula (reference line). The first

measurement is taken at this position which is still inside of the cannula. It is

recorded as picture number one. Picture number two is located one cursor

step to the left at the same distance away from the probe. The angle between

two scan lines is 1°. The cursor is moved further to the left until the detected

velocity is lower than 0.9 m/s or 0.6 m/s respectively. The part of the jet on the

right side of the reference line is scanned in the same manner. The same

procedure is carried out for all other distances from the transducer. The

Appendix 2

91

deepness is changed in intervals of 10 mm. Figure 3 shows the measuring

points that are accessed by using the protocol described above. All

measurements are performed for a constant flow of 1.3 l/min of water with 2

% cornstarch to improve the image quality. Besides PWD measurement also

Color Flow Doppler (CFD) image are studied in two perpendicular planes, a

horizontal and a vertical one.

3. Exit force

The harmfulness of the sandblasting effect depends on the vector of the exit

force of the jet that is perpendicular to the aortic wall (cosine term) (figure 4).

This is in contrast with the shear stress, which is caused by friction force

oriented laterally with respect to the wall. By definition the force is the product

of the pressure p and the area A the pressure works on:

F = pA cosα = ρ u2 A cosα

where u represents the velocity and α the angle between the cannula and the

horizontal plane (the angle of incidence). This can also be written as a

differential equation:

dFdA

= u cos2ρ α

The force is a function of the angle of incidence and the radius and velocity of

the jet. In addition the force depends on the distance from the point of the tip

which makes it a three dimensional problem.

To be able to integrate the formula it is assumed that the jet has a circular

cross section which is a function of the radius (A = πr2 so that dA = 2πrdr).

Furthermore an integration of the equation is only possible if a functional

Appendix 2

92

relationship for the velocity profile is given. For most situations the following

profile is applicable assuming a turbulent velocity profile : u(r) = (1 - r/R)1/n

Umax with R being the maximum radius and Umax the maximum velocity.

Prandtl derived n to 7 from Blasius’ law of friction [5].

Rearranging the equations and inserting the expressions for the assumed

velocity profile leads to:

αρπ cos 1 -

R - 1

2 +/n 2R +

Rr - 1

1 +/n 2R U 2- = F(r)

2 +/n 21 +/n 22max

rr

The force at a certain depth is obtained for r = R (n = 7):

αρπ cos R U 7249 = F 22

max

Appendix 2

93

Results

Figure 5 shows the comparison between the calculated geometric outflow

area (GOA) and the measured effective outflow area (EOA) for all tested

cannulae. These values deviate for cannulae with side holes.

Figure 6 displays the high velocity core of the jet obtained with PWD in the

horizontal plane (left panel) and the vertical plane (right panel) for the Sarns

9484. As mentioned before only a limited range of velocities is measured for

each jet. Plotting these velocities in a chart according to their measuring

position gives a good impression of the spatial distribution of the maximum of

the jet which is referred to as the core of the jet. Distances measured in

respect to the position of the ultrasound probe are drawn on the left hand side

of the diagram (Fig. 6) and distances measured from the point of the tip are

given on the right hand side with the tip being positioned at zero. At the

boundaries of the jet flowing water starts to mix with the resting fluid in the

reservoir. The core decreases in size with increasing distance from the exit of

the cannula. Nevertheless it is observed that even at a distance of 90 mm

away from the tip the maximum of the velocity in the center of the jet remains

constant. This suggests that a velocity drop-off regarding the peak velocity is

not yet present at that distance. These measurements together with the CF

Doppler measurements show an axisymmetric jet indicating that the angled

cut of the point of the tip does not effect basically the profile of the jet. Peak

velocities of all the cannulae vary between 0.85 m/s (“Sarns 5847" and “Sarns

5774") and 1.08 m/s (“Gent Hospital”) for a steady water flow of 1.3 l/min

(blood flow of 4 l/min). Figure 7 summarizes the measured peak velocities.

Appendix 2

94

For all cannulae it is observed that the peak velocities in the vertical plane are

slightly higher compared to the values measured in the horizontal plane

because reflections from the bottom of the reservoir and the water surface are

higher in this case.

An estimation of the maximum exit force of the tested cannulae is obtained

from the average measured values for the velocities and the available cross

section (Tab. 1). Figure 8 shows the variation of the exit force as a function of

the radius.

The peak pressure drop over the cannulae varies between 14.4 (“Sarns

5847") and 31.8 mmHg (“Gent Hospital”) for a water flow of 3 L/min which

corresponds to a pressure drop of 132 - 291 mmHg for a blood flow of 9

L/min.

Figure 9 shows the measured pressure drop for all cannulae. Besides the

pressure versus flow chart (Fig. 10, left panel) the data is also represented in

a dimensionless way where the Euler number is a function of the Reynolds

number (Fig. 10, right panel). The Euler number is defined asup =Eu

2ρ∆ and

the Reynolds number as Re = ρ u D/µ where ρ represents the density, u the

velocity, D the internal diameter and µ the dynamic viscosity. The graphs for

the pressure drop over the tube, the tip and the cannula are drawn separately.

As already can be seen from Figure 9 and 10 the total pressure drop across

the cannulae is mainly influenced by the tip except for the model “Gent

Hospital” where the tube plays the most important part because of its long and

thin origin.

Figure 11 summarizes the measured water flow across the cannulae for a

Appendix 2

95

constant pressure drop of 10.9 mmHg.

Appendix 2

96

Discussion

Sandblasting effect: peak velocities and exit force

Two out of six cannulae (“Sarns 5847" and “Sarns 5774") tested in this study

feature a new tip design where the blood flow exits through four lateral side

holes and not through an open tip like on the other models which is believed

to reduce peak velocities and exit forces [4].

To verify this statement the spatial velocity distribution of all cannulae is

measured with an ultrasound Doppler velocimeter and the exit forces at

certain distances from the tip are calculated out of the peak velocity and the

diameter of the jet by means of an integration.

Although aortic arch cannulae are routinely used during open heart operations

the importance of their hydrodynamics is somewhat overseen. In literature a

few papers are directly dealing with this topic. This is in contrast with the

knowledge that in terms of clinically significant central nervous system

dysfunction the most important embolic hazard of open heart operations in the

current area is atheroembolism from the ascending aorta.

The term “sandblasting effect” is used to describe the erosion process. In fact

this is misleading because it is actually the high pressure exerted on the aortic

wall rather than particles accelerated by a fluid (which would be the technical

meaning of the expression) that causes the dislodge of material. Erythrocytes

are too small and soft as to act like sand corns in the blood jet. Since the

terminology “sandblasting effect” is found in all of the reviewed papers it

seems to be an accepted phrase in the clinical field despite of its actual

meaning.

Appendix 2

97

To quantify the significance of the sandblasting effect Grossi et al. measured

intraoperatively the flow in the aortic arch of 18 patients undergoing CPB by

by epiaortic ultrasonography [3]. All patients were cannulated in the ascending

aorta, 10 with a short (15 mm) and 8 with a long (70 mm) cannula.

The peak forward aortic flow velocities measured on the caudal luminal

surface of the aortic arch were 0.80 m/s (± 0.23 m/s) when the CPB was

turned off and 2.42 m/s (± 0.69 m/s) on CPB for the short cannula. Using the

long cannula velocities of 0.53 m/s (± 0.20 m/s) and 0.18 m/s (± 0.10 m/s)

off/on CPB were measured respectively. For all measurements a handhold

probe was used connected to a Doppler velocimeter set to the continuous

wave mode.

Based on these measurements it was concluded that a long tip cannula

should be used in patients with an atheromatous aortic arch because it

confines the sandblasting effect to the descending aorta beyond the origins of

the cerebral vessels. Grossi’s results are somewhat questionable because

cannulae with different cross sections (long: 22 French versus short: 20

French) were compared. It is not surprising that the cannulae with the larger

cross section (long) provide lower peak velocities. Furthermore the length of

the tip has no influence on the quantity of the exit velocity assuming a

constant flow for all cannulae but only the pressure drop increases with

growing length.

The accuracy of the velocity measurements which is inadequate in one case

and it is peculiar that the long tip cannulae offer velocities that are even lower

than the physiological values in the aortic arch (0.18 versus 0.53 m/s). This

phenomenon is based on the fact that Doppler measurements are very

Appendix 2

98

direction sensitive (handhold probe) and that the continuous wave Doppler

mode (integration of all velocities on one line) rather than the pulsed Doppler

mode (local velocity measurements) had been chosen. Nevertheless our

velocity measurements obtained with PWD deviates also from data published

by Muehrche [4]. They reported very low peak velocities for a water flow of 2

l/min between 0.29 and 0.72 m/s. For example the peak velocity of the model

RMI ARS 024C, which is a straight open tip cannula with an internal diameter

of 6.1 mm, is measured at 0.57 m/s whereas a calculation would suggest a

value of around 1.41 m/s. An explanation for this deviation could be probably

found in an appropriate calibration of the laser Doppler anemometer. It is also

questionable if the measured velocity drop-off obtained with an ultrasound

velocimeter reflects the actual situation. Due to the limited lateral resolution of

ultrasound Doppler velocimeters it is likely to underestimate peak velocities if

the width of the jet is approximately of the same size as the width of the

sampling volume. The calculated velocity drop-off appears too low in this

case.

All open tip cannulae offer equal peak velocities and diameters of the jet

resulting in the same exit forces. The four side hole cannulae provide a larger

EOA which produces lower peak velocities and therefore reduces exit forces.

Pressure drop

The pressure drop across a cannula for a given blood flow is of concern in the

clinical practice because it adds to the total pressure loss of the CPB and

needs to be taken into account to adjust the roller pump of the extracorporeal

system previous to the operation.

Appendix 2

99

Due to this interest pressure versus flow charts are recorded for each of the

tested cannulae. Losses are measured for the tube and the tip of the

cannulae seperately to prove that the tip dominates the total loss. All pressure

versus flow charts are also presented in a dimensionless manner (Euler

versus Reynolds number) to advertise the benefits of dimensionless numbers.

For the flow of interest one has to compute the Reynolds number; look up the

corresponding Euler number in the graph and compute the resulting pressure

drop as ρ u2 * Eu, with u the mean velocity, which is the flow rate divided by

the cross-sectional area. There is also no need to rescale the graphs for blood

although the measurements are performed with water.

This is of great advantage when hemodilution and hypothermia are present,

because they alter the dynamic viscosity and therefore the pressure flow

relationship. However with the dimensionless numbers Re and Eu the graph

is normalized for a Newtonian fluid of any viscosity.

Montoya et al. proposed a standardized system to describe pressure versus

flow relationships in vascular access devices e.g. aortic arch cannulae [7].

Their request is that catheters are usually characterized by the French

number and length only. This description does not provide any information

about the pressure-flow relationship of the catheter nor does it allow for

performance comparisons between catheters.

Their system allows to characterize any vascular access device by a single

number denoted as “M” which may be determinated from the geometry or

from simple in vitro pressure-flow measurements. M is defined as log (LDC-

4.75) where L represents the length and DC the characteristic diameter of the

cannula. The system can be used by surgeons who wish to choose an

Appendix 2

100

appropriate catheter when size or pressure limitations are given or by

manufacturers who may supply M as a specification which will allow for

performance comparisons between catheters.

However the M number does not provide any new information because it

could be replaced by two already existing dimensionless numbers: the Euler

and the Reynolds number. Euler is defined as Eu = ∆p/ρu2 = λL/2DC for a

straight tube with λ a dimensionless friction number. Inserting Blasius’

equation for λ [5] and substituting the velocity by the flow Q, gives the

relationship Eu/Re-0.25 = 0.158 L/DC which is constant for a given geometry. If

all pressure and velocity values were provided in a dimensionless manner in

terms of Reynolds and Euler this approach would be an alternative to the M

number which might have some difficulties to become widely accepted.

Appendix 2

101

Conclusions

In summary the four side hole designs show a superior hydrodynamic

performance in the in vitro study compared to the end hole cannulae.

However the situation might look different in an in vivo setup. E.g. the amount

of the exit force exerted on the aortic wall depends very much upon the angle

of incidence. The jet of the straight open tip cannulae hits the aortic wall

almost perpendicular resulting in a high impact on possibly calcified tissue

whereas the jet of the angled tip cannulae hits the wall at a flatter angle which

results in a lower force on the aortic wall.

The four side hole designs are difficult to judge in this respect because one or

more of the four jets is likely to hit the aorta at a right angle. It must be taken

into account that the exit force is much lower compared to the other cannulae

but since the threshold value to erode calcified plaque is unknown it remains

questionable if the design offers a large advantage compared to the angled

open tip cannulae in an in vivo situation.

It is suggested to determine the threshold value for tissue erosion in an in vitro

setup before starting an in vivo investigation of the flow patterns of the

cannulae to be able to judge the force that needs to be applied to erode

calcified material.

Meanwhile it is advised to use angled tip cannulae to direct the blood flow

away from the aortic wall reducing the impact on the aortic intima.

Appendix 2

102

References

1. Katz E.S., Tunick P.A., Rusinek H., Ribakove G., Spencer F.C., Kronzon I.

Protruding aortic atheromas predict stroke in elderly patients undergoing

cardiopulmonary bypass : experience with intraoperative transesophageal

echocardiography. J. Am. Coll. Cardiol. 20:70-7, 1992.

2. Ribakove G.H., Katz E.S., Galloway A.C. et al. Surgical implications of

transesophageal echocardiography to grade the atheromatous aortic arch.

Ann. Thorac. Surg. 53:758-93; 1992.

3. Grossi E.A., Kanchuger S., Schwartz S., McLoughlin D.E., LeBoutillier M.,

Ribakove G.H., Marschall K.E., Galloway A.C., Colvin S.B. Effect of

cannula length on aortic arch flow : protection of the atheromatous aortic

arch. Ann. Thorac. Surg. 59:710-2, 1995.

4. Muehrche D.D., Cornhill J.F., Thomas J.D., Cosgrove D.M. Flow

characteristics of aortic cannulae. J. Card. Surg. 10:514-519, 1995.

5. Streeter V.L. Handbook of fluid dynamics. Mc Graw-Hill Book Company;

1961.

6. Guiot C. et al. Continous and pulsed Doppler power spectral density in

steady flow : an experimental investigation. Med. & Biol. Eng. & Comput.

35:146-159, 1997.

7. Montoya J.P. et al. A standardized system for describing flow/pressure

relationships in vascular access devices. ASAIO Transactions. 37:4-8,

1991.

Appendix 2

103

Table 1. Geometries, dimensions and materials of the tested 24 French

cannulae

Appendix 2

104

Figure 1. Schematic drawing of a cannula (a connector, a tube and the tip).

Connector Tube Tip

Appendix 2

105

Figure 2. Experimental in vitro setup.

UpperReservoir

Valves

3/8" Tube

FlowProbe

Luer LockCannula

Pressure Transducers

Pump

∆H

Outflow Reservoir

Lower Reservoir

UltrasoundProbe

UltrasoundMachine

Flow Meter

AD

Data Aquisition System

Computer

Appendix 2

106

Figure 3. Schematic representation of measured sample volume in the jet.

-20,00

-15,00

-10,00

-5,00

0,00

5,00

10,00

15,00

20,00

0 20 40 60 80 100 120 140

Depth (mm)

Wid

th (m

m)

Appendix 2

107

Figure 4. Calculation scheme for the exit force on the aortic wall.

α

d um

Tip

JetAortic wall

Flow profile

Core of jet

A

Appendix 2

108

Figure 5. Comparison between geometric outflow area and effective outflow

area of all tested cannulae.

0

10

20

30

40

50

60

70

80

90G

ent

Hos

pita

l

Sarn

s94

84

Sarn

s16

5264

Sarn

s44

01

Sarn

s58

47

Sarn

s57

74

Out

flow

cro

ss s

ectio

n (m

m2 )

GOA

EOA

Appendix 2

109

Figure 6. Measured velocity core of the jet for a “Sarns 9498" in a horizontal

plane (left panel) and a vertical plane (right panel).

Width (mm)-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7

Vel

ocity

(m/s

)

0,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

20

30

40

50

60

70

80

90

100

110

120

Dis

tanc

e fro

m s

enso

r (m

m)

-10

0

10

20

30

40

50

60

70

80

90

Dis

tanc

e fro

m ti

p (m

m)

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

Width (mm)-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7

Velo

city

(m/s

)

0,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

20

30

40

50

60

70

80

90

100

110

120

Dis

tanc

e fro

m s

enso

r (m

m)

-10

0

10

20

30

40

50

60

70

80

90

Dis

tanc

e fro

m ti

p (m

m)

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70,900,951,001,051,101,151,20

Appendix 2

110

Figure 7. Measured peak velocities.

0,75 0,75

0,97 1,00 1,00

0,85 0,85

1,05 1,07 1,07 1,081,05

0,00

0,20

0,40

0,60

0,80

1,00

1,20

Sarn

s58

47

Sarn

s57

74

Sarn

s94

84

Sarn

s16

5264

Sarn

s44

01

Gen

tH

ospi

tal

Peak

vel

ocity

(m/s

)

Appendix 2

111

Figure 8. Calculated exit forces (N) for all tested cannulae.

0

0,05

0,1

0,15

0,2

0,25

0,3

-0,01 -0,005 0 0,005 0,01

Radius (mm)

Exit

Forc

e (N

)

constant

n = 8

n = 7

n = 6

parabolic

linear

Appendix 2

112

Figure 9. Pressure drop (mmHg) for constant water flow of 1.3 l/min.

0

5

10

15

20

25

30

Sarn

s58

47

Sarn

s57

74

Sarn

s94

84

Sarn

s16

5264

Sarn

s44

01

Gen

tH

ospi

tal

Tota

l pre

ssur

e dr

op (m

mH

g)

Tip

Tube

Appendix 2

113

Figure 10. Pressure-flow (upper panel) and Euler-Reynolds (lower panel)

relationships for a “Sarns 9484".

0

2

4

6

8

10

12

14

16

18

0 0,5 1 1,5 2 2,5 3

Flow (l/min)

Pres

sure

dro

p (m

mH

g)

Cannula

Tip

Tube

0

0,5

1

1,5

2

2,5

3

0 1000 2000 3000 4000 5000 6000 7000

Reynolds

Eule

r Cannula

Tube

Tip

Appendix 2

114

Figure 11. Water flow (l/min) across the tested cannulae for a constant

pressure drop of 10.9 mmHg.

1,66

2,31 2,332,51 2,56

3,53

0

0,5

1

1,5

2

2,5

3

3,5

4G

ent

Hos

pita

l

Sarn

s44

01

Sarn

s16

5264

Sarn

s94

84

Sarn

s57

74

Sarn

s58

47

Flow

(l/m

in)

Appendix 3

115

Comparison of two dissimilar designs of paediatric aorticcannulae

D. De Wachter, F. De Somer, PR Verdonck

Int. J. Art. Organs, 2002; 25(9): 867 - 874

Appendix 3

116

Abstract

Any extracorporeal blood treatment requires an adequate and safe connection

to the circulation. For cardiopulmonary bypass procedures, aortic and venous

cannulas are utilized. However, the performance of these cannulas is not only

dependent on their size (diameter), but merely on their complete geometric

design.

In this paper two aortic cannula designs are evaluated haemodynamically for

two different sizes (8, 10 Fr) with both aqueous fluids as well as with blood.

Using the novel concept of equivalent diameter, a new performance

parameter, and the theory of dynamical similarity the results obtained with

different fluids can be compared. Data points of one cannula can be fitted with

a parabolic equation.

There is a significant performance difference between the two 8 Fr cannulas.

The 10 Fr cannulas differ non-signicantly except when water is used.

Equivalent diameters obtained with water in the turbulent region are

significantly higher than those obtained with fluids that have a higher viscosity

(blood and aqueous glycerine mixture). The latter fluids have comparable

viscosities and render an equal equivalent diameter. The coefficients of their

proper parabolic fit lines can be easily recalculated into each other.

This provides a simple method to quickly determine pressure drops over

cannulas in the operating room.

Appendix 3

117

Introduction

During open heart surgery procedures, the heart is arrested. To cover heart

and lung functions during this period, an extracorporeal shunt is established

over the heart (cardiopulmonary bypass), which includes a blood pump to

pump the blood through the extracorporeal circuit and back into the main

circulation, and an artificial lung for gas exchange. Blood is collected on the

venous side by means of a plastic cannula and drained by gravity into a

collecting reservoir. On the positive pressure side of the pump a plastic aortic

cannula is employed to inject the oxygenated blood into the intracorporal

circulation.

Pediatric aortic cannulas are made from soft plastic and on the one end they

allow a connection to PVC tubing, while at the other end they have a small tip,

suitable for introduction in the aorta of children. This tip is either in hard plastic

or it is a wire-reinforced. The dimensions of a paediatric cannula are a

compromise between technical requirements (blood flow vs. pressure drop)

and practical limitations (aortic diameter, small incision). Tip dimensions are

commonly stated in french (Fr); 1 Fr corresponds to an outer diameter of 0.33

mm. However inner diameters (ID) may differ significantly (Table 1).

The purpose of this study is to characterise the hydraulic resistance of

paediatric aortic cannulas on the one hand with an aqueous solution, similar

as has been done for adult size cannulas [1] and secondly with blood. The

second objective is to test whether these values can be used to determine

actual flow resistance in the operating room where the cannulas are perfused

by blood with different haematocrits and temperatures.

Appendix 3

118

Materials and Methods

4 different cannulas are included in the study (Table 1). All cannulas are from

DLP (Medtronic®). Of each type 2 specimens are studied. These cannulas

can be grouped in two ways: according to size (10Fr or 8Fr) or according to

tip design (series 750xx or 77xxx). The 750xx series cannulas have a short

light blue stiff plastic tip with a sudden diameter reduction (Figure 1). The

inner diameter (ID) of their tip is smaller than the ID of a 77xxx series cannula

of comparable French size. The latter series of cannulas have a long wire-

reinforced flexible tip with a smooth diameter reduction halfway the cannula.

The length of the 750xx series is shorter than the 771xx series, mainly

because the flexible tip of the latter can be introduced downstream into the

aorta. Pressure-flow relationships are assessed using three different fluids:

pure water, an aqueous glycerine solution (35 vol. %), both at room

temperature (20-25°C) and bovine blood with a hematocrit of about 32% (at

20°C and 37°C).

The test set-up consists of a tubing set, a variable speed rollerpump and a

small temperature controlled reservoir (Figure 2). The fluid is pumped from

the reservoir through the cannula and back in the reservoir. The tip of the

cannula is positioned in a long large bore tube (length: 50cm, ID=1/2 inch), to

minimise afterload influence. In fact this tube serves also as a sort of reservoir

in which kinetic energy of the fluid jet, that is propulsed from the cannula’s tip,

can be dissipated. Flow rate is measured with an ultrasonic transit-time

clamp-on flow meter (Transonic®, Ithaca NY, USA), that was previously

calibrated for each fluid by a volumetric method. Pressures are measured with

Appendix 3

119

fluid-filled piezo-electrical transducers (Ohmeda, Gent, Belgium), two-point

calibrated before each test run.

To determine a single data point, flow rate and pressures are recorded during

a finite time, through a computer data-acquisition card (PC74, Eagle

Technology, Cape Town, South Africa) and suitable software, typically at a

sampling rate of 200Hz. Due to roller pump operation, these curves exhibit a

periodical pattern. Actual data points are then determined by averaging these

values over an integral number of periods. All averaged data points are used

to fit a parabolic equation through the origin, stating the pressure drop (∆P)

flow rate (Q) relationship of a cannula. The parabolic equation is obtained by

polynomial regression (Sigmaplot, SPSS Inc., Erkrath, Germany).

Data obtained with different fluids render different parabolic equations.

However using the theory of dynamic similarity [2] (see Appendix), the

coefficients of these parabola can be recalculated and compared. Since

during clinical use, these cannulas are perfused with blood, the most useful

conversions are those to blood. The simplest method is to directly convert

pressures (P) and flow rates (Q) according to these ratios:

2

2

fb

bf

f

b

fb

bf

f

b

PP

QQ

µρ

µρ

µρ

µρ== (1)

where subscripts b denote blood and f any other fluid (water, aqueous

glycerine, ..). ρ and µ are respectively the density and the dynamic viscosity.

A second method rescales the coefficients of the fitted parabolic equation. If

the parabola is defined with ∆P the pressure drop and Q the flow rate:

bQaQP +=∆ 2 (2)

Appendix 3

120

Then the new coefficients are determined by rescaling with ratios of density

and dynamic viscosity:

f

bfb

f

bfb bbaa

µµ

ρρ

== (3)

The last method consists of the utilisation of dimensionless numbers, which

are independent of the fluid’s physical properties. The Euler number (Eu) is a

dimensionless measure of the pressure losses and the Reynolds number (Re)

of the flow rate:

µπρ

ρπ

e

e

DQ

QPD 4Re

16Eu 2

42

=∆

= (4)

The equivalent diameter De is determined from the effective diameters. The

effective diameter Deff is defined as the internal diameter that a circular tube

with the same length (L) as the cannula should have to exhibit the same

pressure drop as the cannula under study at a particular flow rate. At fully

developed turbulent flow (Re > 4000; water measurements), it is determined

from the Blasius equation (top of eq. 5). At laminar flow (Re < 2300; blood &

glycerine measurements) it is derived from the well-known Poiseuille equation

(lower part of eq. 5).

4

194

743

128

0541.0

PLQD

PQLD

eff

eff

∆=

∆=

πµ

µρ(5)

Deff should be independent on the fluid’s density (ρ) and dynamic viscosity (µ),

as their effects are cancelled by the flow / pressure drop ratio. However, when

using the Poiseuille equation, Deff depends on the flow rate because of special

pressure losses in the cannula that are not linearly proportional to the flow

Appendix 3

121

rate. As a reasonable approximation, it can be assumed that Deff varies

linearly with Re. Therefore the actual equivalent diameter (De) is obtained as

the Deff at Re=1000 on the linear regression line that is fitted through all Deff in

the laminar range for blood and aqueous glycerine and as the average of the

effective diameters determined from all measurements in the turbulent flow

range (Re> 4000) for water. The Blasius equation is utilized for water since

most water measurements fall in the turbulent flow regimen because of its

much lower viscosity compared with blood.

Dynamic viscosity of the fluids (µ) is either determined from literature data

(water [3], blood [4]), or measured in a viscometer (aqueous glycerine &

blood). Blood viscosity µb can be described by exponential functions, with µp:

plasma viscosity, T: absolute temperature (K) and Hct the fraction of red blood

cell volume [4]:

)35.2exp()180054.5exp(

HctT

pb

p

µµ

µ

=

+−=(6)

Appendix 3

122

Results

Viscosities of the fluids are respectively: 3.66 mPa.s for bovine blood at 37°C;

5.66 mPa.s for blood at 20°C; 3.36 mPa.s for 35%-65% glycerine-water

mixture at 20°C and 1.00 mPa.s for tap water.

In Table 2 the equivalent diameters of the four cannulas are presented. They

range from 2.5 mm for the 75008 8 Fr cannula to 3 mm for the 10 Fr

cannulas. The equivalent diameter obtained with water measurements is 6%

(75xxx series) to 10% (77xxx series) higher than for measurements with

blood.

The coefficients of the parabolic pressure equation (eq. 2) are listed in Table

3. The quadratic coefficients (a) are greater than the linear coefficients (b).

Both coefficients, but especially the linear coefficient (b) are generally greater

for the aqueous glycerine compared to the blood measurements, while for

water they are generally smaller. Also, the ratio of b/a is generally smaller for

water measurements. This ratio is a measure of the flow rate at which the

influence of the special pressure losses in the tip overhaul the pressure losses

in the cannula tube, the latter at these low flow rates being linearly related to

the laminar flow rate of the fluid.

In Figure 3 the pressure-flow relationships of all cannulas as measured with

bovine blood at 37°C are plotted along with their fitted parabolic regression

line proper. If the total pressure drop over each cannula would be limited to

200 mmHg, the maximal blood flow rate through the different cannulas is

respectively 0.78 L/min (77008); 1.05 L/min (77008); 1.60 L/min (77110) and

1.64 L/min (75010).

Appendix 3

123

In Figure 4 the Eu-Re relationship is shown for cannula 77008 on a semi-

logaritmic plot, with their respective transformed parabolic regression lines. As

is observed, the type of fluid does not influence to a great extent the position

of the plots that are in the laminar region (Re<2300, hollow symbols). For the

water measurements, which are mostly recorded in the turbulent flow region,

the curve is shifted up with respect to the laminar case. This indicates that the

dimensionless pressure drop in the turbulent region is larger than for laminar

flow. However, remark that the Euler number is proportional to the equivalent

diameter to the fourth power and that the equivalent diameter for water is

greater than that for the other fluids. Typically the region with Reynolds

numbers between 2000 and 4000 is regarded as a transition region. In this

region the behaviour of the fluid may be either laminar or turbulent or

somewhere in between. The only water measurement point at Re=1400 is

definitely measured with a laminar flow regime and shows a much lower Eu.

In fact this point lies amongst those measured with other fluids.

In Figure 5 the error of the calculated pressure-flow relationships for blood

with respect to the measured pressure-flow relationship is shown in the case

of the 77008 cannula. The coefficients of these calculated pressure-flow

relationships are found in Table 4. These coefficients are found by applying

either equations (3) or using the dimensionless numbers Eu and Re

(equations 4).

Appendix 3

124

Discussion

8 and 10 Fr cannulas are typical sizes used during paediatric surgery of

smaller childeren (up to 10kg). This group of patients constitutes

approximately 50% of the total paediatric population undergoing cardiac

surgery [5]. Size selection today is mainly done on anatomical grounds and on

arbitrary limits. Gates et al. [6] and Hessel [8] propose to limit the pressure

drop to 100 mmHg to avoid turbulent flow. More objective criteria are

necessary to make decent selections.

As is expected from theoretical considerations, cannulas that have a larger

internal diameter and a shorter length will exhibit the lowest pressure drop for

the same flow rate. According to the Poiseuille theory the diameter acts to the

fourth power, while the pressure drop is linearly proportional with the

cannula's length. Therefore, according to Table 1, the cannulas should be

ranked according to the highest pressure drop from 77008 over 75008 and

75010 to 77110. Indeed, this is exactly what is found, using the equivalent

diameter as a measure of performance (Table 2). In the water column this

ranking is indeed apparent, although the differences between the 10Fr

cannulas for blood and glycerine measurements are not significant. This is

also observed in Figure 3, where the two curves of the 10 Fr cannulas do not

differ much in the typical range of blood flow rates (up to 2 L/min). This may

be explained by the fact that the effect of the favorably shorter length of the

75010 is impaired by the much smaller inner diameter (ID) of its tip as

compared with the 77110 cannula (Table 1). For the 8 Fr cannulas the

difference in ID is not so great and therefore their curves do not overlap.

Appendix 3

125

If pressure would be the selection criterium, cannulas with greater equivalent

diameter are to be preferred (e.g. type 75008). If the equivalent diameter is

nearly equal and consequently the pressure gradient over the cannula, the

shortest cannula will produce the lowest pressure drop (e.g. 75010).

The equivalent diameter is shown to be a good predictor of performance as it

ranks the cannulas according to their pressure drop at the same flow rate.

However, the actual value is not useable in clinical practice, especially if it is

obtained by water measurements. In fact this has been tried before [7] and

promoted as the M-number. The reason is easily pointed out, as water

measurements tend to lie in the turbulent region (see Figure 4), whereas

during clinical operation blood is used and the flow is in the laminar region. As

can be observed in Table 2 the values obtained by water measurements are 6

to 10% higher than those obtained with either blood or aqueous glycerine.

However, this difference does not really have much meaning as the

equivalent diameter is arbitrarily determined at a Reynolds number of 1000.

Also, it is stated that the equivalent diameter is non-constant in the laminar

region. Typically it varies by up to 15% from the lower (Re=500) to high

Reynolds numbers (Re=2000). Typical Re values for blood flow through an 8

Fr cannula lie between 500 to 1500 and between 800-2500 for 10 Fr cannula,

depending on blood flow rate and viscosity. This indicates that Re=1000 is a

good compromise for the determination of the equivalent diameter, although it

remains an arbitrarily choosen number. On the other hand, Re=1000 is

considered to be critical for the flow through extracorporeal tubing sets by

Hessel [8]. No significant differences were found between equivalent

diameters of each cannula obtained by either blood or glycerine

Appendix 3

126

measurements. Therefore it can be concluded that the performance that is

assessed with glycerine is well matched by that determined by blood

measurements. Although aqueous glycerine is but a Newtonian fluid that

lacks the non-Newtonian characteristics of blood, it is apparent that the

comparable viscosities of both fluids are sufficient to render equal results.

One way to translate the measured pressure-flow rate relationship is by

utilizing equations (1). The only unknown to perform this conversion is the

blood viscosity which can be approximated by equations (6). However, due to

measurement errors and uncertaincies in the determination of the viscosities,

large deviations are possible, such that generally the error on the calculated

pressure is not less than 20%, except perhaps when using the measurements

obtained with blood at 20°C. These large errors can be expected in part

because of the conversion that has to be done. As is seen in Figure 4, the

water measurements for the 77008 cannula are performed at much higher

Re-numbers (but equal flow rates!) as the blood measurements, due to the

lower water viscosity. Since Re is inverse proportional to the viscosity, it can

be stated from equation (1) that the flow rates are scaled according to the

ratio of the Re numbers. The result is that the translated flow rates range from

0-6 L/min, much higher than the actual range used in clinical settings. Only

the measurements at low water flow rates translate into reasonable blood flow

rate values, but unfortunately these water measurements are made with

limited accuracy.

Therefore a better approach is to use the coefficients of the parabolic

equation (eq. 2) that describe the pressure drop for the cannula as a function

of blood flow rate and scale them according to equations (3). In this

Appendix 3

127

conversion the weight of the viscosity ratio is diminished and acts only on the

small linear coefficients (b). The more general method is with an Eu-Re

dimensionless analysis, which is quite similar, except that also the equivalent

diameters are taken into account (see Appendix). As a case, we have

selected cannula 77008 to test the potential of this conversion. This operation

renders new coefficients (Table 4), that allow to calculate the specific

pressure drop at a given flow rate. The relative errors of this calculation with

the actual blood measurements are plotted in Figure 5. At low flow rates the

errors are generally very high. This effect is caused by the lower relative

accuracy of the measurements in this range. For flow rates in the range from

0.4 to 1.2 L/min, the absolute error remains relatively constant for pressure

drops calculated from water measurements (15-20%) or from blood at 20°C

(10%; no difference between the two methods since the equivalent diameters

are equal). The latter is rather suprising, since it would be expected that the

blood measurement at 20°C is a good predictor of the values for blood at

37°C. However, the same effect as previously explained is active: the limited

accuracy of the blood viscosity measurement yields a new linear coefficient

(b) that is twice as high (see Table 4) as the original for 37°C blood (Table 3).

In Figure 5, the thick black lines show the error between the original blood

measurements at 37°C and the regression line fitted with the coefficients of

Table 3 on the one hand and secondly the lines that delineate the 99%

confidence intervals. This indicates that using the regression line for the

prediction of the pressure, a (maximal) error of up to 10% can be introduced.

So it seems that a 10% error in prediction is not bad at all. The only curves

that yield better results are the glycerine measurements. Since its viscosity is

Appendix 3

128

close to that of blood (respectively 3.36 and 3.66 mPa.s), the error in

predicting the pressure drop for blood remains small: between 0 and 12% for

the Eu-Re method and between -15 and -1% when using eq (3). It should be

remarked that the error is close to zero near Re=1000 for the Eu-Re method,

where the actual equivalent diameter is determined. Therefore, it is

hypothesized that the variable errors in this case are probably caused by the

variation of the equivalent diameter with the flow rate (or Re number). The Eu-

Re method is not the preferred method, as it requires the knowledge of the

equivalent diameters for both fluids. On the other hand, the error by using

glycerine and eq (3) remains small in the blood flow range of 0.8 to 1.2 L/min.

The final selection of a cannula in clinical practice mainly depends on

anatomical considerations and the surgeons' practice. If turbulent flow

conditions are to be avoided, the flow rate should be limited and the

equivalent diameter maximalized. In fact, turbulence will onset inside the

small bore tip. From the Re number (eq. 4), calculated with the tip diameter

(Table 1) the maximal laminar flow rate is obtained. E.g. for the blood

viscosity of 3.66 mPa.s in this paper at Re=2300, the maximal flow rates are

about 800 ml/min for the 8 Fr cannulas and 950-1050 ml/min for the 10 Fr

cannulas (respectively 75010 and 77110). At these flow rates the pressure

drop over the 75008 and 77110 cannulas is indeed about 100 mmHg. For the

77008 cannula it is doubled (202 mmHg) and for the 75010 it is only

68 mmHg, but at a sligthly decreased flow rate. This emphasizes again the

doubtfull nature of a pressure limit for cannula selection.

In conclusion, for the cannulas studied in this paper, we would advise to use

the 75008 when an 8 Fr cannula is desired, due to the lower pressure drop at

Appendix 3

129

equal flow rates and the 77110 cannula 10 Fr cannula, as it allows for higher

blood flow rates before turbulent flow occurs, albeit with slightly higher

pressure drops. Maximal clinical blood flow rates for the 8 Fr cannulas lie

around 750 ml/min and around 1 L/min for the 10 Fr cannulas.

Appendix 3

130

Conclusion

In this paper we have shown that the performance of paediatric cannulas for

cardiopulmonary bypass techniques can be readily evaluated by the concept

of equivalent diameter. To determine the actual pressure drop over the

cannula in the clinical setting, it is sufficient to measure the pressure-flow

relationship with a fluid that has a comparable viscosity as blood and translate

with the theory of dynamical similarity the coefficients of the fitted parabolic

regression line to the values for blood . A simple calculation with the resulting

quadratic equation and the desired blood flow rate is then able to render an

accurate estimation of the pressure drop.

Acknowledgement

Dirk De Wachter is a post-doctoral Fellow of the Fund for Scientific Research

- Flanders, Belgium.

The authors like to thank DLP - Medtronic to provide us with samples of the

cannulas. The kind assistance of Neil Uyttensprot and Jürgen Lissens during

the experiments is greatly appreciated.

Appendix 3

131

References

1. Verdonck P, Siller U, De Wachter D, De Somer F, Hydrodynamical

comparison of aortic arch cannulae, Int J Artif Organs, 1998; 21: 705-713.

2. Douglas JF, Gasiorek JM, Swaffield JA, Part III Dimensional Analysis and

Similarity. In: Fluid Mechanics, 3rd ed. Harlow UK: Longman Scientific &

Technical; 1985; ISBN 0-582-23408-5

3. Touloukian YS, Saxena SC, Hestermans P, Thermophysical properties of

matter, Vol 11: Viscosity. New York, USA: IFI Plenum; 1975: 643p.

4. De Wachter D, Modelling of dialysis treatment for renal failure. PhD

dissertation (in Dutch), Ghent University, Belgium; 1998.

5. Cecere G, Groom R, Forest R, Quinn R, Morton J, A 10-year review of

pediatric perfusion practice in North America. Perfusion, 2002; 17: 83-89

6. Gates RN, Cushen CK, Laks H, Cardiopulmonary bypass in infants and

children. In: Cardiopulmonary bypass. Principles and practice. Gravlee

GP, Davis RF, Utley JR (eds). Williams & Wilkins, Baltimore USA, 1993:

619-624

7. Montoya JP, Merz SI, Bartlett RH, A standardized system for describing

flow/pressure relationships in vascular access devices. ASAIO Trans.

1991; 37(1): 4-8

8. Hessel EA, Cardiopulmonary bypass circuitry and cannulation techniques.

In: Cardiopulmonary bypass. Principles and practice. Gravlee GP, Davis

RF, Utley JR (eds). Williams & Wilkins, Baltimore USA, 1993: 55-92

Appendix 3

132

Appendix: dynamic simililarity

A1. Conversion of pressures and flow rates

Dimensionless numbers are independent of the fluid’s physical properties.

Therefore the ratio of the Reynolds number (equation (4)) for any two fluids

should be unity. The same applies to the Euler number (equation(4)). Using

these properties, the ratio of the two flow rates and the ratio of the pressure

drops can be written as:

bf

fb

ff

bb

ff

fe

be

bb

f

b

f

b

bf

fb

ffe

f

b

bbe

f

b

f

b

QQ

QD

DQ

PP

DD

QQ

ρµρµ

ρρ

ρπ

πρ

ρµρµ

µπρ

ρµπ

2

2

2

2

2

42

42

2

1616

EuEu4

4ReRe

===∆∆

== −

−−

−

(a1)

The subscript f denotes an aqueous fluid and b stands for blood. Since the

equivalent diameter De should also be independent of the fluid’s properties, it

can be eliminated from the coefficient ratios. Equation (a1) is similar to

equation (1) of the main text.

A2. Conversion of the constants of the quadratic pressure drop equation

The parabolic relation between pressure drop en flow rate through the origin

(equation (2)) can be divided by the flow rate to render the flow rate

dependent hydraulic resistance:

baQQPRh +=

∆= (a2)

In dimensionless form this becomes with Eu (equation (4)) as a measure of

pressure drop, Re (equation (4)) as a measure of the flow rate and the Eu.Re

product as a measure of dimensionless hydraulic resistance:

Appendix 3

133

βαµ

πρ

πµ

π+=+=

∆= Re

4Re

164Re.Eu

3423eee DbDa

QPD (a3)

α and β are dimensionless coefficients that are independent of the fluid’s

physical properties. For the cannulas in this study, β ranges from a few

hundred to a few thousand and α lies between 1 and 2. The fluid dependent

coefficients a and b can be deduced from these dimensionless coefficients,

the fluid’s viscosity µ and density ρ and the equivalent diameter De:

342416

ee Db

Da

πµ

βπ

ρα == (a4)

The dimensions of these coefficients are in SI units: a in Pa/(m3/s)2 and b in

Pa/(m3/s), otherwise conversion constants should be added. The ratios of the

coefficients for different fluids are easily determined from equation (a4) and

simplified under the already stated assumption that the equivalent diameter is

independent of the fluid’s properties :

f

b

be

fe

f

b

f

b

f

b

be

fe

f

b

f

b

DD

bb

DD

aa

µµ

µµ

ρρ

ρρ

≈=≈=−

−

−

−3

3

4

4

(a5)

The subscript f denotes an aqueous fluid and b stands for blood. Since the

equivalent diameter De should also be independent of the fluid’s properties, it

can be eliminated from the coefficient ratios. Equation (a5) is equivalent to

equation (3) in the main text.

Appendix 3

134

TABLES

Type & Size Tip ID (mm) Length (mm)75008 8 Fr 2.03 17875010 10 Fr 2.46 17877008 8 Fr 2.13 22977110 10 Fr 2.79 229

Table 1: Pediatric cannulas types and dimensions (ID=inner diameter)

Blood (20°+37°C) Aq. Glycerine Water75008 2.530 ± 0.025 2.501 ± 0.005 2.681 ± 0.018 (ª)75010 ** 2.969 ± 0.049 ** ** 2.917 ± 0.042 ** * 3.143 ± 0.064 (º) *77008 2.271 ± 0.032 2.182 ± 0.061 * 2.497 ± 0.004 (") *77110 2.917 ± 0.038 2.991 ± 0.400 3.247 ± 0.125 (')

Table 2: Equivalent diameter (in mm) obtained with different fluids (mean ± stdev)

(º): p<0.05; ("): p<0.005; (ª): p<0.001 between water & blood;

*: p<0.05; **: p<0.005 between cannulas

Non-significant differences between blood and aqueous glycerine for each cannulas.

No significant difference between cannulas 75010 and 77110 (blood & glycerine)

Appendix 3

135

Blood 37°C Aq. Glycerine WaterB a b a b a

75008 15.54 165.02 26.01 173.17 2.32 136.4975010 9.80 68.43 12.30 77.13 10.23 55.3077008 36.55 269.92 85.75 236.35 9.99 200.6177110 35.64 55.58 44.86 66.42 9.78 52.85

Table 3: Coefficients of the parabolic regression line (a: mmHg/(L/min)²,

b: mmHg/(L/min)) as obtained with different fluids

Blood 20°C Aq. Glycerine WaterB a b a b a

Eq (3) 67.10 274.36 93.41 228.33 36.56 211.88Re-Eu 67.10 274.36 82.85 194.58 48.60 309.66

Table 4: Calculated coefficients as an estimate to describe the parabolic regression

line for blood 37°C in cannula 77008 using the regression coefficients from different

fluids.

Appendix 3

136

FIGURES

Figure 1. A picture of the two designs, shown here for the 10Fr cannulas.

The 75xxx series has a small short plastic tip. The 77xxx series has a long

wire-enforced tip.

Figure 2. The test set-up. P: pressure transducers.

Reservoir

Cannula

Rollerpump

Flow

P

P

Appendix 3

137

Blood flow rate (L/min)0,2 0,4 0,6 0,8 1,2 1,4 1,6 1,80,0 1,0 2,0

Pres

sure

dro

p (m

mH

g)

0

100

200

300

40075008a 77110a 77008a 75010a

Figure 3. Pressure flow relationships of different cannulas

and their parabolic regression line

Appendix 3

138

Figure 4. Eu-Re plot of 77008 cannula for different fluids and

their transformed regression lines

Re200 300 500 2000 3000 5000 200001000 10000

Eu

2

3

4

5

6

7

Blood 37°CAequous glycerine 25°CBlood 20°CWater 25°C

Appendix 3

139

Figure 5. Error in predicted pressure for blood flow at 37°C.

The pressure is predicted from curves measured with different fluids (symbols)

or from its fitted parabolic regression line and 99% confidence intervals (thick lines)

Flow rate (L/min)0,2 0,4 0,6 0,8 1,20,0 1,0

Erro

r with

resp

ect t

o bl

ood

37°C

(%)

-20

-10

0

10

20

30

40

50Blood 20°CWater eq (3)Water Eu-ReGlycerine eq (3)Glycerine Eu-ReParabolic fit blood 37°C

Re=1000

+99%

-99%

Appendix 3

140

Appendix 4

141

D-901 Neonatal oxygenator: a new perspective

F. De Somer, K. François, L. Foubert, Y. Deryck, D. De Smet, M. Vanackere,

G. Van Nooten

Perfusion 1994; 9: 349-355

Appendix 4

142

Abstract

Five infants with congenital heart disease were perfused with the D-901

neonatal oxygenator at the time of their cardiac surgery. The ability to reduce

the prime volume below the bloodvolume as well as the blood handling and

gas transfer characteristics were studied.

In all cases the prime volume was less than or equal to the bloodvolume of

the patient. This resulted in a reduction in the use of homologous blood

products. Due to the concept of the D-901 it was possible to adapt the tubing

in such a way that a complete prime of 220 ml was obtained.

The device had a maximum oxygen transfer of 45 ml/minute. The maximum

carbon dioxide removal was 50 ml/minute at a blood gas ratio of 1.

The mean platelet count post bypass decreased to 91% of the baseline value.

Mean free haemoglobin levels increased to 86 mg/100 ml at 120 minutes of

bypass.

We conclude that the D-901 oxygenator opens new perspectives for perfusion

in small babies in terms of priming volume and use of homologous blood

products while maintaining good gas transfer characteristics. However, larger

series are necessary to expand our experience with this device and its

limitations.

No specific problems related to the device were encountered and all infants

had an uneventful postoperative course.

Appendix 4

143

Introduction

Perfusion techniques in neonates and small infants vary significantly from

those in adults. However, little attention has been paid to the design of

oxygenators for these patients (1), the former being only smaller copies of

their larger brothers currently used in adult cardiac surgery. For this reason it

is very difficult to find oxygenators with a priming volume which is lower than

the total blood volume of the patient.

Recently Dideco (Mirandola, Italy) released a small oxygenator (D 901) which

is especially designed for use in infants up to 7 kilograms.

This device was tested in two ways. First the priming volume of the complete

system was investigated, and secondly the blood handling and gas transfer

characteristics were evaluated.

Appendix 4

144

Materials and methods

The D-901 was used in five babies undergoing elective cardiac surgery for

congenital heart disease (Table 1). The D-901 hollow fiber membrane

oxygenator is built around a central heat exchanger core. The heat exchanger

is made of molded stainless steel and has a surface area of 0.02 m2. The

oxygenator uses a polypropylene microporous hollow fiber mat to separate

blood and gas pathways. Blood flow is channeled around the outside of the

fibers and gas flows through the lumen of the fibers. The priming volume of

the heat exchanger/oxygenator structure is 60 ml. If the small flexible venous

reservoir is included, the minimal priming volume is 90 ml. The device has an

effective membrane surface area of 0.34 m2 and a nominal blood flow of 0.8

l/min. All connectors can be used with both 3/16" and 1/4" tubings.

The extracorporeal system comprised a Cobe heart lung machine (Cobe

Cardiovascular, Arvada, CO, USA), custom tubing packs made of PVC and

silicone (International Medical Products, Brussels, Belgium) and a Dideco

Midicard cardiotomy reservoir (Dideco, Mirandola, Italy). No arterial filter was

used in the system. Since the priming volume of the circuit becomes more

important when the priming volume of the oxygenator module decreases, we

adapted our circuit to this new situation. We used an 3/16" PVC arterial line of

150 cm and a 1/4" silicone venous line of 90 cm. The pump boot was 1/4"

silicone tubing with a wall thickness of 3/32". This setup resulted in a final

priming volume of 220 ml. One must take into account that the use of every

vent or sucker during the procedure will take away a considerable amount of

blood out of the circulation. For this reason one could be obliged to fill the

circuit with extra fluid to compensate for this loss. To anticipate this problem

we reduced the diameter of all venting and sucker lines to 3/16". This resulted

in a reduction of the dead volume in venting and sucker lines by 50%.

Appendix 4

145

Perfusion technique

Each oxygenator was inspected and set up in accordance with the

manufacturer's enclosed instructions. The circuit was flushed with carbon

dioxide prior to gravity priming. Blood flow rates were maintained to ensure

adequate tissue perfusion. An alpha-stat regimen (2) was used in all cases for

acid/base and blood gas management. Gas flow was delivered through

Sechrist air/O2 blenders (Sechrist Industries, Anaheim, CA, USA) with sweep

rates sufficient to maintain uncorrected PaCO2 within a normal range or

subnormal range in patients with pulmonary hypertension. Continous PaO2

measurements were done using the Polystan polytrode (Polystan A/S,

VærlØse, Denmark). Gas analysis (oxygen and carbon dioxide concentration)

was performed on both inlet and outlet of the oxygenator (Ohmeda RGM

5250). Activated coagulation times were kept above 400 seconds during

bypass. St Thomas II solution (20 ml/kg) was used for cardioplegic arrest in all

cases.

The priming consisted of a mixture of 20% human albumin, 15% mannitol (0.5

g/kg) and plasmalyte A. Packed red blood cells were added if necessary to

obtain a haematocrit of 30% during the extracorporeal circulation (ECC).

The lowest temperature during the procedures was 25° Celsius

(oesophageal). The patients were vasodilated to keep the mean arterial

pressure during ECC between 40 and 50 mmHg as described by the Marie-

Lannelongue group in Paris (3).

Data collection

ECG, central venous pressure, arterial pressure, pump flow, core

temperature, pressure drop over the oxygenator, arterial, venous and water

temperature were recorded manually every 15 minutes. Arterial and venous

blood gas samples together with electrolytes (sodium and potassium) were

analysed on a Corning 288 blood gas analyser (Ciba Corning, Medfield,

Appendix 4

146

USA). White blood cell count, red blood cell count, haematocrit, haemoglobin

and platelets were processed using STKS-Coulter counter apparatus. Serum

concentrations of free haemoglobin and haptoglobin (Hp) (as markers of

haemolysis) were determined using immunonephelometry on a BN

nephelometer (Behringwerke, Marburg, Germany). Blood samples were taken

prior to institution of bypass, after mixing (fifteen minutes bypass), every thirty

minutes during bypass, post bypass and daily during the first three

postoperative days. Total protein concentration used to estimate oncotic

pressure, was measured the day before surgery, immediately post bypass

and the first three postoperative days.

Data analysis

Following parameters were calculated: oxygen consumption (VO2/min), shunt

fraction of the oxygenator (Qs/Qt) and the oxygen transfer slope (OTS).

Calculations and techniques to obtain those values have been published

previously (4).

The correction for haemodilution of the platelet values was obtained

considering the baseline platelet count in the prime to be zero. Even when

blood prime was used no platelets were added to the prime as packed red

cells where used instead of whole blood. However one unit of packed red

cells does contain an important amount of free haemoglobin (mean 87

mg/100 ml) and haptoglobin (mean 0,24 g/l), which was added to the ECC in

the four patients where blood priming was used. Therefore haemolysis

markers were also quantified as the increment between measurements.

Appendix 4

147

Results

Priming volume

In cases 1 and 2 a priming volume of 240 ml for the whole system was

obtained. In cases 3 to 5 the priming volume had decreased to 220 ml by the

use of slightly shorter tubing. The latter was a reduction with 37% compared

to our previous system (Cobe VPCML, Pall arterial filter, 3/16” arterial line,

1/4” venous line) (Figure 1). In all cases, except one, it was possible to obtain

a priming volume which was lower than the estimated blood volume of the

patient (Figure 2). Recently the priming volume was reduced to 200 ml by

changing the pump boot tubing from 1/4" to 3/16".

In one case we did not add homologous blood.

No extra fluids were added to the circuit during ECC in two patients. In the

three other cases an average of 80 ml was added. Mean diuresis in all cases

during ECC was 31 ml (Table 2).

Gas transfer analysis

The oxygen transfer slope (change in FiO2/change in VO2/min) for the D-901

is shown in Figure 3. The maximal oxygen transfer was 45 ml/min or 132

ml/min/m2. Maximum carbon dioxide removal was 50 ml/min at a blood to gas

ratio of 1. Mean venous oxygen saturation was 64%.

Haemolysis

Serum free haemoglobin increased from a mean of 7.84 mg/100ml pre

bypass to 86 mg/100ml at 120 minutes of ECC. The increment in free

haemoglobin between the first measurement during ECC (15 minutes) and

120 minutes of ECC was 54 mg/100ml.

Appendix 4

148

Serum Hp level fell from 1.27 g/l prior to institution of bypass to 0.38 g/l at 120

minutes of ECC. The decrease between the first measurement during ECC

and after 120 minutes of ECC was 0.26 g/l.

Haematology and blood chemistry

Mean haematocrit during ECC was 30%. Mean platelet count (Figure 4)

decreased slightly during ECC. A mean of 91% of the baseline value was

obtained post ECC. Mean total protein level decreased from 65.3 g/l on the

preoperative day to 44.4 g/l immediately post ECC. At the end of the study the

total protein level was 54.4 g/l.

In the patient where no blood was added to the prime, haematocrit was

28.8%, platelet count 111% of the baseline value and protein level 51 g/l at

the end of bypass.

Appendix 4

149

Discussion

For many years there is great demand for smaller oxygenators especially

designed for the difficult problems encountered in cardiac surgery for

babies.(1,3,5,6,7) The D-901 is designed to deal with such problems. Due to

the unique design of heat exchanger and connectors a low complete system

prime can be obtained when used with short appropriate tubing. This may

result in a reduction in coagulation disorders and complement activation. The

reduced use of homologous blood products however is important (8). The

device allowed us to avoid the use of homologous blood products in one

patient in whom haematocrit, platelet count and protein levels were preserved

after ECC.

The D-901 gas transfer characteristics meet the requirements set by the

Association for the Advancement of Medical Instrumentation (9) and

compares favourably with other devices. The device is very predictable and

by using the oxygen transfer slope it is easy to anticipate sudden changes in

metabolic needs of the babies. A reliable continous venous oxygen saturation

device operating with an acceptable error at low flows (up to 0.5 l/min) would

make it even easier to steer the oxygenator.

The miniaturisation of tubing and connectors did not result in a higher

haemolysis in the range of blood flows used (3). There was also no

remarkable loss of platelets showing an acceptable biocompatibility of the

device.

We could easily prevent a drop in protein levels, thus reducing the risk for

capillary leak (3). There was no need for ultrafiltration in our series.

Conclusion

We conclude that the D-901 oxygenator opens new perspectives for perfusion

in small babies in terms of priming volume and use of homologous blood

Appendix 4

150

products while maintaining good gas transfer characteristics. However, larger

series are necessary to expand our experience with this device and its

limitations.

Appendix 4

151

References

1. Menghini A. Oxygenation design: a global approach. Perfusion 1993; 8:

87-92

2. Swan H. Acid-base management during hypothermic circulatory arrest for

cardiac surgery. In:Rahn H, Prakash O. eds. Acid-base Regulation and

Body Temperature. Boston: Martinus Nijhoff

3. Nicolas F, Daniel J-P, Bruniaux J et al. Conventional cardiopulmonary

bypass in neonates. A physiological approach - 10 years of experience at

Marie-Lannelongue Hospital. Perfusion 1994, 9: 41-49

4. De Somer F, De Smet D, Vanackere M et al. Clinical evaluation of a new

hollow fibre membrane oxygenator. Perfusion 1994, 9: 57-65

5. Molina G Neonate, Infant and pediatric perfusion: a Review of recent

product selection. Presentation held at the American Academy of

Perfusion Siences meeting on 2/1991 in San Francisco.

6. Elliot M. Minimizing the bypass circuit: a rational step in the development

of paediatric perfusion. Perfusion 1993; 8: 81-86

7. Elliot M, Rao PV, Hampton M. Current paediatric perfusion practice in the

UK. Perfusion 1993; 8: 7-25

8. Tyndal M, Berryessa RG, Campbell DN, Clarke DR. Micro-Prime Circuit

Facilitating Minimal Blood use during Infant Perfusion. J. Extra-Corpor.

Technol. 1987, 19: 352-357

9. Association for the Advancement of Medical Instrumentation. Standard for

blood/gas exchange devices-oxygenators. 1982.

Appendix 4

152

Table 1: Patient population.

Appendix 4

153

Table 2: Patient data, fluid administration and priming volume.

Appendix 4

154

Figure 1 Evolution of priming volume for babies smaller than 7 kg in UZ Gent.

VPCML = Variable Prime Cobe Membrane Lung; D-901 = Dideco Lilliput 1

oxygenator; AF = Pall 1440 arterial filter; AL = arterial line; VL = venous line;

PB = pump boot.

0

50

100

150

200

250

300

350

400

450

500

1990 1991 1993 1994

ml

Appendix 4

155

Figure 2 Blood volume versus priming volume.

0

100

200

300

400

500

600

700

800

900

1 2 3 4 5

Case

ml

Priming volume Blood volume

Appendix 4

156

Figure 3 Oxygen transfer slope (OTS)

y = 0.018x + 0.1664

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 5 10 15 20 25 30 35 40 45

Oxygen transfer (ml/min)

FiO

2 (%

)

Appendix 4

157

Figure 4 Platelet count expressed as percent of baseline.

Prae = 30’ pre ECC; Post = immediately post ECC; PO = postoperative day.

Appendix 4

158

Appendix 5

159

Low extracorporeal priming volumes for infants: a benefit?

F. De Somer, L. Foubert, J. Poelaert, D. Dujardin, G. Van Nooten, K. François

Perfusion 1996; 11: 455-460

Appendix 5

160

Abstract

An extracorporeal circuit consisting of an oxygenator especially designed for

neonatal use and appropriately sized tubings, with an average total priming

volume of 205 ml, was used on 80 infants undergoing cardiac surgery for

congenital heart disease. The priming volume and foreign surface area of the

circuit were determined. The influence of low priming volumes on the use of

blood products and the management of cardiopulmonary bypass was studied.

No whole blood nor platelets were used in this study. The mean volume of

packed red blood cells used over the hospital stay was 202 ± 67 ml. The

mean volume of fresh frozen plasma (FFP) used until the second

postoperative day was 62 ± 72 ml. The mean total blood loss until the second

postoperative day was 15.8 ± 9.2 ml/kg.

The priming volume of the extracorporeal circuit was 62 % lower than values

commonly reported in the literature. The low priming volume had a strong

influence on the use of platelets and FFP and in a lesser extent on the use of

packed red blood cells.

Appendix 5

161

Introduction

For many years the average priming volume of an extracorporeal circuit for

infants up to 8 kg is approximately 500 ml (1). After the start of

cardiopulmonary bypass (CPB) the cardioplegia volume is almost immediately

added to this priming volume (PV), which results in a massive haemodilution.

The latter may cause serious side effects such as alterations in coagulation

mechanisms and changes in extracellular/interstitial fluid distribution (2).

In an attempt to prevent this extreme haemodilution several techniques have

been described. Reduction of the priming volume by tailoring the circuit (3)

and modified ultrafiltration are well known techniques(4). Until recently the first

technique was not very popular because of the lack of specific neonatal

oxygenators and the fact that perfusion techniques had to be changed

dramatically. The latter removes water from the patient at the end of bypass

and perfuses the pulmonary artery with oxygenated blood at the same time.

However, it may not prevent the onset of the inflammatory response by the

large foreign surface at the beginning of and during CPB.

Recently two new oxygenators especially designed for neonatal use were

released: Dideco D-901 Lilliput (Mirandola, Italy) and Polystan Microsafe

(Polystan, VærlØse Denmark). Both of them can be easily used with priming

volumes as low as 200 ml (5). The aim of this study was to determine the

priming volume and foreign surface area of a low volume system and to

evaluate its influence on the use of blood products and the management of

CPB.

Appendix 5

162

Materials and Methods

The D-901 (n=76) and the Microsafe (n=4) were used in cardiac surgery for

congenital heart disease (Table 1). The D-901 is a closed system and was

used together with a Midicard (Dideco, Mirandola, Italy) cardiotomy reservoir.

This reservoir automatically reduces the filter material depending on the

suction volume. Since suction never exceeded 1000 ml a minute, only one-

third of the filter medium was used in all cases. The Microsafe is an open

system and has a venous reservoir of 400 ml. Since this can be rather small

for larger infants (5-8 kg) it was only used in infants below 5 kg, in contrast to

the D-901 which was used for infants up to 8 kg. Both systems have quite

comparable characteristics (Table 2) and both oxygenators can be used with

3/16 inch or 1/4 inch tubing.

The extracorporeal system consisted of a Cobe heart lung machine (Cobe

Cardiovascular Inc., Arvada, CO, USA), custom tubing packs of

polyvinylchloride (PVC) and silicone (International Medical Products,

Brussels, Belgium) and a Dideco Midicard cardiotomy reservoir (Dideco,

Mirandola, Italy). No arterial filter was used in the system. Since the relative

importance of the circuit increases when the priming volume of the oxygenator

decreases, we adapted our circuit to this new situation. We used an 3/16 inch

PVC arterial line of 150 cm and a 1/4 inch silicone venous line of 90 cm. The

pump boot was 3/16 inch silicone tubing with a wall thickness of 3/32 inch.

This set-up resulted in a final priming volume of 205 ml. One must take into

account that the use of every vent or sucker during the procedure will remove

a considerable amount of blood from the circulation. For this reason one could

Appendix 5

163

be obliged to fill the circuit with extra fluid to compensate for this loss. To

anticipate this problem we reduced the diameter of all venting and suction

lines to 3/16 inch. This resulted in a 44 % reduction of the dead volume in

venting and suction lines.

Perfusion Technique

Each oxygenator was inspected and set up in accordance with the

manufacturer's instructions. The circuit was flushed with carbon dioxide prior

to gravity priming. Since the system was primed with a 50 ml syringe the

exact priming volume was easy to determine. Blood flow rates were

maintained to ensure adequate tissue perfusion. An alpha-stat regimen (6)

was used in all cases for acid/base and blood gas management. Gas flow

was delivered through Sechrist air/oxygen gas blenders (Sechrist Industries,

Anaheim, CA, USA) with sweep rates sufficient to maintain uncorrected

PaCO2 within a normal range or subnormal range in patients with pulmonary

hypertension. Continuous PaO2 measurements were done using the Polystan

Polytrode (Polystan A/S, VærlØse, Denmark). Gas analysis (oxygen and

carbon dioxide concentration) was performed on both inlet and outlet of the

oxygenator (Ohmeda RGM 5250). Activated coagulation times were kept

above 400 seconds during bypass. St Thomas II solution (15-20 ml/kg) was

used for cardioplegic arrest in all cases.

The priming consisted of a mixture of 20% human albumin, 15% mannitol (0.5

g/kg) and Plasma-Lyte-A. Packed red blood cells were added if necessary to

obtain a haematocrit of 30% at the end of CPB.

The lowest oesophageal temperature during the procedures was 15° for deep

hypothermic circulatory arrest (DHCA) and 25° for continuous flow. The

Appendix 5

164

patients were vasodilated to keep the mean arterial pressure during CPB

between 30 and 40 mmHg as described by others (7).

Data Collection.

Haematocrit and platelets were determined the day before the operation, at

the end of the operation and two days postoperatively. The total amount of

blood products (packed red blood cells (PC), fresh frozen plasma (FFP),

platelets (Plts)) used in the priming, in the perioperative and postoperative

period were noted, as was total blood loss. One patient was removed from the

due to a perforation of the right atrium by a central venous catheter in the

postoperative period.

Blood loss was compared between patients who received FFP and those who

did not. Blood loss was expressed as ml/kg.

Haemodilution was calculated by following formulae:

Total blood volume (TBV) = weight x 85 ml

Red cell volume (RCV) = TBV x haematocrit (Hct)

Haemodilution (HD) by the priming volume = RCV/(TBV+PV)

Complete HD = RCV/(TBV+PV+CPL) where CPL = cardioplegia volume

Surface area of the tubing was calculated using following formula,

Surface = Πdh where d = inner diameter in cm and h = height in cm.

Volume in the tubing was calculated using following formula: Volume = Πr2h

where r is radius (cm) and h is height (cm). The values used for the

oxygenators were those mentioned in the brochures of the companies. The

information not available in the brochures was sent to us by the research and

development department of both companies.

Appendix 5

165

Results

Surface area and priming volume.

The surface area of the membrane, heat exchanger and their housing was

3750 cm2 for the Lilliput and 4060 cm2 for the Microsafe. The overall foreign

surface area (including the venous bag) was 3815 cm2 for the Lilliput plus

1725 cm2 for the cardiotomy reservoir resulting in a total of 5540 cm2 .

Including the venous reservoir, defoamer and filters the total foreign surface

area was 4710 cm2 for the Polystan Microsafe.

The surface of the extracorporeal lines (arterial line, venous line, pumphead)

was 554 cm2.

The total surface area of the neonatal systems was 6094 cm2 (D-901) and

5264 cm2 (Microsafe) respectively.

The priming volume for each system varied between 180 and 250 ml.

Use of blood products.

The mean use of packed red cells until the second postoperative day was 202

± 67 ml. The mean use of packed cells in the priming volume was 93.5 ± 60

ml. Three infants (3.7%) did not receive packed red cells in the priming

volume nor in the perioperative period. Twenty infants (25%) did not receive

packed red cells in the priming volume.

The mean use of FFP until the second postoperative day was 62 ± 72 ml. The

mean use of FFP in the priming volume was 2 ± 19 ml. In total 30 infants

(37.5%) of whom 4 below 3 kilogram did not receive FFP during their hospital

stay. Sixty eight infants (85%) did not have FFP in the priming volume.

No homologous platelets nor whole blood was used.

Appendix 5

166

Haematology.

Platelet count the day before the operation was 378 ± 144 x 1000/mm3. It

decreased towards the end of the operation to 156 ± 71 x 1000/mm3. At the

second postoperative day the value was 251 ± 140 x 1000/mm3.

Mean haematocrit the day before the operation was 38 ± 8 %. The mean

lowest value during CPB was 26 ± 4 %. At the end of CPB a mean

haematocrit value of 29 ± 3% was obtained. At the second postoperative day

mean haematocrit was 34.7 ± 4 % .

Blood loss.

The overall mean blood loss was 15.8 ± 9.2 ml/kg. There was no statistical

difference (ANOVA) in blood loss between those patients who received FFP

and those who did not.

Appendix 5

167

Discussion

For many years the use of CPB in cardiac surgery for congenital heart

disease has induced a massive haemodilution up to 300% in infants below 5

kg (7). This causes some adverse effects such as decrease in the

concentration of nutrients as well as in oxygen content of blood, alteration in

coagulation mechanism with potential for increased bleeding,

extracellular/interstitial fluid accumulation, redistribution of coronary blood flow

with myocardial ischaemia and possible contribution to immunosuppression

with increased risk of infection (2). This study demonstrates that reduction of

priming volume and hence, limitation of haemodilution can be performed

safely. Compared with the average priming volume of 500 ml reported in the

literature (1,7-9) our neonatal system offers a reduction of 60% in priming

volume. As a result, the risk for adverse effects may be limited. In infants

below 5 kg the venous line, which was 1/4 inch in our study, can be changed

to 3/16 inch, decreasing the priming volume to 180 ml. Due to the small

priming volume the amount of foreign surface area exposed to blood is less

than half that compared to a conventional system (e.g. the Cobe VPCML used

on its smallest compartment has a foreign surface area of 10975 cm²

(Personal communication with Mark Miller, Cobe Cardiovascular Inc., Arvada,

CO, USA) . This concept of low volume-low foreign surface area might be

beneficial, although still in debate, for reducing complement activation (10,11).

If, from a theoretical point of view a conventional system of 500 ml had been

used in CPB for the infants in our study, and if we had used the same

management of CPB, the calculated mean amount of packed red cells used in

Appendix 5

168

the priming volume would have been 245 ± 50 ml. This is 62 % more than

what was used in our study, and even 8 % more than the total amount of

packed red cells used during the whole hospital stay of the infants in the study

group.

Beside the decreased need for packed red cells, the use of other homologous

blood products is also favourably influenced since neither homologous

platelets neither whole blood were administered. The use of FFP was also

limited to a minority of the patients. Since there is no difference in blood loss

between the group which received FFP and the group which did not, the use

of FFP is probably not justified and was based on the experience we had in

the past with larger priming volumes. For this reason the use of FFP will be

limited in the future to those patients who have a pathologic

thromboelastography or disturbed coagulation tests. Due to the low volume of

blood products used a high number of infants can be operated on with

exposure to only one blood product donor.

No major differences in management with a conventional system were

observed, except for the very small residual volume in the circuit. The

Microsafe however, will be possibly restricted for infants under 5 kilogram due

to its small venous reservoir of 400 ml.

In our study there was no need for modified ultrafiltration, since, due to the

small priming volume, it was much easier to control fluid shifts and hence

extracellular interstitial fluid accumulation, possibly related to it. Due the small

amount of PC used in the priming volume the use of calcium to counteract the

effects of sodium citrate could be avoided, which might reduce reperfusion

injury (12).

Appendix 5

169

This relatively new concept of small neonatal oxygenators used with

appropriately sized tubings, may solve some of the problems caused by

haemodilution in infants below 8 kg. Future studies should examine the

influence of these systems on coagulation and complement activation.

Appendix 5

170

References

1. Elliott M, Rao PV, Hampton M. Current paediatric perfusion practice in the

UK. Perfusion 1993; 8: 7-25

2. Cooper MM, Elliott M. Haemodilution. In: Jonas RA, Elliott M eds.

Cardiopulmonary bypass in neonates, infants and young infants. Oxford:

Butterworth-Heinemann Ltd, 1994: 82-100

3. Tyndall Jr. CM, Berryessa RG, Campbell DN, Clarke DR. Micro-prime

circuit facilitating minimal blood use during infant perfusion. J. Extra-corpor.

Technology 1987; 19: 352-357.

4. Naik SK, Elliott MJ. Ultrafiltration and paediatric cardiopulmonary bypass.

Perfusion 1993; 8: 101-112

5. De Somer F, François K, Foubert L et al. D-901 neonatal oxygenator: a

new perspective. Perfusion 1994; 9: 349-355.

6. Swan H. Acid-base management during hypothermic circulatory arrest for

cardiac surgery. In:Rahn H, Prakash O. eds. Acid-base Regulation and

Body Temperature. Boston: Martinus Nijhoff , 1985: 81-107

7. Nicolas F, Daniel J-P, Bruniaux J et al. Conventional cardiopulmonary

bypass in neonates. A physiological approach - 10 years of experience at

Marie-Lannelongue Hospital. Perfusion 1994, 9: 41-49

8. Hill AG, Groom RC, Akl BF, Lefrak EA, Kurusz M. Current paediatric

perfusion practice in North America. Perfusion 1993; 8: 27-38

9. Groom RC, Hill AG, Kurusz M, Munoz R et al. Paediatric perfusion practice

in North America: an update. Perfusion 1995; 10: 393-401

Appendix 5

171

10. Bonser RS, Vergani D. The role of the complement system during

cardiopulmonary bypass. In: Kay HK, editor. Techniques in Extracorporeal

Circulation. Third Edition. Butterworth-Heinemann, 1992: 156-177

11. Gu YJ, Boonstra PW, Akkerman C, Mungroop H, Tigchelaar I, Van

Oeveren W. Blood compatibility of two different types of membrane

oxygenator during cardiopulmonary bypass in infants. Int.J.Artif.Organs.

1994 Oct; 17: 543-548

12. Vinten-Johansen J, Hammon J. Myocardial protection during cardiac

surgery. In: Gravlee GP, Davis RF, Utley JR, eds. Cardiopulmonary Bypass

Principles and Practice. Baltimore: Williams and Wilkins, 1993: 172-173

Appendix 5

172

Figure 1 Dideco D-901 Lilliput.

Appendix 5

173

Figure 2 Polystan Microsafe

Appendix 5

174

Table 1: Patient data.

N = 80 Mean ± SD

Age (days) 148 ± 174

Gender (female/male) 36 F / 44 M

Weight (kg) 4.6 ± 1.6

BSA (m²) 0.26 ± 0.07

CPB time (minutes) 107 ± 44

Aortic cross clamp time

(minutes)

52 ± 23

Table 2: Oxygenator characteristics.

Dideco Lilliput Polystan Microsafe

Maximum blood flow (ml/min) 800 800

Priming volume in membrane and heat

exchanger (ml)

60 52

Minimum volume in venous reservoir (ml) 20 25

Connector size (inch) 3/16 and 1/4 3/16 and 1/4

Appendix 6

175

Hydrodynamic characteristics of artificial lungs

Peter W. Dierickx, Filip De Somer, Dirk S. De Wachter, Guido Van Nooten,

and Pascal R. Verdonck

ASAIO, 2000; 46(5): 532-535

Appendix 6

176

Abstract

An artificial lung is used during cardiopulmonary bypass to oxygenate blood

and to control blood temperature. The pressure drop-flow rate characteristics

of the membrane compartment in three hollow fiber membrane oxygenators

were determined in vitro to characterize design features. Results are

presented in a unique dimensionless relationship between Euler number, NEu

(ratio of pressure drop to kinetic energy) and Reynolds number, NRe (ratio of

inertial to viscous forces) and are a function of the device porosity, ε, and a

characteristic device length, ξ, defined as the ratio of the mean blood path

and manifold length: ( )ε1εNβα

εξN

Re2Eu −⋅⋅

+=⋅ .

This dimensionless approach allows us (1) to compare oxygenators

independently, and (2) to relate water tests to blood.

Appendix 6

177

Introduction

An artificial lung is used during cardiopulmonary bypass to oxygenate blood

and to control the blood temperature. The blood-material interaction in the

artificial lung induces a complex systemic inflammatory reaction. To control

this reaction, more biocompatible surfaces, in combination with blood outside

the fiber geometry’s (less surface for the same mass transfer) were

introduced. In general, a membrane oxygenator is placed between the pump

and the patient to overcome the resistance exerted by the device. Resistance

can be monitored by measuring blood flow in combination with inlet and outlet

pressure and is related to the geometry of the way the fluid flows (flow

pattern) through the membrane oxygenator. Few attempts have been made to

characterize hydrodynamics, including geometry and flow pattern, of an

artificial lung.1-3 Vaslef et al. proposed a dimensionless flow-friction

relationship that only incorporated the viscous losses, eliminating nonlinear

effects in the pressure-flow rate relationship. Our study investigated the value

of a unique relationship between pressure drop in the membrane

compartment and flow rate as a function of geometry and flow pattern,

incorporating nonlinear effects. Such a relationship may facilitate the design of

new devices.

Appendix 6

178

Materials and Methods

Dimensional Analysis

An artificial lung can be characterized by different geometrical parameters:

membrane surface area, A; diameter of the fiber, d; length of the fiber

compartment, L; inside housing outer diameter, Di; outside housing inner

diameter, Do and gross frontal area of the blood path, Af. The “void fraction” or

device porosity, ε, is defined as the ratio of the volume of voids (volume in the

membrane compartment occupied by blood) to the volume of the bed (total

volume of the membrane compartment). A characteristic length for flow

through porous beds or packed fiber bundles, is hydraulic radius, Rh.

Hydraulic radius is expressed in terms of device porosity, ε, and wetted

surface, a, per unit volume of bed 4,5:

abed of volume

surface wettedbed of volume

voidsof volume

R hε

=

= (1)

Manifold length, Lm, is defined as the length by which the total oxygenator flow

is divided per unit width of fiber stack. Mean blood path length, Lb, is the

average distance blood has to travel through the fiber stack. Consequently, a

dimensionless characteristic device length, ξ, can be defined as the ratio of

mean blood path length, Lb, and average manifold length, Lm. We assume that

blood is uniformly distributed over the fiber stack. In Sarns Turbo 440 (3M,

Michigan) and Optima (Cobe, Arvada) membrane oxygenators, Lm and Lb are

determined as shown in Figure 1 (left panel). Blood enters the membrane

Appendix 6

179

evenly distributed over the length, L, of the manifold, and blood flow is split in

two for left and right sides. Blood flow rate per unit membrane width Q´ is:

LQ

LQQ

m ⋅==′

2 (2)

Mean blood path length is :

+

⋅π

=2

DD2

L oib (3)

In the Dideco D703 (Dideco, Miranda, Italy), blood flow enters the fiber stack

circumferentially (Figure 1 right panel). Manifold length is:

+

⋅π=2

DDL oim (4)

whereas mean blood path length is determined as :

LLb = (5)

The measured pressure drop, ΔP, and flow rate, Q, relationship is presented

as a polynomial of second order:

QbQaP 2 ⋅′+⋅′=∆ (6)

Reynolds number, NRe, is defined as the ratio of inertial and viscous forces.

Characteristic length in the Reynolds number for flow through a fiber stack, is

hydraulic diameter, Rh. Consequently, Reynolds number, NRe, is defined as

µρ⋅⋅

⋅ε⋅

= h

fRe

R4A

QN (7)

in which Af represents the gross frontal area, ρ represents the density, and μ

the dynamic viscosity of the fluid. Euler number, NEu, is defined as the ratio of

pressure drop and kinetic energy:

Appendix 6

180

2

f

Eu

AQ

PN

ε⋅

⋅ρ

∆=

(8)

If eq. (6) is divided by the denominator of eq. (8), one can describe Euler

number, NEu, as a dimensionless function of the reciprocal of Reynolds

number, NRe, under the assumption of laminar flow conditions in the fiber

stack.

Re

Eu NβαN ′

+′=(9)

An analogous approach has already been successfully applied to the flow

characteristics of aortic canulae.6 A similar, but not identical approach is

described by Ergun,5 Bird et al.,4 and by Macdonald et al.7 for flow through

porous media. Our starting point for a dimensionless relationship is a pressure

drop-flow rate relationship, whereas Ergun5 related the pressure gradient to

fluid velocity. α’ and β’ in Eq. 9 are model parameters that characterize the

porous medium and, therefore, must be functions of the medium5,7 rather than

universal constants. It is assumed that the medium can be characterized by

the device porosity, ε, and the dimensionless characteristic device length, ξ,

and that the functional form of α’ and β’, in analogy with Ergun5 and

Macdonald et al.,7 can be represented as a power function of device porosity

and dimensionless characteristic device length:

( ) ( )( ) ( ) 'pε1ξεβξε,β

ε1ξεαξε,αmn

pmn

−⋅⋅⋅=′

−⋅⋅⋅=′′′

(10)

in which α, β, n, m, p, n´, m´, and p’ are constants that may be determined by

nonlinear regression analysis. The relationship between Euler and Reynolds,

Appendix 6

181

therefore, is determined solely by geometry and flow pattern in the artificial

lung.

Materials

Pressure drop-flow rate characteristics of the membrane compartment of

three different hollow fiber membrane oxygenators (Sarns Turbo 440 (n=2),

Optima (n=3), Dideco D703 (n=3)) were measured in vitro and analyzed to

characterize design features.

Experimental Setup

Experiments were performed with water, and steady flow is applied using an

upstream reservoir with a constant head. Pressure was measured between

the heat exchanger and membrane compartment, and at the inlet and outlet of

the artificial lung, using fluid-filled pressure transducers (Ohmeda, Gent,

Belgium). Flow rate was measured with an ultrasonic transit time flow meter

(Transonic, Ithaca, NY). Downstream of the artificial lung, static pressure was

kept constant at 150 mmHg.

Statistical Analysis

Fitting of the parabolic pressure drop and flow rate relationship is performed

using the non-linear regression Marquardt-Levenberg algorithm (Sigmastat

2.0, Jandel Scientific, Germany). The same technique is used to fit

geometrical parameters within the Euler-Reynolds relationship. Results are

presented with upper and lower confidence limits (95%), asymptotic standard

errors of fit parameters and coefficient of determination, R², for non-linear

regression.

Appendix 6

182

Results

Figure 2 presents pressure drop and flow rate characteristics for the

membrane compartments of three different artificial lungs, indicating a

parabolic relationship between ∆P and Q (Eq. 6). The technique to measure

the pressure between the heat exchanger and the membrane compartment

does not influence the pressure drop-flow relationship. Hence, the variance in

pressure drop–flow rate relationship in Figure 2 for the three Dideco D703

artificial lungs may be attributed to difference in construction of the membrane

compartment. The corresponding Euler-Reynolds relationship for the

membrane compartment (Eq. 9) is depicted in Figure 3. The geometric data

and model parameters α’ and β’ are tabulated in Table 1, indicating a similar

trend among the different artificial lungs in NEu-NRe for device porosity ε, and

dimensionless characteristic device length ξ: α’ and β’ increase with ε and

decrease with ξ. Based on this finding, the functions of Eq. (10) are

determined and yield the following dimensionless relationship:

( )ε1εNβα

εξN

Re2Eu −⋅⋅

+=⋅ (11)

The results of the non-linear regression are listed in Table 2 and depicted in

Figure 4. Figure 5 shows that for each oxygenator the original ∆P versus Q

data, along with the predicted ∆P versus Q relationship (with 95% regression

intervals for α and β) obtained by converting the dimensionless fit from figure

4 back to dimensional form.

Appendix 6

183

Discussion

The importance of the effect of a given resistance and flow characteristic in an

artificial lung on blood elements and the degree of inflammatory response

have not yet been established. However, it is well known however that shear

stress plays an important role in the activation of blood platelets and white

blood cells. However, from an engineering point of view, a certain pressure

drop over the device is necessary for an even distribution of blood flow. In the

past, few attempts have been made to characterize hydrodynamics, including

the geometry and flow pattern of an artificial lung. Pressure drop across a

membrane compartment can be studied using a dimensionless relationship

between Euler and Reynolds number as a function of two dimensionless

characteristic geometrical parameters, namely, device porosity ε and the

dimensionless characteristic device length ξ. This relationship indicates that

the (total) pressure across heat exchanger and membrane compartment is

directly related to (1) length of the blood path, (2) length of the manifold and

(3) flow pattern. Figure 4 can be presented as a device specific scaling of

Figure 3, resulting in one curve representing the three artificial lungs.

In Figure 5, the dimensionless Eu-Re equation gives a good prediction of the

∆P-Q relationship for water flow rates up to 3.5 lpm (Re<10). However, at

higher water flow rates (Re>10), the predicted ∆P-Q data deviate from the

measured ∆P-Q data, especially for the Optima.

The dimensionless approach is independent of fluid density and viscosity and

enables one to relate water tests to blood, advertising the benefits of

dimensionless numbers. There is no need to rescale the graphs for blood,

Appendix 6

184

although the measurements are performed with water. This is of great

advantage when hemodilution and hypothermia are present, because they

alter the dynamic viscosity and, therefore, the pressure drop-flow relationship.

However, with the dimensionless numbers NRe and NEu, the graph is

normalized for a Newtonian fluid of any viscosity. Assuming a blood

temperature of 28°C and a hematocrit of 30% during cardiopulmonary bypass,

density and dynamic viscosity of blood can be calculated 3 ρ = 1.037 kg/m³

and µ = 0.0015 Pa.s, yielding operational ranges for NRe between 0 and 4.

Blood pressure drop can then be calculated using the corresponding NEu

number, dimensionless geometric parameters and kinetic energy.

With the help of the proposed dimensionless format, one can (1) compare

oxygenators independently, (2) relate water tests to blood tests, and (3)

predict pressure drop of a new design in an artificial lung. We believe that this

dimensionless analysis can be an excellent tool for the study of better

designs.

We demonstrate that pressure drop across a membrane compartment can be

studied by using a dimensionless relationship between Euler and Reynolds

number as a function of two dimensionless characteristic geometric

parameters, namely device porosity ε and a newly defined dimensionless

characteristic device length ξ.

Acknowledgement

This research is funded by a grant of the Flemish Institute for the Promotion of

the Scientific-Technological Research in Industry (IWT961181). The authors

Appendix 6

185

like to express their gratitude to Nico Vincart for doing most of the

measurements.

Appendix 6

186

References

1. S.N. Vaslef, L.F. Mockros, R.W. Anderson, R. Leonard: Use of a

mathematical model to predict oxygen transfer rates in hollow fiber

membrane oxygenators. ASAIO Trans 40: 990-996, 1994

2. S.N. Vaslef, L.F. Mockros, K.E. Cook, R. Leonard, J. Sung, R.W.

Anderson: Computer-assisted design of an implantable, intrathoracic

artificial lung. Artif Organs 18: 813-817, 1994

3. L.F. Mockros and R. Leonard : Compact Cross-Flow Tubular Oxygenators.

ASAIO Trans 31:628-633, 1985.

4. R.B. Bird, W.E. Stewart, E.N. Lightfoot : Transport Phenomena. New York,

John Wiley & Sons, 1960.

5. S. Ergun: Fluid flow through packed columns. Chem Eng Prog 48: 89-94,

1952

6. P.R. Verdonck, U. Siller, D.S. De Wachter, F. De Somer, G. Van Nooten :

Hydrodynamical comparison of aortic arch cannulae. Int. J. Artif. Organs

21(11): 705-13, 1998.

7. I.F. Macdonald, M.S. El-Sayed, K. Mow, F.A. Dullien: Flow through porous

media: the Ergun equation revisited. Industrial Engineering Chem

Fundament 18: 199-208, 1979.

Appendix 6

187

Figure 1: Geometrical characteristics of artificial lungs. Left panel: Sarns

Turbo 440 and Optima; right panel: Dideco D703.

Do Di

L

LbLm

Q

Do

L LmLb

Q

Di

3M Sarns Turbo 440Cobe Optima

Dideco D703

Af

Appendix 6

188

Figure 2: Measured membrane compartment pressure drop (∆P) flow rate Q

relationship with confidence limits.

Q [lpm]

0 1 2 3 4 5

∆P [m

mH

g]

0

20

40

60 Sarns Turbo 440.Cobe Optima.Dideco D703.∆

Appendix 6

189

Figure 3: Euler–Reynolds relationship for the Sarns Turbo 440, Optima and

Dideco D703 on a bilogaritmic plot.

NRe

1 10

NEu

103

104

105

Sarns Turbo 440.Cobe Optima.Dideco D703.

Appendix 6

190

Figure 4: Dimensionless pressure drop-flow rate relationship as a function of

device porosity ε and dimensionless characteristic device length ξ on a

bilogaritmic plot.

NRe . ε . (1−ε)

0,1 1

NEu

. ξ

/ ε2

104

105

NEu-NRe as a function of ε and ξ.Sarns Turbo 440.Cobe Optima.Dideco D703.

Appendix 6

191

Figure 5. Original pressure drop-flow rate data along with the predicted

pressure drop-flow rate data obtained by converting the dimensionless fit back

to dimensional form. Left panel: Sarns Turbo 440; middle panel: Optima; Right

panel: Dideco D703.

0 1 2 3 4

)p

[mm

Hg]

0

10

20

30

40

50Sarns Turbo 440

Q [lpm]0 1 2 3 4

0

10

20

30

40

50

Cobe Optima

0 1 2 3 40

10

20

30

40

50

Dideco D703

Appendix 6

192

Table 1: Geometrical characteristics of artificial lungs, including the model

parameters α’ and β’.

Appendix 6

193

Table 2: Non-linear regression results.

α ± SD 2365 ± 52

β ± SD 8509 ± 25

R² 0.998

Appendix 6

194

Appendix 7

195

Impact of oxygenator design on hemolysis, shear stress,white blood cell and platelet count

De Somer F, Foubert L, Vanackere M, Dujardin D, Delanghe J.,Van Nooten G

J. Cardiothor.Vasc. Anesth. 1996; 10: 884-889

Appendix 7

196

Abstract

Objective: To determine whether relative pressure drop, shear stress,

hemolysis, white blood cell and platelet count are influenced by different

oxygenator designs. To compare the oxygenator results with the average

shear stress over an arterial cannula.

Design: Prospective; patients enrolled consecutively.

Setting: University Hospital.

Participants: 3 times 12 adult patients, scheduled for routine cardiac surgery.

Interventions: Each group was submitted to a different oxygenator design,

group 1 to a high pressure hollow fibre membrane oxygenator (Sarns Turbo),

group 2 to a medium pressure hollow fibre membrane oxygenator (Cobe

Optima) and group 3 to a flat sheet membrane oxygenator (Cobe Duo).

Measurements and Main results: Although the investigated oxygenators have

important differences in pressure drop and shear stress no statistical

differences were found in hemolysis generation or blood handling between the

different groups. Actually the study shows much higher shear stress levels

over an average arterial cannula than over any of the evaluated oxygenators.

Conclusions: The pressure drop over an oxygenator does not correlate well

with shear stress and hemolysis because the dimensions of the system

(radius and length) must be included in the calculation of shear stress from

pressure drop.

Appendix 7

197

Introduction

During the last years low prime, hollow fibre oxygenators have become first

choice in most cardiac centres. The main rationale is the reduction in total

system prime and blood foreign material interface. However, the effects of

pressure drop over an oxygenator on blood trauma still appear to be a subject

of debate (1,2). According to the literature (1,3,4) it is not the pressure drop as

such that causes hemolysis and cellular activation, but shear stress. Red

blood cells are less sensitive than platelets and white cells to shear, with a

critical shear stress level of 2000 - 3000 dynes/cm², below which hemolysis is

limited (3,4). Platelets and white cells are activated at significantly lower shear

stress levels of 100 and 75 dynes/cm², respectively (5-7)

The aim of this study is to evaluate the relative pressure drop, shear stress,

and ex vivo blood handling characteristics for three different oxygenator

designs.

Appendix 7

198

Materials and methods

The study consisted of three groups of 12 patients (Table 1). In group 1

(Sarns Turbo, 3M, Ann Arbor, MI) a hollow fibre oxygenator with a high

pressure drop, in group 2 (Cobe Optima, Cobe Cardiovascular, Arvada, CO).

a hollow fibre oxygenator with a moderate pressure drop and in group 3 a low

prime flat sheet oxygenator design (Cobe Duo, Cobe Cardiovascular, Arvada,

CO) were used. The Cobe Duo consists of two oxygenators in one device.

This is accomplished by two flat sheet membrane compartments, each 1.3 m2,

placed in parallel. The heatexchanger is the same whether one or two

compartments. Depending on the oxygen consumption of the patient one can

start on one or two compartments are used. If one starts on the first

compartment the second still can be opened when needed during the

extracorporeal circulation. Only one patient in group three had an oxygen

consumption which exceeded the oxygen transfer capacity of one

compartment. This patient was perfused with both oxygenator compartments,

which resulted in a total system prime of 1500 ml. There were no statistical

differences between demografic data of the patients in the three groups.

The specific characteristics of the oxygenators are presented in table 2.

The extracorporeal system comprised in all groups the Cobe heart-lung

machine (Cobe Cardiovascular, Arvada, CO), custom tubing packs made of

polyvinyl chloride (PVC) tubing with exception of the arterial pumphead

(silicone), arterial line filtration (40 micron), cardiotomy reservoir with a 20

micron filter and a collapsible venous reservoir. The arterial line had an

internal diameter of 3/8 inch and was 175 cm long, the venous line was 1/2

Appendix 7

199

inch and had a length of 190 cm. All suction lines were 1/4 inch and had an all

over length of 360 cm (from top to cardiotomy reservoir). Occlusion setting of

the arterial rollerpump was completely occlusive at a back pressure of 330

mmHg. The occlusion setting was verified and eventually adjusted before

every use. The prime solution was a mixture of Plasma-Lyte A, human

albumin 20% and mannitol. The complete system had a priming volume of

1300 ml in every group.

In group 1 one patient, in group 2 five patients and in group 3 two patients

received homologous blood in order to obtain a post bypass hematocrit of

25%.

Each oxygenator configuration was used on 12 consecutive patients

undergoing routine cardiac surgery.

Perfusion technique

All oxygenators were inspected and set up in accordance with the

manufacturer’s enclosed instructions. Blood flow rates were maintained to

ensure adequate tissue perfusion. The arterial pump was a standard

rollerpump (Cobe Cardiovascular, Arvada, CO). An alpha-stat regimen was

used in all cases for acid-base and blood gas management. The detailed

perfusion protocol was published elsewhere (8).

Since suction is an important determinant of hemolysis (9), it was controlled

within strict limits. Suction was only applied when necessary and the

revolutions of the roller pumps were kept as low as possible. The aspiration

on the aortic needle was pressure controlled and automatically stopped at a

negative pressure of minus 100 mmHg.

Appendix 7

200

Activated clotting times were kept above 400 seconds. Cardioplegia was

instituted by using St. Thomas cardioplegia two solution (400-1000 ml).

Data collection

White blood cell count, red blood cell count, hematocrit, hemoglobin and

platelets were processed using STKS-Coulter counter apparatus. Serum

concentrations of free hemoglobin (free Hb), haptoglobin (Hp) and hemopexin

(Hpx) were determined as markers of hemolysis using immunonephelometry

(10) on a BN nephelometer (Behringwerke, Marburg) and expressed

according to IFCC standards (11). Blood samples were taken prior to

institution of bypass, after mixing (five minutes bypass), every 20 minutes

during bypass, five minutes after bypass and 30 minutes after the

administration of protamine.

For correction of the obtained results for hemodilution (due to the priming fluid

and the cardioplegia) a neutral plasma protein IgG was monitored. Correction

was done using the following formula:

corrected concentration = measured concentration x initial IgG

concentration/IgG concentration at time of measurement

Values of platelets (PLT) were expressed as percent of baseline according to:

% of baseline = (prebypass IgG/IgG at time of measurement) x (measured

PLT count/prebypass PLT count) x 100

Shear stress calculations were made for both the cannula and each

oxygenator group at standard blood conditions of 37° Celsius and 35%

hematocrit. These blood characteristics were chosen because the data from

the cannula manufacturer for the determination of pressure drop were

Appendix 7

201

obtained under these conditions (14). For comparative reasons the same

blood conditions were used for the oxygenators. (Blood viscosity at 35 percent

hematocrit and 37° Celsius is 2.65 kg/m*sec; blood viscosity at 25 percent

hematocrit and 28° Celsius is 2.50 kg/m*sec). For the cannula the calculation

of wall or maximum shear stress (12,13) was done by the following formula:

τ =((∆P)(r))

2L

where: ∆P = pressure drop, L = length (cm), r = radius (cm)

An effective radius was calculated by superposition of the manufacturer’s

pressure versus flow data on a model for turbulent flow in a smooth tube (15).

Flow through an oxygenator can be considered as flow through a porous

medium. According to Bird (13) the shear stress in each oxygenator was

calculated by considering the flow equivalent to the flow in a packed column

governed by:

τ =((Rh)(∆P))

(L)

where: ∆P = pressure drop in mmHg, L= blood path length in cm, Rh =

hydraulic radius

Rh =Q(25 / 6)µL(∆P)ε(Ae)

where: ε = porosity of membrane area that fills that cross section

Q = volumetric pump flow

µ = fluid viscosity

Appendix 7

202

Ae = is empty housing cross sectional area for flow in cm²

25/6 = experimental derived factor.

The hydraulic radius for each group was calculated from superposition of

pressure drop data on the packed column flow model (13).

Statistics

Statistical analysis was performed using analysis of variance for repeated

measurements. Statistical analysis of bypass data was performed up to 60

minutes of bypass. There were insufficient data available in all groups for

further analysis beyond 60 minutes. In order to test the control value against

all other values the Dunnett test was used. All values are expressed as mean

± SD where appropriate.

Appendix 7

203

Results

Hematology

Platelet depletion is shown in table 3. The prebypass level of 100% dropped

to 97% for group1 versus 86% for group2 and 95% for group 3 at the end of

the operation. (p=0.119)

The evolution in white blood cell count is demonstrated in table 3. After an

initial decline in white blood cell count it increased to 15.73 ± 4.09 x1000/mm3

in group 1, 13.06 ± 5.27 x1000/mm3 in group 2 and 15.99 ± 5.23 x1000/mm3

in group 3. (p=0.119)

Hemolysis markers (Table 4)

The free plasma hemoglobin (Figure 1) increased in group 1 from 26.91 ±

7.77 mg/100ml to 47.82 ± 19.65 mg/100ml versus 14.14 ± 5.11 mg/100ml to

26.49 ± 18.72 mg/100ml in group 2 and 11.52 ± 5.70 mg/100ml to 25.29 ±

9.90 mg/100ml in group 3 at 60 minutes of bypass. There is a statistical

difference between group 1 and 2 (p=0.01) but not within both groups

(p=0.126). Between group 1 and 3 there is a statistical difference between

groups (p<0.001) and within both groups (p=0.012).

Haptoglobin levels (Figure 2) decreased from 1.19 ± 0.53 g/l to 0.85 ± 0.45 g/l

in group 1 versus 0.92 ± 0.54 g/l to 0.63 ± 0.24 g/l in group 2 and 1.64 ± 0.73

g/l to 1.29 ± 0.69 g/l in group 3 at 60 minutes of bypass. At the end of the

operation they were 1.12 ± 0.58 g/l, 0.77 ± 0.66 g/l and 1.16 ± 0.69 g/l (NS),

respectively.

Hemopexin levels decreased from 0.78 ± 0.17 g/l to 0.64 ± 0.15 g/l in group 1

versus 0.84 ± 0.13 g/l to 0.79 ± 0.19 g/l in group 2 and 0.84 ± 0.11 g/l to 0.71

Appendix 7

204

± 0.09 g/l in group 3 at 60 minutes of bypass. At the end of the operation

hemopexin content was 0.65 ± 0.16 g/l in group 1, 0.74 ± 0.15 g/l in group 2

and 0.74 ± 0.08 g/l in group 3. There was no statistical difference between the

groups (p=0.10).

Shear stress (Table 5)

The calculated shear stress in group 1 was 40 dynes/cm² at 2 LPM, 84

dynes/cm² at 4 LPM and 126 dynes/cm² at 6 LPM. In group 2 the values were

23 dynes/cm² at 2 LPM, 52 dynes/cm² at 4 LPM and 88 dynes/cm² at 6 LPM. In

group 3 with only one compartment in use the shear stress was 38 dynes/cm²

at 2 LPM and 80 dynes/cm² at 4 LPM. With two compartments following

values were obtained: 25 dynes/cm² at 2LPM, 52 dynes/cm² at 4 LPM and 82

dynes/cm² at 6 LPM.

For a 24 french Bard “Opticlear” straight arterial cannula (Bard

Cardiopulmonary, Haverhill, MA) with a length of 24 cm the pressure drop is

50 mmHg at 4 LPM and 100 mmHg at 6 LPM (13). This results in a calculated

shear stress of 375 dynes/cm² at 4 LPM and 749 dynes/cm² at 6 LPM.

Appendix 7

205

Discussion

In general, the pressure drop over an oxygenator does not correlate well with

shear stress and hemolysis, because the dimensions of the system (radius

and length) must be included in the calculation of shear stress from pressure

drop. An oxygenator with a high pressure drop over a long blood path length

may have a smaller shear stress than an oxygenator with a low pressure drop

over a short length. For example, the Cobe Duo with only one compartment in

use has a much higher pressure drop than the Sarns Turbo (235 mmHg vs.

183 mmHg at 4 LPM), although the shear stress of the Duo is lower than that

of the Sarns Turbo (80 dynes/cm² vs. 84 dynes/cm²). The Cobe Optima with

the lowest pressure drop, never more than 62% of that of the Cobe Duo with

two compartments, has low shear stress levels, however the latter are

comparable with those of the Cobe Duo (88 dynes/cm² vs. 82 dynes/cm²)

The calculated shear stress levels for the cannula and oxygenator presented

here demonstrate that the average wall shear stress levels for the cannula are

greater than those for oxygenators.

The duration the shear stress is applied to the blood is an important

consideration in the relative comparison of shear stress within a circuit. Blood

flow in the arterial cannula, which would be subjected to a higher shear stress

than in the membrane oxygenators, would be subjected to that cannula stress

for a shorter period of time due to a higher velocity. Both the level of the shear

stress, and the exposure time, have been related to the extend of cellular

activation (5-7,16).

Appendix 7

206

The calculated average shear stress values in this study for the oxygenators

and the cannula are all well below the critical value of 2000 - 3000 dynes/cm²

(3,4) for hemolysis. Therefore the lack of statistically significant differences in

the markers of hemolysis between the groups is not surprising. However, the

shear values calculated herein do exceed those reported for platelet and

white cell activation (5-7), yet there were no statistically significant differences

in the depletion of these components between the groups. This might be

explained by the small differences in the oxygenator shear levels relative to

the cannula levels, or by the fact that platelet and white cell depletion is not an

accurate measure of activation (17). The good preservation of platelets in our

study might be explained by the small priming volume and foreign surface

area (18). The values above 100% might find their origin in the fragmentation

of large platelets in smaller pieces (18).

Our present study does not show any correlation between the different

oxygenator designs as perhaps expected. In all groups the accumulation of

free hemoglobin was counteracted by rapid elimination of the Hb/Hp

complexes by specific hepatic receptors (19). In all patients the residual

capacity of serum Hp to protect against hemolysis was satisfactory. Free

hemoglobin is not always a correct predictor of the degree of hemolysis (20)

since this parameter can be influenced by various pre-analytical factors (e.g.

the suction applied on the syringe, the sampling site, etc.). However pressure

drop as a design parameter, although important for centrifugal pump users, is

not the only single element that may influence hemocompatibility.

As a matter of fact two key factors must be taken into account

Appendix 7

207

1. The total pressure drop of the bypass system (oxygenator, filter,

connectors, cannulae etc...) and not only the oxygenator itself.

2. The instantaneous shear forces or stresses within the entire bypass circuit

as a function of blood flow rate. One should be very carefull not to draw any

hasty decision based on normal average shear stress calculations as they do

not indicate the instantaneous shear stresses that can sometimes exceed

what the red blood cells, the platelets and the leukocytes can withstand before

they are damaged. And again the full circuit has to be analyzed (21,22) and

not only the oxygenator.

On the other hand measuring, monitoring or calculating instantaneous shear

forces during cardiopulmonary bypass is a rather difficult task. Because of the

practical difficulty any total bypass design has to be assessed against

hemocompatibility to assure the best preservation as possible of our patient

formed blood elements. Although the challenge for us remains to agree on

standard markers to best characterize and clinically objectivate

hemocompatibility.

Acknowledgement.

The authors wish to express their gratitude to Mr. Ben F. Brian for his valuable

help in the preparation of this manuscript.

Appendix 7

208

References

1. Bearss MG, The Relationship Between Membrane Oxygenator Blood Path

Pressure Drop and Hemolysis: An In-vitro Evaluation. The Journal of

Extra-Corporeal Technology 25: 87-92, 1993

2. Personnal communication Ned Evans, product specialist 3M

3. Nevaril CG, Lynch EC, Alfrey CP, Hellums JD, Erythrocyte damage and

destruction induced by shearing stress. J.Lab. & Clin. Med. 71: 781-790,

1968

4. Blackshear PL, Dorman FD, Steinbach EJ, et al: Shear, Wall Interaction

and Hemolysis. Trans. Amer. Soc. Artif. Int. Organs 12: 113-120, 1966

5. Hellums JD, Biorheology in Thrombosis Research. Annals of Biomedical

Engineering. 22: 445-455, 1994.

6. Hellums JD, Hardwick RA.: Response of Platelets to Shear Stress - a

Review. In Gross DR, Hwang NHC eds. The Rheology of Blood Vessels

and Associated Tissues. Alphen aan den Rijn: NATO Advanced Study

Institute Series - E, No 41, Sijthoff & Noordhoff, 1981

7. McIntire LV, Martin RR. Mechanical Trauma Induced PMN Leucocyte

Dysfunction. In Gross DR, Hwang NHC eds. The Rheology of Blood

Vessels and Associated Tissues. Alphen aan den Rijn: NATO Advanced

Study Institute Series - E, No 41, Sijthoff & Noordhoff, 1981

8. De Somer F, De Smet D, Vanackere M, et al: Clinical evaluation of a new

hollow fibre membrane oxygenator. Perfusion. 9: 57-64, 1994

Appendix 7

209

9. de Jong JCF, ten Duis HJ, Smit Sibinga C. Th, Wildevuur Ch. R. H.

Hematologic aspects of cardiotomy suction in cardiac operations. J.

Thorac. Cardiovasc. Surg. 79: 227-236, 1980

10. Fink et al. Measurement of proteins with the Behring Nephelometer. J.

Clin. Chem. Clin. Biochem. 27: 261-276, 1989

11. Johnson AMA New international reference for proteins in human serum.

Arch. Pathol. Lab. Med. 117: 29-31, 1993

12. Data provided by Bard Cardiopulmonary, PRMA#93-006 Rep 6/94 1.5M.

13. Giles RV, Fluid Mechanics and Hydraulics 2nd edition, New York:

McGraw-Hill Book Company, 1977, p 101.

14. Bird RB, Stewart WE, Lightfoot EN. Transport Phenomena, New York:

John Wiley and Sons, 1960: p 197.

15. Montoya JP, Merz SI, Bartlett RH. A Standardized System for Describing

Flow/Pressure Relationships in Vascular Access Devices. Trans ASAIO

37: 4-8, 1991.

16. Lambert J. In: Schmid-Schönbein H, Teitel P eds. Basic aspects of blood

trauma The Hague: Martinus Nijhoff Publishers, 300-311, 1979

17. O’Brien JR, Etherington MD, Rebleeding, the reversal of shear activation

of platelets - a possible clue to thrombogenesis. Thromb. Res. 65: 821-

822, 1992;.

18. In: Casthelhy PA, Bregman D. eds. Cardiopulmonary Bypass: Physiology,

Related complications and Pharmacology. New York: Futura Publishing

Company, 71, 196, 1991

19. Kino K et al. Hemoglobin - Haptoglobin receptor in rat liver plasma

membrane. J. Biol. Chem. 255: 9616-9620, 1980.

Appendix 7

210

20. Lammers M, Gressner AM. Immunonephelometric quantification of free

haemoglobin. J. Clin Chem Clin Biochem 25: 363-367, 1987.

21. Craddock PR, Hammerschmidt D, White JG, et al: Complement (C5a) -

induced granulocyte aggregation in vitro. A possible mechanism of

complement mediated leukostasis and leukopenia. J Clin Invest 60: 260-

64, 1977

22. Hammerschmidt DE, Stroncek DF, Bowers TK et al. Complement

activation and neutropenia occuring during cardiopulmonary bypass. J

Thorac Cardiovasc Surg 81: 370-77, 1981

Appendix 7

211

Table 1. Patient demographics.

Mean Group1

Sarns Turbo (n=12)

Group2

Cobe Optima (n=12)

Group3

Cobe Duo (n=12)

Age (years) 64±8 63±9 66±10

Sex 11 M / 1 F 9 M / 3 F 9 M / 3 F

Weight (kg) 77±12 74±10 76±16

Length (cm) 169±7 172±10 169±6

BSA (m²) 1.87±0.15 1.87±0.15 1.85±0.18

Bloodflow (LPM) 4.5±0.4 4.5±0.4 4.4±0.5

ECC time (minutes) 71±19 87±16 82±22

Aortic cross clamp

time (minutes)

38±10 47±14 43±15

CABG 12 11 11

AVR 0 0 1

ASD 0 1 0

Appendix 7

212

Table 2. Oxygenator characteristics.

Sarns Turbo Cobe Optima Cobe Duo

1 compartment

Cobe Duo

2 compartments

Geometry hollow fibre hollow fibre flat sheet flat sheet

Surface Area1, m² 1.9 1.7 1.3 2.6

Priming1, ml 270 260 260 460

Flow Range1,

LPM

1 - 7 0.5 - 8 0.5 - 5 0.5 - 8

Pressure Drop2

2 LPM (mmHg)

88 38 111 73

Pressure Drop2

4 LPM (mmHg)

183 87 235 152

Pressure Drop2

6 LPM (mmHg)

279 148 304

(5LPM)3

241

1 Manufacturer’s Information

2 In vitro measurements with bovine blood at 37° C and 35% haematocrit

3 The maximum bloodflow over one compartment is 5 LPM

Appendix 7

213

Table 3. Evolution of platelet and white blood cell count.

Appendix 7

214

Table 4. Hemolysis parameters.

Appendix 7

215

Table 5. Measured oxygenator parameters and calculated shear stress.

Sarns

Turbo

Cobe

Optima

Cobe Duo

1

compartment

Cobe Duo

2

compartments

Blood Path

Length1, cm

9.42 9.19 11 11

Hydraulic radius2, cm 0.0032 0.0041 0.0028 0.0028

Shear Stress 2LPM

(dynes/cm²)

40 23 38 25

Shear Stress 4LPM

(dynes/cm²)

84 52 80 52

Shear Stress 6LPM

(dynes/cm²)

126 88 103 (5LPM) 82

1 measured from dissected units

2 calculated from saline pressure drop superposition on packed flow model

Appendix 7

216

Figure 1. Free plasma hemoglobin.

ECC = extracorporeal circulation; ‘ = minutes.

0

10

20

30

40

50

60

70

80

90

pre-ECC 5' ECC 20' ECC 40' ECC 60' ECC 5' post ECC 30' post ECC

SMO +/- SE

Optima +/- SE

DUO +/- SE

Appendix 7

217

Figure 2. Haptoglobin levels.

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

2.00

pre-ECC 5' ECC 20' ECC 40' ECC 60' ECC 5' post ECC 30' post ECC

SMO +/- SE

Optima +/- SE

Duo +/- SE

Appendix 7

218

Appendix 8

219

Can an oxygenator design potentially contribute to airembolism in CPB? A novel method for the determination of

the air removal capabilities of neonatal membraneoxygenators

F. De Somer, P. Dierickx, D. Dujardin, P. Verdonck, G. Van Nooten

Perfusion, 1998; 13: 157-163

Appendix 8

220

Abstract

At the moment air handling of a membrane oxygenator is in general studied

by using an ultrasonic sound bubble counter. However this is not a

quantitative method and it does not give any information where air was

entrapped in the oxygenator and if it eventually was removed through the

membrane for gas exchange.

This study presents a novel technique for determination of the air handling

characteristics of a membrane oxygenator. It is aimed at defining not only the

amount of air released by the oxygenator but also the amount of air trapped

within the oxygenator and or removed through the gas exchange membrane.

Two neonatal membrane oxygenators without the use of an arterial filter are

investigated: Polystan Microsafe and Dideco Lilliput. Although the air trap

function of both oxygenators when challenged with a bolus of air was simular

the Microsafe obtained this effect mainly by capturing the air in the heat

exchanger compartment while the Lilliput did remove a large amount of air

through the membrane. The difference in trap function was most striking

during the continous infusion of air.

Immediate contact with a microporous membrane, avoidance of high

velocities within the oxygenator, pressure drop, transit time and construction

of the fibre mat all contribute to the air handling characteristics of a membrane

oxygenator.

Appendix 8

221

Introduction

With the introduction of neonatal oxygenators total system priming volumes of

180 ml are feasible 1. The use of an arterial filter in these systems is

debatable. The priming volume of a paediatric arterial filter with its bypass is

almost as large as the membrane heat exchanger compartment of a neonatal

oxygenator. Beside this an arterial screen filter is only effective in the removal

of air emboli with an open purge line. However, especially in small babies it is

not desirable to keep this purge line open since one wants to know exactly the

tissue perfusion.

By not using an arterial screen filter there is a risk for pumping gaseous

microemboli (GME) or particles towards the patient. However most of the data

supporting this risk is generated using bubble oxygenators2-4. When using

membrane oxygenators a lower incidence of GME generation is reported.

However, membrane oxygenators can vary in their air handling performance,

as has been frequently reported 4-7.

The assessment of air handling capabilities of an oxygenator includes

generally an ultrasonic bubble counter 8. However the latter is not a sufficient

quantitative method and has a low reproducibility due to the limitations of the

working principle 9-11. This matter of fact is preventing an objective

assessment that could contribute to the air handling improvement of a given

device.

This study investigates the risk for microemboli and macroemboli when using

a neonatal membrane oxygenator without an arterial screen filter. It is aimed

at defining not only the amount of air released by the oxygenator but also the

Appendix 8

222

amount of air trapped within the oxygenator and or removed through the gas

exchange membrane.

Appendix 8

223

Method

Circuit (Figure 1)

To determine the amount of air released by the oxygenator a standpipe is

used as described by Miller5. This standpipe (e) acts as a long cylindrical

bubble trap. A dilato meter (4) measures the amount of gas evacuated

through the gas exchange membrane. This meter consists of a calibrated

glass capillary with a mercury drop inside. After sealing the gas outlet port and

the overpressure relief with silicone the capillary is connected to the gas inlet

port. As gas goes from the blood to the gas compartment the mercury drop

will move due to the volume increase until equilibrium is established with the

atmospheric pressure. The following formula converts the displacement of the

mercury drop into volume:

Where: V=volume in ml, ∆L=displacement of the mercury drop in mm,

d=diameter of the glass capillary in mm

Gross filtered bovine blood (22% haematocrit, 22° Celsius) is sequestered in

a 10-liter reservoir (f). It is pumped by a calibrated roller pump (a)(Cobe

Cardiovascular Inc., Arvada, CO, USA) through a Cobe Excel flat sheet

membrane oxygenator (b), which is known for its good air removal capability.

Blood flow is then directed via 1/4 inch PVC line versus the test device (d) into

the standpipe. Blood returns back to the reservoir via a 3/8 inch PVC line.

Arterial line pressure is maintained at 150 mmHg. The blood flow during the

4000.. 2dLV π∆

=

Appendix 8

224

experiment is set at 600 ml / minute. Pressures were measured before (1) and

after (2) the membrane oxygenator by fluid filled pressure transducers

(Ohmeda - Spectramed, Gent, Belgium)

A connector and stopcock (c) are added in-line to the test device inlet prior to

the test. This gives the possibility to directly inject a bolus of air ((STP) 1, 5

and 20 ml) in order to simulate a gross air embolus, or to attach a syringe

pump and slowly inject air (20 ml at 3 ml/min) to simulate an adverse event

leading to GME. After injecting a bolus a three-minute period is allowed for

equilibration after which the pump is stopped and the data recorded. During

the slow injection the pump flow is maintained during 10 minutes before the

data are recorded.

A “Y” connector is added prior to the entrance of the standpipe, so that a

recirculation line back to the aliquot can be used during priming and

debubbling of the test device. The test device is carbon dioxide flushed and

primed according to the manufacturers’ instructions. A syringe (5) is employed

at the top of the standpipe to volumetrically quantify the amount of air that

passes through the membrane test device. The reservoir is positioned to

minimise head pressure effects.

A Hatteland CMD-10 (Hatteland Instrumentering, Royken, Norway) pulsed

doppler microbubble counter (3) is attached to the tubing 20 cm after the

arterial outlet of the test device. The device is set at a maximum sensitivity

and is utilised to detect micro bubbles not macro bubbles. This device is

connected to a computer with a COMAC computer interface supported by

BUBMON (version 1.6 Hatteland Instrumentering, Royken, Norway). The

Appendix 8

225

addition of the bubble counter in the test circuit is to compare the bubble

counts with the amount of air collected in the standpipe after the test device.

Calibration

The circuit is verified by initially injecting a 0.5-ml, 1 ml, 5 ml and 10 ml bolus

into the circuit with no test oxygenator. The injected air is then collected at the

pipette to verify that the standpipe is effectively collecting all the air injected.

Injecting 0.5 ml, 1 ml, 5 ml and 10 ml before the mercury drop after which the

change in distance is recorded does calibration of the dilato meter.

Test devices

At this time only two neonatal oxygenators have been investigated, Polystan

Microsafe and Dideco Lilliput. Although their performance characteristics are

very comparable (Table 1) their design is not. The Microsafe consists of two

cylinders. The first cylinder forms the heat exchanger made of stainless steel

tubes. The second cylinder holds the gas exchange membrane. A rigid tube

connects both cylinders. Although blood flow will be the same everywhere in

the device the velocity will not. Blood velocity will be high in the inlet

connector, outlet connector and the connecting tube between both cylinders

and much lower in the cylinders itself. In the Lilliput blood flows through the

inner side of a spiral corrugated pipe heat exchanger after which it enters the

fibre stack for gas exchange. This design gives high velocities in inlet and

outlet connectors and lower velocity in the rest of the oxygenator. Both

devices have a top to bottom flow path in order to establish a bubble trap

function in the oxygenator. The fibres in the Microsafe are in an angle of

Appendix 8

226

approximately 15° to the length axis of the membrane module. In the Lilliput

the fibres are in parallel with the length axis of the oxygenator module and the

clearly visible spacing wires are perpendicular to the length axis. The

pressure drop over the oxygenator is lower in the Microsafe design compared

to the Lilliput design.

Appendix 8

227

Results

Calibration

Calibration of the standpipe (R²= 0.99) and the dilato meter (R²=0.99) shows

an excellent correlation (Figure 2).

Bolus injection

The results of the bolus and slow air injections are shown in Table 2.

Injection of 1 ml air in the Microsafe resulted in a visible air collection of 1 ml

at the bottom of the heat exchanger, no air entrapping in the bubble trap and

no removal of air by the membrane. In the Lilliput no air was observed in the

oxygenator module or in the bubble trap. However the oxygenator contained

0.9 ± 0.1 ml air and 0.1 ± 0.1 ml was removed by the membrane.

The 5 ml bolus injection resulted in visible air (4.7 ± 0.1 ml) in the heat

exchanger module of the Microsafe, no air in the bubble trap and a evacuation

of 0.4 ± 0.2 ml by the membrane. In the Lilliput air was found at the top of the

oxygenator (1.8 ± 0.2 ml), in the bubble trap (0.1 ± 0.2 ml ) and 3.1 ± 0.1 ml

was evacuated by the membrane.

With injection of 20 ml of air, air was found in both heat exchanger and

membrane modules of the Microsafe (8.3 ml) as in the bubble trap (4.9 ml)

while 6.8 ml was evacuated by the membrane. In the Lilliput air was found in

the oxygenator (3.2 ml) and in the bubbletrap (4.2 ml). At the end of the three-

minute period for stabilisation there was still some small movement of the

mercury drop. At that time the membrane evacuated 12.6 ml.

Appendix 8

228

Slow injection

When injecting 20 ml of air at 3 ml / minute the Microsafe entrapped air in

both heat exchanger module and membrane module (12.2 ± 1.9 ml). As more

air entered the oxygenator module it was captured in the same plane as the

arterial outlet connector of the oxygenator after which it travelled along the

fibres towards the arterial outlet connector (Photograph 1). This resulted in 2.7

± 0.1 ml of entrapped air in the bubble trap. The membrane evacuated 5.1 ±

1.7 ml.

In the Lilliput air was observed in the membrane compartment where the

membrane easily removed it. Even after the 10 minutes of stabilisation there

was still some evacuation by the membrane as detected by the dilato meter.

The oxygenator entrapped 2.9 ± 0.1 ml of air, no air was found in the bubble

trap and the membrane removed 17.1 ± 0.1 ml.

Pressure measurements

The pressure drop over the oxygenator during the experiment was 30.1 ± 1.4

mmHg for the Microsafe and 53.1 ± 1.5 mmHg for the Lilliput.

During the bolus injection of 5 and 20 ml there was a sudden transient

increase in pre membrane pressure with the Lilliput for a few seconds after

which it returned to normal.

Appendix 8

229

Discussion

Arterial screen filtration has become a standard procedure in the USA for

paediatric and adult perfusion. The rationale is reducing the risk for gaseous

and/or solid microemboli. However the fear for these adverse effects was

mainly based on the large experience with bubble oxygenators 2-4. Membrane

oxygenators in opposition do have a complete different working principle. The

packed fibres in the gas exchange compartment do not only provide a control

of the blood path but will also work as an effective depth filter for solid

particles. This was confirmed by recent research in a pig model showing no

difference in embolisation of vital organs after three hours of extra-corporeal

circulation using an extra-luminal hollow fibre oxygenator with or without

arterial screen filter 12. Several papers have described a much lower or even

non-existent generation of GME with the introduction of membrane

oxygenators 13,14. However there is a lot of contradiction amongst authors

about the GME generation of a given device 6,13. This inconsistency finds

probably its origin in the non-quantitative nature of the data generated by

bubble counters and in the severe limitations to count accurately GME by

ultrasonic techniques 10,14. Indeed problems can occur from the angle and

coupling of the ultrasonic transducer, the frequency and pulse length of the

device, the electrical circuitry employed, bubble diameter and shape, tubing

diameter and curvature, amount of air and speed at which air is introduced,

red cell interference, and the rate of blood flow 5, 11,. Finally the bubbles

themselves may produce the most serious problems. Signal differences can

Appendix 8

230

result from multiple bubbles clumping, bubbles of different sizes blocking

others, the beam missing multiple bubbles etc.

It was the assumption of the authors that theoretically a membrane

oxygenator must be able to evacuate air because of its combination of a

microporous membrane with a depth filter. The results in this study show

clearly that this hypothesis is correct but highly dependent on the oxygenator

design. Both oxygenators had more or less the same bubble trap efficiency

when confronted with an air challenge. However the Microsafe obtained this

effect mainly by capturing the air in the heat exchanger compartment. Since

the heat exchanger is not permeable for gas the risk for a sudden release of

this air by movement of the oxygenator (taking samples, repositioning) or by

changes in temperature is still existing over time. In the Lilliput is almost

immediately contact between the air and the gas exchange membrane as a

result of this a large portion of the air is evacuated. This difference is most

obvious during the slow injection of air where the Lilliput due to the fact that it

is confronted with smaller quantities of air over time is capable to remove

almost 90% of the air via the membrane after 10 minutes. The Microsafe will

in the same circumstance still release emboli versus the patient.

Repositioning of the arterial outlet connector in another plane could probably

reduce this. Nevertheless this design always will keep an amount of gas

accumulated in the heat exchanger module. Also the construction of the

membrane mat and the existence of high velocities within the body of the

oxygenator seems to be an important aspect. In the Microsafe air is actually

pushed by the high velocity generated in the connection tube between the two

cylinders in the plane of the arterial outlet connector, where it is guided by the

Appendix 8

231

fibres towards the arterial outlet. Interesting was also the influence of the

pressure drop. Theoretically a higher-pressure drop should be beneficial in

the removal of air, especially when there is a good contact with the membrane

material. The Lilliput with the higher-pressure drop and the largest immediate

contact with the membrane is the most effective. As a consequence of this

one should not only open the purge line on an oxygenator when air has

entered the unit but also at the same time raise the pressure downstream the

oxygenator.

Removal of air in a neonatal oxygenator seems to be positively influenced by

following aspects: rapid and large contact with the gas exchange membrane,

contact time between the fibres and the gas, avoidance of zones with high

velocity within the oxygenator module, pressure drop (higher seems

favourable) and the construction of the fibre mat. The dilato meter also

showed an important consideration however that the evacuation of air through

the membrane is the most effective in the beginning and will decrease over

time. This is explained by the reduction of the contact area of the bubble

against the fibre over time.

The results of this study show an important influence of oxygenator design on

the air removal capability of a neonatal oxygenator. Although others have tried

to speculate what was happening to gas once it had entered the oxygenator

they had no means to measure or quantify it7. The technique described in this

paper gives the possibility to actually measure and quantify the capability of a

given device to trap, release or evacuate air. Because of the existing lack in

data we believe this test protocol should become a standard procedure in

combination with pulsed Doppler bubble counting for the evaluation and study

Appendix 8

232

of the air removal and air trap capabilities of membrane oxygenators or

artificial lungs.

Acknowledgements

This study was supported by a specialisation grant from the Flemish Institute

for the promotion of the Scientific-Technological Research in Industry (no.

961181).

Appendix 8

233

References

1. F. De Somer, L. Foubert, J. Poelaert, D. Dujardin, G. Van Nooten, K.

François. Low extracorporeal priming volumes for infants: a benefit?

Perfusion 1996, 11: 455 - 460

2. DT Pearson, MP Holden, SJ Poslad, A Murray, PS Waterhouse. A clinical

evaluation of the performance characteristics of one membrane and five

bubble oxygenators: gas transfer and gaseous microemboli production.

Perfusion 1986; 1: 15 - 26

3. PLC Smith Interventions to reduce cerebral injury during cardiac surgery –

introduction and the effect of oxygenator type. Perfusion 1989; 4: 139 -

145

4. M Sellman, T Ivert, P Stensved, M Högberg, BKH Semb. Doppler

ultrasound estimation of microbubbles in the arterial line during

extracorporeal circulation. Perfusion 1990; 5: 23 - 32

5. M J Miller, R R Johnson. Comparative Analysis of Air Handling in

Membrane Oxygenators. Cobe Cardiovascular, 1996.

6. AP Mehra, A Atkins, A Maisuria, BE Glenville. Air handling characteristics

of five membrane oxygenators. Perfusion 1994; 9: 357 - 362

7. PD Beckley, PD Shinko, JP Sites. A comparison of gaseous emboli

release in five membrane oxygenators. Perfusion 1997; 12: 133 - 141

8. B D Butler, M Kurusz. Gaseous microemboli: a review. Perfusion 1990, 5:

81 – 89

9. W Pugsley The use of Doppler ultrasound in the assesment of

microemboli during cardiac surgery. Perfusion 1989; 4: 115 - 122

Appendix 8

234

10. G Wright, A Furness, S Haigh. Integral pulse frequency modulated

ultrasound for the detection and quantification of gas microbubbles in

flowing blood. Perfusion 1987; 2: 131 - 138

11. M Kurusz, B D Butler. Embolic Events and Cardiopulmonary Bypass. In: G

P Gravlee, R F Davis, J R Utley eds. Cardiopulmonary Bypass Principles

and Practice, Baltimore: Williams & Wilkins, 1993: 267 – 290.

12. M K Dewanjee, S M Wu, M Kapadvanjwala et al. Emboli From an

Extraluminal Blood Flow Hollow Fiber Oxygenator With and Without an

arterial Filter During Cardiopulmonary Bypass in a Pig Model. ASAIO

Journal 1996, 42: 1010 – 1018

13. T Gourlay, J Fleming, K M Taylor, M Aslam Evaluation of a range of

extracorporeal membrane oxygenators. Perfusion 1990; 5: 117 – 133

14. B D Butler Biophysical aspects of gas bubbles in blood. Biomedical

Instrumentation 1985, 19: 59 -62

Appendix 8

235

Table 1: Oxygenator characteristics.

Dideco Lilliput Polystan Microsafe

Maximum blood flow (ml/min) 800 800

Priming volume in membrane and heat

exchanger (ml)

60 52

Minimum volume in venous reservoir (ml) 20 25

Connector size (inch) 3/16 and 1/4 3/16 and 1/4

Appendix 8

236

Table 2: Air removal capabilities of the Lilliput and the Microsafe.

* two measurements, ° one measurement

Appendix 8

237

Figure 1: Test circuit.

(a) calibrated roller pump, (b) Cobe Excel, (c) port for air injection, (d) testdevice, (e) standpipe, (f) reservoir, (1) and (2) pressure measurements, (3)Doppler probe, (4) dilatometer, (5) syringe

Appendix 8

238

Figure 2: Calibration of dilato meter and bubble trap.

R2 = 0.99p<0.001

0

5

10

15

20

0 5 10 15 20

Injected air (ml)

Mea

sure

d ai

r (m

l)

bubble trapdilato meter

Appendix 8

239

Figure 3

Appendix 8

240

Appendix 9

241

In vivo evaluation of a phosphorylcholine coatedcardiopulmonary bypass

F. De Somer, Y. Van Belleghem, L. Foubert, K. François, F. Dubrulle,

D. De Wolf and G. Van Nooten

Journal of Extra-corporeal technology, 1999; 31 (2): 62-67

Appendix 9

242

Abstract

A complete phosphorylcholine coated cardiopulmonary bypass circuit,

including the Dideco D901 oxygenator, was tested for gas transfer, blood path

resistance and biocompatibility in a standardized setting. Blood compatibility

was tested by measuring complement and platelet activation.

Three dogs (mean body weight : 28 ± 3 kg) were placed on cardiopulmonary

bypass at a flow rate of 600 mL/min during six hours. The animals were

weaned from cardiopulmonary bypass and sacrificed electively after seven

days.

Oxygen and carbon dioxide transfer were 26.6 ± 2.4 mL/min and 33.0 ± 1.9

mL/min, respectively. Mean pressure drop across the oxygenator was 52.6 ±

0.2 mmHg. The respective baseline values for thromboxane B2, prostaglandin

E2 and platelet factor 4 were 1817 ± 283 pg/mL, 12783 ± 2109 pg/mL, 0.35 ±

0.08 IU/mL. Thromboxane B2 and prostaglandin E2 increased slightly to 2881

± 868 pg/mL and 18083 ± 3144 pg/mL at 30 minutes of bypass, whereas

platelet factor 4 values remained stable during the procedure. Concentrations

of tumor necrosis factor α and complement split products C5a were only

mildly increased.

After use scanning electron microscopy was performed on the inner housing,

heat exchanger and outer surface of the hollow fibres. No thrombi nor

organised cellular deposits were found on any of the components.

Phosphorylcholine coating of CPB seems to be very promising regarding

platelet activation and complement activation.

Appendix 9

243

Introduction

Materials used in a cardiopulmonary bypass (CPB) circuit are not originally

developed for this application. In general these materials activate the

coagulation, complement and fibrinolysis cascades. Together with turbulent

flow patterns, zones of blood stasis and the aspiration of shed blood this

contributes to the bio-in-compatibility of CPB. In order to reduce this bio-in-

compatibility several approaches have been proposed : reduction of foreign

surface area, more even distribution of blood flow, avoiding stasis and blood-

air interfaces, use of anticoagulant and antifibrinolytic drugs and surface

modification of the different materials. An alternative approach is the

development of bio-membrane-mimetic surfaces. Such surfaces are designed

to mimic the outer surface of blood cells [1, 2]. This outer surface is

predominantly composed of phosphorylcholine groups, which contribute

largely to the non-thrombogenic properties of blood cells. Recent research

shows that polymers containing phosphorylcholine reduce protein adsorption

and complement activation markedly [3].

This study investigates the impact of a complete phosphorylcholine coated

CPB circuit on the oxygen transfer, blood elements, coagulation and

complement activation.

Appendix 9

244

Material and Methods

The study group comprised three male Labrador dogs with an average weight

of 28 ± 3 kg. All animals received care in accordance with institutional

guidelines and national laws.

A CPB circuit was phosphorylcholine coated from cannula to cannula a. The

circuit consisted of PVC tubing, a D901 neonatal oxygenator with closed

venous reservoir a and a venous and arterial cannula b. The total priming

volume of the circuit was 208 ± 9 mL. The dogs were instrumented, and

cannulated via the right carotid artery and right jugular vein. Partial bypass

was instituted by means of a roller pump c for a six-hour period at a blood flow

of 600 mL per minute. The gas to blood ratio was one to one. Before

cannulation animals were heparinised with 300 IU/kg body weight. Activated

clotting time was measured with a Hemochron celite tube d and was

maintained above 300 seconds during the procedure. Animals were kept at

normothermia during the whole procedure.

Pre and post membrane pressures were automatically recorded every 10

seconds by means of a Cobe Perfusion Controller (Cobe Cardiovascular,

Arvada, USA) connected to a Personal Computer. Arterial and venous blood

gases were taken every hour. Red blood cell count, haematocrit,

haemoglobin, white blood cell count and formula, platelet count, electrolytes,

free plasma haemoglobin, APTT, PTT, fibrinogen, thromboxane B2 (TXB2),

prostaglandin E2 (PGE-2), platelet factor 4 (PF4), C5a and TNFα were

a Dideco, Mirandola, Italyb Stöckert, Münich, Germanyc Cobe Cardiovascular, Arvada, Co

Appendix 9

245

analysed before instituting CPB and at 30, 60, 120, 240 and 360 minutes of

CPB. All values were corrected for haemodilution using following formula:

Corrected value = measured value * (start haematocrit/actual haematocrit).

Haemolysis rate (HR) was calculated using following formula:

HR=free plasma haemoglobin (mg/mL)/ haematocrit

Data are expressed as mean value ± standard error of the mean.

At the end of the experiment, the extracorporeal circuit was checked for

visible clots and fibrinogen deposits. Electron microscopy was performed on

the oxygenator inner housing, heat exchanger, fibres and knots of the weft

yarn.

One week after the experiment the animals were sacrificed and autopsy of the

lungs, kidneys and heart was performed.

Analysis techniques

Radioimmunoassay was used for the determination of TXB2 e, PGE-2 f and

C5a e. Platelet factor 4 g and TNFα e were analysed using enzyme-linked

immuno-sorbent assay (ELISA).

Data analysis

All data are presented as mean ± standard error of the mean. Statistical

analysis was done by using the Friedman test corrected for multiple

comparisons. Results were significant when p < 0.05.

d International Technidyne Corporation, Edison, NJe Amersham International, UKf Perseptive Biosystems, USAg Boehringer, Germany

Appendix 9

246

Results

Mass transfer

Mean oxygen transfer was 26.6 ± 2.4 mL/min (Figure 1). Mean carbon dioxide

removal was 33.0 ± 1.9 mL/min.

Inlet and outlet oxygenator pressures

Mean inlet and outlet pressure before and after the oxygenator was 160.9 ±

0.3 mmHg and 108.4 ± 0.2 mmHg, respectively. Mean pressure drop across

the D901 was 52.6 ± 0.2 mmHg.

Haematology and haemolysis (Table 1)

Platelet count started at a mean value of 159 ± 55 /mm³ pre CPB and

decreased to a mean value of 123 ± 34 at 30 minutes of CPB, after which it

normalized to a mean value of 150 ± 22 /mm³ at the end of the experiment (p

= NS).

White blood cell count started at 6700 ± 300 /mm³, decreased to 5300 ±

200/mm³ at 30 minutes and then steadily increased to 12200 ± 1100 /mm³ at

the end of CPB (p = 0.01). Differentiation of the white blood cell count showed

no major changes with exception of the eosinophils, which decreased from a

baseline average value of 6.3 % to 0.3 % at the end of the experiment.

Free plasma haemoglobin levels started at a mean value of 63 ± 15 mg/100

mL pre CPB and stabilized at a mean value of 43 ± 11 mg/100 mL at the end

of CPB (p = NS). Haemolysis rate started at a mean value of 0.26 ± 0.04

Appendix 9

247

mg/mL cells and decreased over time to an average value of 0.10 ± 0.01

mg/mL cells.

Inflammatory response (Figure 2)

C5a levels raised from 14 ± 2.1 IU/mL to 25.9 ± 9.5 IU/mL (p = NS) at 30

minutes of CPB after which they returned to baseline values. TNFα started

from a baseline value of 2.8 ± 0.1 pg/mL, then increased to 3.6 ± 1.5 pg/mL (p

= NS) at 30 minutes, and finally decreased to 2.8 ± 0.9 pg/mL at 360 minutes.

Platelet activation (Figure 3)

Thromboxane B2 and PGE-2 levels increased from a starting value of 1817 ±

283 pg/mL and 12783 ± 2109 pg/mL respectively, to 2881 ± 868 pg/mL (p =

NS) and 18083 ± 3144 pg/mL (p = NS) at 30 minutes, after which the levels

returned to 1595 ± 353 pg/mL and 9897 ± 3175 pg/mL at the end of CPB.

Platelet factor 4 and LDH values remained stable during the experiment.

Autopsy

Autopsy of heart, lungs and kidneys did not reveal any pathologic lesions in

the dogs.

Scanning electron microscopy

Examination of the polycarbonate housing (Figure 4), the stainless steel heat

exchanger (Figure 5), polypropylene fibres (Figure 6) and knots of the weft

yarn (Figure 7) showed almost no deposition of proteins and platelets.

Appendix 9

248

Discussion

Although phosphorylcholine coatings have already been applied with good

results on chest tubes and coronary stents they have never been used in a

complete CPB circuit [4, 5].

Coating of micorporous hollow fibres can cover the micropores with a small

layer of coating what can result in a higher resistance to diffusion. The

application of a small layer of phosphorylcholine on the gas exchange fibres

does not influence the oxygen transfer of the oxygenator (Figure 1). Scanning

electron microscopy photographs of the hollow fibre showing open pores

support this finding (Figure 6). The obtained oxygen transfer data are not only

comparable with the ones provided by the manufacturer but also show a high

reproducibility amongst the different oxygenators.

Whole body inflammatory response to CPB is highly complex, and

complement appears to be just one component. The alternative complement

pathway is activated during CPB and results in the activation of C5 to C5a

and C5b. C5a activates neutrophils and C5b initiates the formation of the

membrane attack complex, which is capable of producing cell lysis and death

[6]. Whereas complement levels up to four times the baseline are observed

during CPB [7, 8], during our experiment only a relatively small increase in

C5a level of 46% is noted at 30 minutes of bypass. During CPB, leukocyte

count first decreases, in response to haemodilution, after which it increases

moderately during the procedure. At the same time monocytes and

neutrophils are activated, while lymphocytes count decreases resulting in a

higher susceptibility to infection postoperatively [6]. Our results show a

Appendix 9

249

comparative evolution in white blood cell count, but no major changes in white

blood cell differentiation with exception of the eosinophil count. Although no

real markers for neutrophil activation, such as Mac-1, were measured, one

could speculate on less activation of neutrophils due to the lower generation

of C5a. This is in line with previous research [9]

In uncoated circuits platelet activation is expressed by an increase of both

thromboxane B2 and PGE-2 up to 4 times the baseline [7, 10]. In our study

only a mild activation of platelets is observed at 30 minutes of bypass, but the

overall activation by the surface is much lower as shown by the constant

values of thromboxane B2, platelet factor 4, PGE-2 and LDH. This suggests

that a phosphorylcholine coated CPB has excellent non-thrombogenic

characteristics.

Free plasma haemoglobin levels and haemolysis rates show values

comparable with those reported for uncoated circuits [7].

When heparinised blood comes into contact with nonendothelial surfaces,

plasma proteins are instantaneously adsorbed onto the surface. All these

nonendothelial surfaces produce a thrombotic stimulus, but the stimulus

seems to vary between surfaces. Heparin-bound surfaces seem to be more

thromboresistant. This study, in agreement with the literature, shows that

phosphorylcholine coated surfaces are at least equal in thromboresistance as

is shown by the SEM analysis of housing, fibres, heat exchanger and weft

yarn. However, in contrast to heparin bound surfaces, which loose their

antithrombotic properties after exposure to protamine, phosphorylcholine

coating can be expected to resist to contamination of the circulating blood with

protamine [11].

Appendix 9

250

Phosphorylcholine coating of CPB seems to be very promising regarding

platelet activation and complement activation, which makes it a full alternative

for heparin bound surfaces. However, these promising results should be

confirmed by expanding the series. Clinical studies should clarify if these

results can be reproduced during cardiac surgery with a certain degree of

organ ischaemia and reperfusion, which are both known to activate

complement and platelets.

Appendix 9

251

References

1. Yianni YP: Biocompatible surfaces based upon biomembrane mimicry. In

Quinn PJ, Cherry RI (eds.). Structural and Dynamic Properties of Lipids

and Membranes. Portland. Press Research Monograph. 1992, pp 187-216

2. Campbell EJ, O’Byrne V, Stratford PW, Quirk J, Vick TA, Wiles MC, Yianni

YP. Biocompatible surfaces using methacryloylphosphorylcholine

laurylmethacrylate copolymer. ASAIO 1994; 40: 853-857.

3. Yu J., Lamba NMK, Courtney JM et al. Polymeric biomaterials: influence of

phosphorylcholine polar groups on protein adsorption and complement

activation. Int. J. Artif. Organs 1994; 17: 499-504.

4. Hunter S, Angelini GD: Phosphorylcholine coated chest tubes improve

drainage after open heart surgery. Ann. Thorac. Surg. 56: 1139-1342,

1993.

5. Nordrehaug JE, Chronos NAF, Sigwart U: A biocompatible coating applied

to metallic stents (abstract). J.Am.Coll.Cardiol. 5A 1994

6. Edmunds LH Jr. Cardiopulmonary bypass and blood. In Pifarré R. (ed.)

Blood conservation with aprotinin. Philadelphia. Hanley & Belfus Inc. 1995,

pp 45 – 67.

7. Hatori N, Yoshizu H, Haga Y, Kusama Y, Takeshima S, Segawa D,

Tanaka S. Biocompatibility of heparin-coated membrane oxygenator

during cardiopulmonary bypass. Artificial Organs 1994; 18: 904-910.

8. Ashraf S, Tian Y, Cowan D, Entress A, Martin PG, Watterson KG. Release

of proinflammatory cytokines during pediatric cardiopulmonary bypass:

Appendix 9

252

Heparin-bonded versus nonbonded oxygenators. Ann Thorac Surg 1997;

64:1790-4.

9. DeFife KM, Yun JK, Azeez A, Stack S, Ishihara K, Nakabayashi N, Colton

E, Anderson JM. Adhesion and cytokine production by monocytes on poly

(2-methacryloyloxyethyl phosphorylcholine-co-alkyl methacrylate)-coated

polymers. J Biomed Mater Res 1995; 29: 431-439.

10. Butterworth JF, Utley JR, Swain JA. Neuroendocrine and electrolyte

responses to cardiopulmonary bypass. In Gravlee GP, Davis RF, Utley JR

(eds). Cardiopulmonary Bypass. Principles and Practice. Baltimore.

Williams and Wilkins. 1993, pp 305-307.

11. von Segesser LK, Gyurech DD, Schilling JJ, Marquardt K, Turina MI. Can

protamine be used during perfusion with heparin surface coated

equipment? ASAIO J. 1993; 39: M190-4

Appendix 9

253

Table 1. Evolution of haematocrit, leukocyte count, white blood cell

differentiation, platelets, free plasma haemoglobin. Values are expressed as

mean ± standard error of the mean.

Time (minutes) 0 30 60 120 240 360

Haematocrit (%) 39.7±2.

8

37.4±2.

7

37.0±2.

3

41.8±4.

1

30.8±2.

8

31.5±1.

8

WBC (x1000/mm³) 6.7±0.3 5.3±0.2 6.2±0.5 7.4±0.5 10.6±0.

7

12.2±1.

1

Segmented (%) 68.7±1.

2

71.0±2.

1

74.3±2.

4

77.3±2.

4

83.3±1.

3

84.7±0.

9

Lymphocytes (%) 20.7±0.

9

18.7±1.

9

16.0±1.

0

13.7±1.

5

12.0±1.

5

11.3±0.

3

Monocytes (%) 4.0±1.7 5.3±0.9 5.3±1.2 6.0±1.5 3.7±1.7 3.7±1.2

Eosinophils (%) 6.3±0.9 5.0±0.0 4.3±0.7 3.0±0.0 1.0±0.0 0.3±0.3

Platelets (x1000/mm³) 159±55 123±34 127±31 121±30 132±28 150±22

Free plasma haemoglobin

(mg/100mL)

63±15 66±27 81±29 58±21 54±20 43±11

Appendix 9

254

Figure 1. Oxygen transfer.

30 40 50 60 70 80 90 100Venous saturation (%)

10

20

30

40

Oxy

gen

trans

fer (

mL/

min

)r2 = 0.88

Appendix 9

255

Figure 2. Evolution of complement C5a and TNFα during the experiment.

Data are expressed as mean ± standard error of the mean.

0 30 60 120 240 360Time (minutes)

10

15

20

25

30

35

IU/m

L

C5a

0 30 60 120 240 360Time (minutes)

1.5

2.5

3.5

4.5

pg/m

L

TNFα

Appendix 9

256

Figure 3. Evolution of thromboxane B2, PGE-2, PF4 and LDH. Data are

expressed as mean ± standard error of the mean.

0 30 60 120 240 360Time (minutes)

1000

2000

3000

4000

pg/mL

ThromboxaneB2

0 30 60 120 240 360Time (minutes)

10000

20000

30000

pg/mL

PGE-2

0 30 60 120 240 360Time (minutes)

0.2

0.3

0.4

0.5

IU/mL

PF4

0 30 60 120 240 360Time (minutes)

70

90

110

130

IU/L

LDH

Appendix 9

257

Figure 4. Scanning electron microscopic view of the inner side of the

polycarbonate housing. There are no thrombi or organised cellular structures.

A few cells and some protein deposition can be seen.

Appendix 9

258

Figure 5. Scanning electron microscopic view of the stainless steel heat

exchanger. There is absence of thrombi and cellular structures.

Appendix 9

259

Figure 6. Scanning electron microscopic view of a polypropylene hollow fibre.

The open pores can be clearly seen. There is no evidence of thrombi or

cellular structures.

Appendix 9

260

Figure 7. Scanning electron microscopic view of a weft yarn. There are no

organised cellular structures or thrombi.

Appendix 10

261

Phosphorylcholine coating of extracorporeal circuits providesnatural protection against blood activation by the material

surface

F. De Somer, K. François, W. van Oeveren, J. Poelaert, D. De Wolf, T. Ebels,

G. Van Nooten

European Journal of Cardio-Thoracic Surgery, 2000; 18(5): 602-606

Appendix 10

262

Abstract

Objective: The aim of this study is to evaluate the use of a new coating,

mimicking the outer cell membrane, in paediatric cardiac surgery.

Methods: Two groups of ten patients with a body weight below 8 kg,

undergoing elective cardiac operations for different congenital anomalies,

were prospectively enrolled in this study. In one group the whole

extracorporeal circuit, including the cannulas, was coated with

phosphorylcholine (PC). In the second group the same circuit was used

without coating. Platelet activation (Thromboxane B2, β-Thromboglobulin),

activation of the coagulation system (F1+2), leukocyte activation

(CD11b/CD18) and complement activation (TCC) were analysed prae CPB, at

15, 60 minutes of CPB, at the end of CPB, 20 minutes post CPB and at

postoperative day 1 and 6.

Results: No statistical differences were found for F1+2 and CD11b/CD18.

After onset of CPB mean levels of TCC remained stable in the PC group

whereas an increase was observed in the control group. During CPB βTG

values in both groups increased to a maximum at the end of CPB. Within

groups the increase in βTG levels during CPB was statistically significant (p <

0.05) from baseline in the control group starting from 60 minutes of CPB

whereas no statistical difference was observed in the PC group. After the start

of CPB TXB2 mean levels increased to 405 ± 249 pg/mL in the PC group

versus 535 ± 224 pg/mL in the control group. After this initial increase there

was a small decline in the PC group with further increase. This was in contrast

Appendix 10

263

to the control group were TXB2 levels further increased up to a mean of 718 ±

333 pg/mL at the end of CPB (p = 0.016).

Conclusions: Phosphorylcholine coating had a favourable effect on blood

platelets, which is most obvious after studying the changes during

cardiopulmonary bypass. A steady increase of TXB2 and β thromboglobulin

was observed in the control group, whereas plateau formation was observed

in the phosphorylcholine group. Clinically, this effect may contribute to

reduced blood loss and less thromboembolic complications. Complement

activation is lower in the coated group.

Keywords: Phosphorylcholine coating; Paediatric surgery; Cardiopulmonary

bypass; Platelets; Complement.

Appendix 10

264

Introduction

Approximately 56000 paediatric cardiopulmonary bypass operations were

performed in Europe and the United States in 1996. The anticipated continued

growth of paediatric cardiac surgical practice due to improvement of

technology has shown a 10% increase in last years. Although small babies

are much more vulnerable to inflammatory response due to the larger volume

and foreign surface area of the extracorporeal circuit, the smaller neonatal

oxygenators became available only a few years ago. A further improvement of

the extracorporeal circuit is expected to be related to the surface

characteristics. Depending on the treatment of polymer materials, its surface

may be modified to reduce thrombogenic or inflammatory reactions. Heparin

coating, which is known to reduce the inflammatory reactions, was just

recently introduced for use in paediatric bypass [1]. An antithrombogenic

coating is not commonly used as yet, but may be achieved by application of

phosphorylcholine (PC). This coating will produce interfacial characteristics,

which largely mimic the main lipid headgroup component of the outer cell

membrane [2]. In contrast to the negatively charged phospholipids of the inner

membrane, these neutral phospholipids do not activate the clotting system

and are therefore non-thrombogenic, as would be expected for a major

component of the outer surface of an erythrocyte [2,3]. Till today only limited

experience with phosphorylcholine coatings is available [4,5]

Since coagulation in infants is more delicate than in adults, if not only by the

reduced availability of inhibitors, an antithrombogenic coating was anticipated

to be most profitable for paediatric cardiopulmonary bypass.

Appendix 10

265

The use of PC coated circuits as compared to uncoated extracorporeal

circuits in elective paediatric cardiac surgery was evaluated in this study, by

means of clinical and biochemical evaluation.

Appendix 10

266

Materials and Methods

Two groups of ten patients with a body weight below 8 kg, undergoing elective

cardiac operations for different congenital anomalies (Table 1). Patient

selection was consecutive from 9/6/1998 to 20/1/1999 including all patients. In

the PC group the whole extracorporeal circuit, including the cannulas, was

coated with phosphorylcholine (Dideco, Mirandola, Italy). In the control group

the same circuit was used without coating. Informed parental consent was

obtained for all patients, according to the regulations of the hospital medical

ethics committee.

Cardiopulmonary bypass (CPB) consisted of a D901 neonatal oxygenator with

integrated collapsible venous reservoir (Dideco, Mirandola, Italy), cardiotomy

reservoir (Dideco, Mirandola, Italy) and a custom tubing pack made of PVC.

Priming volume was 200 mL. Priming solution consisted of Plasmalyte-A

(Baxter, Lessines, Belgium), human albumin (Red Cross, Brussels, Belgium)

and packed red cells were added in order to obtain a 4% concentration of

human albumin in the priming solution and an intraoperative haematocrit of

30%. Five hundred IU of porcine heparin was added to the prime (Roche,

Brussels, Belgium). Before cannulation patients were heparinised with 300

IU/kg body weight. Activated clotting time was measured with a Medtronic

kaolin cartridge (Medtronic Hemotec, Parker, CO) and was maintained above

400 seconds during the procedure. Patients were systemically cooled to an

eosophageal temperature of 25°C and weaned of CPB when rectal

temperature was above 34°C. Blood flow rates were maintained to ensure

adequate tissue perfusion.

Appendix 10

267

Arterial and venous blood gases were taken at 15 minutes, 30 minutes and

subsequently every 30 minutes of CPB. Blood samples for determination of

complement activation (Terminal Complement Complex), platelet activation

(Thromboxane B2, β-Thromboglobulin), activation of the coagulation

(Fragment 1+2) and white blood cell activation (CD11b/CD18) were taken

after induction, at 15 and 60 minutes of CPB, at the end of CPB, post CPB

and at postoperative day 1 and 6.

Analysis methods

TCC (C5b-9) was determined by means of an Enzyme Linked

ImmunoSorbent Assay (ELISA) (Quidel, San Diego, CA).

Thromboxane represents activation of the arachidonic pathway in platelets,

and was determined by means of ELISA (Biotrak, Amersham, UK).

β-Thromboglobulin was obtained by an ELISA technique (Diagnostica Stago,

Boehringer Mannheim, BRD) and represents the release of α-granules from

platelets.

Fragment 1+2 is released after cleavage of prothrombin to thrombin.

Fragment 1+2 has no biological activity and remains in blood indicating

activation of the clotting system. Fragment 1+2 was determined by ELISA

(Dade Behring, Marburg, BRD).

Fifty µL of whole blood was incubated with 10µL CD18 antibody (clone 130,

Becton Dickinson, USA) conjugated with FITC and 10 µL CD11b antibody

(clone D12, Becton Dickinson, USA) conjugated with phycoerythrin. The cells

were incubated during 20 minutes at room temperature in the dark, then RBC

were lysed and WBC fixed with Uti-Lyse (Dako) and two color flow cytometric

analyses were performed on a FACSort (Becton Dickinson, USA) equipped

Appendix 10

268

with a single argon ion laser. A minimum of 10000 cells was analysed per

sample. Analyses were performed on a lymhogate with CellQuest software.

Statistics

All data are presented as mean ± standard deviation. Statistical analysis was

done using a Friedman test for the within variation, a Wilcoxon test for the

paired comparison and a Kruskal-Wallis test for the between comparison. The

individual p-values were corrected using following formula: αind=1-(1-αjoint)1/m.

Results were considered to be significant when p<0.05.

Appendix 10

269

Results

Terminal Complement Complex (TCC)

Baseline levels of TCC were different for both groups (145 ± 94 ng/mL (PC)

versus 64 ± 32 ng/mL (Control); p = 0.04) (Figure 1). After onset of CPB mean

levels stayed stable in the PC group (130 ± 146 ng/mL) whereas an increase

to 138 ± 110 ng/mL was observed in the control group (not significant). With

progress of CPB an increase in TCC was noticed in both groups. Within

groups the increase in TCC was statistical significant from baseline at end of

bypass (p = 0.012) and after protamine administration (p = 0.005) in the PC

group, while in the control group statistical difference was reached at 60

minutes (p = 0.018), end of CPB (p = 0.005) and after protamine

administration (p = 0.005). On postoperative day 1 levels in both groups were

at baseline again.

β-thromboglobulin (βTG)

Baseline levels of βTG were different in both groups, 427 ± 202 ng/mL in the

PC group versus 233 ± 158 ng/mL in the control group (p = 0.013). During

CPB values in both groups increased to a maximum at the end of CPB

(Figure 2). Within groups the increase in βTG levels during CPB was

statistically significant (p < 0.05) from baseline in the control group starting

from 60 minutes of CPB whereas no statistical difference was observed in the

PC group.

Thromboxane B2 (TXB2)

Baseline levels of TXB2 were similar in both groups. (PC, 117 ± 109 pg/mL

versus control 125 ± 163 pg/mL, not significant). After start of CPB TXB2

Appendix 10

270

mean levels increased to 405 ± 249 pg/mL in the PC group versus 535 ± 224

pg/mL in the control group. After this initial increase there was a small decline

in the PC group with further increase (Figure 2). This in contrast to the control

group were TXB2 levels further increased up to a mean of 718 ± 333 pg/mL at

the end of CPB (p = 0.016).

Fragment 1+2

Fragment 1+2 mean values were low in both groups and did not exceed 4

nmol/L. No statistical differences were observed between and within both

groups.

CD11b/CD18

CD11b/CD18 expression rose progressively in both groups and peaked at a

value of 4 to 5 times the baseline level at 60 minutes of CPB, being in most

cases, the first measurement after release of the aortic crossclamp.

Subsequently the expression declined towards normal values on

postoperative day 1.

Mass transfer

The mean oxygen transfer was 4.0 ± 1.3 mL O2/100 mL blood in the PC group

versus 4.4 ± 1.3 mL O2/100 mL blood (p = NS) in the control group. Mean

CO2 removal was 3.2 ± 1.5 mL C02/100 mL blood in the PC group and 3.1 ±

1.4 mL C02/100 mL blood in the control group (p = NS).

Appendix 10

271

Discussion

We studied the thrombogenic and inflammatory response after the use of

phosphorylcholine coating, by evaluating progression of βTG release and

thromboxane production, both related to platelet activation, and complement

activation. Also the interaction of the phosphorylcholine coating on the gas

transfer properties of the hollow fibre membranes was evaluated.

Although the literature shows an improved biocompatibility in adult surgery

when using coatings [6], the thrombogenic and inflammatory response is

usally mild in routine adult surgery which makes it difficult to demonstrate

differences in postoperative clinical response. Small babies are much more

vulnerable to the adverse effects of cardiopulmonary bypass due to the

relatively high priming volume and relative large blood foreign material surface

in contact with blood. Additionally several organ systems are still immature.

The characteristic feature of biological membranes is their functional and

compositional lipid asymmetry, which has been described in several cell types

and is thought to stem from the requirement of biological membranes to have

asymmetric protein distributions across the bilayer. In all of the cells for which

lipid compositional asymmetry has been described, negatively charged

phospholipids are found predominantly on the inner cytoplasmatic side of the

membrane, while the neutral zwitterionic PC-containing antithrombotic lipids

predominate in the outer membrane leaflet. Negatively charged phospholipids

are thrombogenic and it has been proposed that this membrane asymmetry

may serve the biological purpose in the maintenance of the delicate balance

between haemostasis and thrombosis. In vitro experiments, in which various

Appendix 10

272

phospholipid coatings were applied to surfaces, showed a very high

procoagulant activity of negatively charged phospholipids was shown. This is

in contrast to the PC-containing surfaces that were not active in coagulation

tests [2,3]. We did not observe an inhibition of activation of the clotting

system, which may indicate a merely passive effect of the PC coating towards

the clotting system. Additionally, F1+2, a cleavage product of prothrombin

during thrombin generation, was very low in our study, indicating proper

anticoagulation during CPB and proper sample collection throughout. Since

F1+2 concentrations of 4 nmol/L are even not noteworthy in a clinical sense, a

comparison between the systems cannot be made under the present

conditions. However both markers of platelet activation showed that the PC

coated circuits were activating mildly and for a short period of time, whereas

the uncoated circuits continued to activate platelets. A difficulty is that the

platelet release product β thromboglobulin is sensitive to release during blood

sampling and processing, especially in non-coagulated blood. Typical for this

parameter is a large individual difference. This may have caused an increase

of the “baseline” β-thromboglobulin concentrations, which was determined in

samples collected after thoracotomy.

Concentrations of TXB2 in uncoated systems followed the pattern of previous

observations with a gradual increase towards end of CPB. In contrast, TXB2

concentrations increased in the phosphorylcholine coated group for only a

short period of time and were already reduced at 60 minutes in 5 out of 7

determinations. It indicates a short exposure of platelets to an activating

surface that rapidly became passive. TXB2 formation appeared most of all

Appendix 10

273

restricted to the operating period, since postoperatively a return to baseline

was observed.

Cell adhesion to biomaterials is a surface dependent event, which is

additionally influenced by the dynamic interaction between proteins and the

material surface [7-9]. The low platelet activation may be due to the affinity of

the phosphorylcholine coating for phospholipids, which may immediately

adsorb to the polymer surface because they are smaller and more

concentrated than most proteins [10]. The adsorbed phospholipids may then

assemble by themselves and form an organised layer on the surface just like

real biomembranes [10], which then interacts minimally with proteins and

cells.

Few series have evaluated heparin coating in paediatric CPB [1,9-10].

Reduced complement activation has been observed as in adult CPB

[1,11,12]. To our surprise, also the PC coating appeared to generate less

complement activation than the uncoated systems. Although baseline

concentrations were slightly different between both groups the increase of

TCC was far more pronounced in the uncoated group (6 times baseline)

compared to the coated group (2 times baseline). For the first 60 minutes of

CPB the differences can be mainly attributed to material dependent activation

by the extracorporeal circuit. Thereafter in both groups further TCC generation

was observed. In the coated group a few patients showed very high TCC

generation probably due to longer reperfusion time. It is known that rewarming

and return of suctioned blood markedly contribute to complement activation

during the later period of CPB, which may have caused the large individual

differences. After CPB no further increase of TCC was observed, although

Appendix 10

274

protamine can cause some additional complement activation. The return to

baseline at day 1 shows rapid recovery from the CPB insult.

In vitro experiments showed decreasing complement activation with

increasing surface phoshorylcholine mole fractions [10], suggesting that the

phosporylcholine is responsible for the reduction. The working mechanism is

probably related to lesser activation of the complement protein C5 [13] and

the inhibition of monocyte and macrophage adhesion [14].

Two of the biochemical tests showed a different baseline, namely β-

thromboglobulin and TCC. For both of these tests it is known that particularly

in infants large individual differences exist. Comparison of these variables with

historical data obtained in a similar group of patients showed that β-

thromboglobulin baseline values ranged between 150 and 450 IU/mL [15].

Historical baseline TCC values in infants ranged between 40 and 460 ng/mL

[1]. Obviously, values from most samples in our study fell within those ranges

and must be considered normal baselines.

General conclusion

Phosphorylcholine coating appears to have a favourable effect on blood

platelets, which is most obvious after studying the changes during

cardiopulmonary bypass. A steady increase of TXB2 and β thromboglobulin

was observed in the control group, whereas plateau formation was observed

in the phosphorylcholine group. Clinically, this effect may contribute to

reduced blood loss and less thromboembolic complications. Also complement

Appendix 10

275

activation is lower in the coated group. The limited number of patients in this

study, however, only allows speculations as to the clinical relevance.

Limitations of the study

Due to the fact that our study concerns a biological system with relatively

large standard deviations in a limited number of patients, our data should be

interpreted with caution. Moreover, the relative extensive use of blood suckers

during many cases in this study, will cause an important activation of the

coagulation and complement cascades. For these reasons large randomised

studies are necessary to investigate in depth the efficacy of coated CPB

circuits during paediatric open heart operations.

Appendix 10

276

References

1. Scheurs HH, Wijers MJ, Gu J. van Oeveren W, van Domburg T, de Boer

JH, Bogers AJJC. Heparin-coated bypass circuits: effects on inflammatory

response in pediatric cardiac operations. Ann Thorac Surg 1998;66:166-

71.

2. Zwaal RFA, Hemker HC. Blood Cell Membranes and Haemostasis.

Haemostasis 11: 12 – 39, 1982.

3. Yianni YP. Biocompatible surfaces based upon biomembrane mimicry. In:

Quinn PJ and Cherry RJ, editors Structural and dynamic properties of

lipids and membranes. London: Portland Press Ltd, 1992: 182-217.

4. Hunter S, Angelini GD. Phosphatidylcholine-Coated Chest Tubes Improve

Drainage after Open Heart Operation. Ann Thorac Surg, 1993, 56: 1339-

42.

5. von Segesser LK, Tonz M, Leskosek B, Turina M. Evaluation of

phospholipidic surface coatings ex-vivo. Int J Artif Organs, 1994, 17: 294-

300.

6. Fukutomi M, Kobayashi S, Niwaya K, Hamada Y, Kitamura S. Changes in

platelet, granulocyte and complement activation during cardiopulmonary

bypass using heparin-coated equipment. Artif Organs, 1996; 20: 767-776.

7. Lewis JC, Hantgan RR, Stevenson SC, Thornburg T, Kieffer N, Guichard

J, Breton-Gorius J. Fibrinogen and glycoprotein IIb/IIIa localization during

platelet adhesion. Am J Pathol 136:239-252, 1990

Appendix 10

277

8. Lee JH, Lee HB. Platelet adhesion onto wettability gradient surfaces in the

absence and presence of plasma proteins. J Biomed Mater Res 41:304-

311, 1998

9. Lindon JN, McManama G, Kushner L, Merrill EW, Salzman E. Does the

conformation of adsorbed fibrinogen dictate platelet interactions with

artificial surfaces? Blood, 1986; 68: 355-362.

10. Ishihara K, Nakabayashi N. Hemocompatible Cellulose Dialysis

Membranes Modified with Phospholipid Polymers. Artif Organs 1995;

19(12): 1215-1221.

11. Kagaisaki K, Masai T, Kadoba K, Sawa Y, Nomura F, Fukushima N,

Ichikawa H, Ohata T, Suzuki K, Taketani S, Matsuda H. Biocompatibility of

heparin-coated circuits in pediatric cardiopulmonary bypass. Artif Organs

1997; 21:836-840.

12. Ashraf S, Tian Y, Cowan D, Entress A, Martin PG, Watterson KG. Release

of proinflammatory cytokines during paediatric cardiopulmonary bypass:

Heparin-bonded versus nonbonded oxygenators. Ann Thorac Surg 1997;

64: 1790-4.

13. Yu J, Lamba NMK, Courtney JM, Whateley TL, Gaylor JDS, Lowe GDO,

Ishihara K, Nakabayashi N. Polymeric biomaterials: influence of

phosphorylcholine polar groups on protein adsorption and complement

activation. Int J Artif Organs 1994; 7: 499-504.

14. DeFife KM, Yun JK, Azeez A, Stack S, Ishihara K, Nakabayashi N, Colton

E, Anderson JM. Adhesion and cytokine production by monocytes on

poly(2-methacryloyloxymethyl phosphorylcholine-co-alkyl methacrylate)-

Appendix 10

278

coated polymers. Journal of Biomedical Materials Research, 1995; 29:

431-439.

15. Gu YJ, Boonstra PW, Akkerman C, Mungroop H, Tigchelaar I, van

Oeveren W. Blood compatibility of two types of membrane oxygenator

during cardiopulmonary bypass in infants. Int J Artif Organs, 1994; 17:534-

548

Appendix 10

279

Table 1. Demographic and surgical data.

Phosphorylcholine

(n = 10)

Control

(n = 10)

Variable Mean SD Mean SD p value

Age (days) 136 127 167 134 NS

BSA (m²) 0.287 0.064 0.292 0.063 NS

Anomaly

TGA 2 0

VSD 4 3

TOF 2 3

TGA/TA/PS/VSD 1 0

TA/VSD/ASD 0 1

VSD/PS/ Ebstein 0 1

DORV 1 2

Bypass time

(minutes)

91.7 35.7 94.2 23.2 NS

Cross-clamp time

(minutes)

50.6 28.9 50.6 25.7 NS

TGA: Transposition of the great arteries, VSD: ventricle septum defect, TOF:

tetralogy of Fallot, DORV: double outlet right ventricle, TA: tricuspid atresia,

PS pulmonic stenosis.

Appendix 10

280

Figure 1. Complement activation.

Pre 15' 60' End Post PO1 PO6-100

0

100

200

300

400

500

600

700

800

900

1000

1100

% D

iffer

ence

from

bas

elin

e va

lue

Terminal Complement Complex (TCC)

CoatedUncoated

Appendix 10

281

Figure 2: Platelet activation.

Pre 15' 60' End Post PO1 PO6-100

0

100

200

300

400

500

% d

iffer

ence

from

bas

elin

e va

lue

Thromboxane B2

CoatedUncoated

Pre 15' 60' End Post PO1 PO6-50

0

50

100

β-Thromboglobulin

Appendix 10

282

Appendix 11

283

Tissue factor as main activator of the coagulation systemduring cardiopulmonary bypass

F. De Somer ECCP, Y. Van Belleghem MD, F. Caes MD, K. François MD, H.

Van Overbeke MD, J. Arnout MD, PhD, Y. Taeymans MD, PhD,

G. Van Nooten MD, PhD

The Journal of Thoracic and Cardiovascular Surgery, 2002; 123: 951-958

Appendix 11

284

Abstract

Objective: This study investigates the influence of foreign material and blood

aspirated from nonvascular structures on activation of coagulation, hemolysis

and blood loss.

Methods: The series comprises three randomized groups (C, S and S+P) of

10 patients undergoing routine coronary artery bypass grafting with

cardiopulmonary bypass. In group C, the control group, all aspirated blood

was returned into the circulation. In group S suction blood was discarded

whereas Group S+P was identical to Group S, with surfaces coated with

phosphorylcholine. Plasma concentrations of β-thromboglobulin, thrombin

generation, haptoglobin and free hemoglobin, as well as blood loss, were

measured.

Results: A steady increase in free plasma hemoglobin, as well as an

increased generation of thrombin, was noticed in group C. Moreover, a close

correlation (r = 0.916) between the generation of thrombin and its inhibition

(thrombin-antithrombin complexes) was observed. Platelets were clearly

activated in group C and, to a lesser extend, in group S. In contrast, platelet

activation in group S+P was negligible, resulting in a 30% decrease in blood

loss (p=0.05).

Conclusions: Aspirated blood contaminated by tissue contact is the most

important activator of the coagulation system and the principal cause of

hemolysis during cardiopulmonary bypass. Contact with foreign surface is not

a main variable in the procoagulant effect of bypass. Mimicking the outer cell

Appendix 11

285

membrane structure resulted in decreased platelet activation and decreased

blood loss.

Ultramini-abstract

This study demonstrates that aspiration of blood from non-vascular structures

is the main activator of coagulation. The influence of foreign surface in

procoagulant activity is small. Mimicking the outer cell membrane on the

foreign surface resulted in decreased platelet activation and a significant

reduction of blood loss.

Appendix 11

286

Introduction

During cardiopulmonary bypass (CPB), blood is diverted into an

extracorporeal circulation. Those foreign surfaces exert a strong procoagulant

effect [1]. Over the years, many improvements have been made to the

components of the CPB circuit. Heparin coating and, more recently,

phosphorylcholine coating definitively reduce inflammatory response [2,3].

The characteristic feature of biological membranes is their functional and

compositional lipid asymmetry, which has been described in several cell

types. It is thought to stem from the requirement of biological membranes to

have asymmetric protein distributions across the bilayer. In all of the cells for

which lipid compositional asymmetry has been described, negatively charged

phospholipids are found predominantly on the inner cytoplasmatic side of the

membrane, whereas the neutral zwitterionic phosphorylcholine containing

antithrombotic lipids predominate in the outer membrane leaflet. Negatively

charged phospholipids are thrombogenic. This membrane asymmetry may

serve the biological purpose in the maintenance of the delicate balance

between hemostasis and thrombosis. However, reduction in activation of the

coagulation cascade and cell trauma is not conclusive [3]. This might be

related to the fact that, in most clinical studies, aspirated blood, which is

recognised as one of the most injurious components [4], is still reused. The

purpose of this study is to investigate the contribution of aspirated blood

versus foreign material in the activation of the coagulation cascade and cell

trauma.

Appendix 11

287

Materials and methods

Patients

Thirty patients were prospectively randomized into 1 of 3 groups. All patients

were subjected to elective coronary artery bypass grafting. The sole exclusion

criterion was an ejection fraction less than 40%. There were no statistical

differences in demographics and operative data between groups (Table 1).

The medical ethical committee of the hospital approved the study, and written

informed consent was obtained from all patients.

In the control group (Group C; n=10) CPB was performed in a standard

fashion, with recuperation of all suction blood into the circulation. In second

group (Group S; n=10) the same circuit was used as in the control group, but

aspirated blood collected from mediastinal cavities, pleural cavities or both

was discarded. The third Group (Group S+P; n=10) was identical to group 2

except for the coating of all foreign material with phosphorylcholine.

Operative techniques

Before cannulation, porcine heparin (300 IU/kg; Roche Pharmaceuticals,

Mannheim, Germany) was injected. Activated coagulation time (kaolin ACT;

Medtronic Hemotec, Inc, Englewood, Colo) was kept above 400 seconds

throughout CPB. CPB consisted of custom tubing pack made of polyvinyl

chloride, an arterial filter, a membrane oxygenator and an open venous

reservoir with separated cardiotomy reservoir (Dideco, Mirandola, Italy).

Circuits were identical in the different groups, with exception of Group S+P in

which all surfaces in contact with blood were coated with phosphorylcholine.

The heart-lung machine (COBE Cardiovascular, Arvada, Colo) was primed

Appendix 11

288

with a mixture of gelatine solution (Pasteur Merieux, Lyon, France), Mannitol

(Baxter Healthcare Corporation, Deerfield, Ill), 2 million KIU of aprotinin

(Bayer AG, Leverkusen, Germany) and 5000 IU of heparin (Roche, Brussels,

Belgium). Total priming volume was 1300 mL. Esophageal temperature was

lowered to 28°C. If possible, autologous blood was removed after induction,

aiming at a hematocrit level of 25% during CPB. During aortic crossclamping,

the aortic root was vented with a pressure-controlled roller pump. Myocardial

preservation during aortic crossclamping was obtained with approximately

800 mL (600 – 900) of crystalloid, antegrade, modified St. Thomas’ Hospital

cardioplegic solution.

Blood sampling

Blood samples were taken after induction, at 15 minutes of CPB, 5 minutes

after release of the aortic cross clamp, at the end of CPB, 20 minutes post

CPB and on postoperative days 1 and 2. Total blood loss was documented at

4, 8 and 12 hours postoperatively.

Laboratory assays

Serum concentrations of free hemoglobin and haptoglobin were determined

as markers of hemolysis by using immunonephelometry [5] on a BN

nephelometer (Behringwerke AG, Marburg, Germany) and expressed

according to Instructional Faculty Consortium Committee standards [6].

The prothrombin fragment (F1+2), split off during conversion of prothrombin to

thrombin, was measured on citrated plasma by using a quantitative enzyme-

linked immunosorbent assay (ELISA; EnzygnostR F1+2 micro, Behring

Diagnostics GmbH, Frankfurt, Germany). The capture antibodies in this

Appendix 11

289

sandwich ELISA are highly specific polyclonal antibodies raised in rabbits

against a synthetic peptide from the negatively charged region of F1+2

fragment. Peroxidase-conjugated rabbit anti-human prothrombin antibodies are

used as the tagging antibody. Normal level, determined in 24 healthy

volunteers, is 1,16 ± 0,39 nmol/L (range 0,5 - 2,6 nmol/L; median, 1,1 nmol/L).

Thrombin-antithrombin complexes (TATs), reflecting thrombin generation

followed by inhibition by antithrombin, were determined on citrated plasma by

means of ELISA ( EnzygnostR TAT micro, Behring Diagnostics GmbH),

according to the manufacturer’s instructions. This ELISA employs a polyclonal

antibody specific for neoantigenic determinants on thrombin as capture

antibody and peroxidase-labelled polyclonal rabbit anti-human antithrombin III

as the tag antibody. Normal TAT level, determined in 24 healthy volunteers

are 4,07 ±2,33 ng/mL (range: 2 - 14,9 ng/mL; median, 3,4 ng/ml).

β-thromboglobulin (β−TG), released from α-granules at platelet activation was

recorded with commercially available ELISA testing (Asserachrom βTG;

Diagnostica Stago, Parsippany, NJ). Normal values determined in 40 healthy

donors ranged from 15 to 42 IU/mL (mean, 24.4 IU/mL).

Statistics

The overall differences among the 3 groups were analyzed with a Kruskal-

Wallis test. The comparison between each individual group was done with a

Mann-Whitney test corrected for repeated comparisons. The sample points

were related to the progress of the operation and differed in each patient.

Comparisons at each sample point were therefore not considered relevant.

Hence, the values were treated individually for each patient, calculating the

surface under the curve representing the total release during CPB.

Appendix 11

290

The correlation between the generation of thrombin (PF 1+2) and its inhibition

(TAT) was calculated with a Spearman R test.

Appendix 11

291

Results

Coagulation

Lactic dehydrogenase

Lactic dehydrogenase (LDH) levels increased in all groups on postoperative

day 1 and 2 compared with baseline values (Figure 1). Mean total LDH

release was 296318 ± 137924 U/L/procedure in group C, 112170 ± 75153

U/L/procedure in group S (p=0.005) and 136212 ± 91602 U/L/procedure in

group S+P (p=0.01).

β-thromboglobulin

The concentration of β-TG remained stable during CPB in group S+P,

whereas an increase over time was observed in groups S and C (Figure 1). In

all groups an increase of baseline values was noted at 20 minutes after CPB.

Mean total release of β-TG during CPB was 3790 ± 4103 IU/mL/CPB in the

S+P group, 18870 ± 20479 IU/mL/CPB in group C (p=0.016) and 8040 ± 3986

IU/mL/CPB in group S (p=0.004).

Prothrombin fragment 1+2

In the control group an important increase in F1+2 levels was noted during

CPB (Figure 2) from a mean baseline value of 1.9 ± 1.8 µg/L to 5.0 ± 3.0 µg/L

at the end of CPB, which further increased to 5.4 ± 2.3 µg/L at 20 minutes

after CPB. In group S and S+P the values remained stable during and after

CPB. Mean total F1+2 during CPB was 20594 ± 21733 µg/L/CPB in group C,

2534 ± 2365 µg/L/CPB in group S (p=0.001) and 2197 ± 2095 µg/L/CPB in

S+P (p=0.001).

Appendix 11

292

Thrombin-antithrombin complex

Mean values of TAT complex decreased slightly in groups S and S+P, from

25.3 ± 42.7 nmol/L and 24.6 ± 21.9 nmol/L to 8.6 ± 8.1 nmol/L and 7.3 ± 2.2

nmol/L at 15 minutes of bypass, respectively (Figure 2). Subsequently, the

values remained stable during CPB and returned to baseline values at 20

minutes after CPB. This in opposition to group C, were the mean baseline

value of 56.9 ± 85.8 nmol/L, after a first decrease to 30.8 ± 22.4 nmol/L,

started to increase to a value of 128.0 ± 96.1 nmol/L at the end of CPB. Mean

total generation of TAT during CPB was 62926 ± 61907 nmol/L/CPB in group

C versus 4009 ± 2958 nmol/L/CPB in group S (p=0.001) and 3925 ± 1593

nmol/L/CPB in group S+P (p=0.001). Moreover, close correlation was

established between levels of TAT and F1+2 (r = 0.916; p<0.001) in group C

(Figure 3).

Hemolysis

Free plasma hemoglobin

Whereas mean free plasma hemoglobin levels remain stable in group S and

S+P, there is a steady increase from 9.9 ± 4.3 mg/dL to 46.6 ± 17.6 mg/dL at

20 minutes after CPB in group C (Figure 4). Mean total generation of plasma

free hemoglobin during the procedure was most pronounced in group C

(353800 ± 193475 mg/dL/procedure) compared with that seen in group S

(70140 ± 56462 mg/dL/procedure; p=0.001) and group S+P (130390 ± 86308

mg/dL/procedure; p=0.001).

Haptoglobin

Mean haptoglobin levels decreased in all groups over time (Figure 4).

Appendix 11

293

Blood loss and tranfusions

In group S an average of 295 ± 136 mL and in group S+P an average of 370 ±

172 mL (p=not significant) of blood was aspirated during CPB and discarded

at the end of the procedure. The hematocrit levels on postoperative day 1

were 30.6 ± 4.1 % in group C, 29.9 ± 3.0 % in group S and 32.8 ± 3.0 % in

group S+P (p=0.196).

The average blood loss during the first 4 hours postoperatively was 210 ± 80

mL (p=0.05) in group S+P, 326 ± 170 mL in group C, and 338 ± 223 mL in

group S. Blood losses between 4 and 8 hours and 8 and 12 hours

postoperatively were not statistical different between groups (Figure 5).

Dividing the total population in patients who lost more or less than 250 mL

during the first 4 postoperative hours revealed duration of CPB (p<0.001),

prolonged cross-clamp time (p=0.002) and number of bypasses (p=0.03) to

be incremental risk factors for bleeding. By using the same division with

regard to the 3 groups, a significant difference in reduced blood loss was

found in favor of group S+P (p=0.05, Table 2).

In none of the groups were blood products given during CPB. In group S no

blood products were given, whereas in group S+P 1 patient and in group C 4

patients received packed red cells postoperatively (p=0.05).

Appendix 11

294

Discussion

Despite anticoagulation with high doses of heparin during CPB, this procedure

is associated with considerable activation of the coagulation system [1]. The

important rise in F1+2 and TAT levels obtained in our control group confirms

the procoagulant effect of CPB. In addition, significant activation of blood

platelets and generation of hemolysis was observed. After unclamping, most

surgeons reinfuse blood aspirated from the mediastinal and pleural cavities.

Recirculation of suction blood is documented to decrease the mean arterial

pressure [7], to activate the coagulation cascade [8,9] and to generate

hemolysis [8,10]. As soon as blood comes into contact with tissue factor, the

coagulation cascade is activated [12,13]. As a result of surgical trauma, tissue

factor can be present in both mediastinal and pleural cavities. Therefore,

blood recuperated from these cavities will be activated, and thrombin will be

generated, leading to elevation of both TAT and F 1+2. In our intervention

groups in which reinfusion of aspirated blood was omitted, almost no rise in

TAT and F1+2 levels was observed, clearly suggesting aspirated blood to be

the main cause of thrombin generation. A discussion is ongoing whether

heparin dosing during CPB based on ACT measurement is optimal or whether

heparin by itself is an adequate anticoagulant in this setting [14-16]. Several

studies have shown a poor correlation between ACT and plasma heparin

levels as measured with an anti-Xa method [14]. However, rapid point-of-care

methods to measure heparin levels are still in the process of validation.

Therefore, in our study heparin dosing was still adjusted on the basis of the

ACT. Despotis and colleagues [15] described that a more effective

Appendix 11

295

suppression of the hemostatic system in CPB may be obtained when heparin

dosing is based on heparin blood concentrations rather than on ACT. In their

study a negative correlation was found between F1+2 and TAT levels on the

one hand and plasma heparin concentrations, as measured with an anti-Xa

method, on the other hand. In contrast to this, the study of Knudsen and

coworkers [16] clearly showed that high levels of F1+2 may be generated

during CPB, despite adequate heparin anticoagulation, as measured with a

plasma anti-Xa method. In this study, suction blood was also reinfused, and

the highest F1+2 levels were similarly to those in our study found shortly after

unclamping. The high degree of comparability of the F1+2 results of our

control group and the results reported by Knudsen and coworkers makes it

unlikely that the low F1+2 values in patients in whom no aspirated blood was

reinfused could be due to higher heparin levels. Differences in amount of

aspirated blood volume may account for the differences found in the literature.

However, blood aspirated from cavities covered with endothelium does not

activate the coagulation [12].

Destruction of red blood cells in contact with the pericardium, pleural cavities,

or both, was recognized in the early days [11]. In our control group, free

plasma hemoglobin started to increase after the release of the aortic

crossclamp, simultaneously with a steady decrease in haptoglobin levels over

time. In a recent study where aspirated blood was kept separated until the

end of CPB, a similar increase in hemolysis was noticed after reinfusion of

this aspirated blood [10]. Major hemolysis is caused by blood aspirated from

nonvascular cavities. This is most likely caused by shear forces, negative

pressure, and the blood-air interaction. The effect of mechanical destruction

Appendix 11

296

(arterial roller pump) is partially neutralized by rapid elimination of the

haptoglobin-hemopexin complexes at specific hepatic receptors. Hemolysis

generation by means of the arterial roller pump remains negligible during

short-term cardiac surgery and was confirmed by low free plasma hemoglobin

values during CPB in both retainment groups.

In addition to high circulating levels of heparin, attempts have been made to

control activation of the coagulation system by coating the foreign surface

area of the CPB. However, generation of TAT and F1+2 in most studies was

not conclusive [3].

Phosphorylcholine coating mimics the characteristic feature of biological

membranes. In vitro experiments, in which various phospholipid coatings were

applied to surfaces, showed a very high procoagulant activity of negatively

charged phospholipids. This in contrast to the absence of activation of

phosphorylcholine-containing surfaces in coagulation tests [17,18]. Blood

platelets are not only essential for the coagulation but also interfere with white

blood cell and complement activation. Platelets were activated predominantly

in group C, by means of reinfusion of damaged and activated platelets with

aspirated blood. However, also in group S, moderate platelet activation is

noticed starting over time, whereas absolutely no increase is observed in

group S+P. Better platelet preservation in group S+P is also reflected by lower

blood loss in the immediate postoperative period. The difference between

group S and group S+P can be seen as the representation of the damage

caused by the contact with untreated foreign material. This finding is in

agreement with previous observations [2].

Appendix 11

297

No statistical differences regarding duration of CPB, crossclamp time and

number of bypasses were observed between groups. Nevertheless, a

significant higher number of patients lost less than 250 mL blood in group

S+P. In the population who lost more than 250 mL during the first 4

postoperative hours, there was a positive correlation with the duration of CPB,

crossclamp time and number of bypasses, which is in agreement with

previous findings [19].

Development of a dedicated venous reservoir makes it possible to separate

aspirated blood coming from different sources. Blood from vascular structures

can be safely returned into the circulation, whereas highly activated blood

caused by contact with tissue factor can be kept separated. Depending on the

amount of blood loss, the latter can be processed with a cell salvage system

or discarded. Moreover, recent in vitro research also points out that

generation of fat emboli is negligible in groups without recuperation of the

mediastinal blood compared with that in a control group [20].

General conclusion

Retainment of blood aspirated out of nonvascular structures will significantly

reduce morbidity of CPB. Blood activated by means of tissue factor should be

discarded or processed with a cell salvage system. Phosphorylcholine coating

is not a main participant for control of the procoagulant effect of CPB but

results in decreased platelet activation and decreased blood loss.

Appendix 11

298

Limitations of the study

Because our study concerns a biological system with relatively large SDs in a

limited number of patients, our data should be interpreted with caution. Large

randomized studies are necessary to investigate the influence of reinfusion of

aspirated blood on morbidity.

Acknowledgement

We thank Sorin-Biomedica, Mirandola, Italy, for providing the

phosphorylcholine coated oxygenators, cannulas and custom packs.

Appendix 11

299

References

1. Tanaka T, Takao M, Yada I, Yuasa H, Kugasawa M, Degushi K.

Alterations in coagulation and fibrinolysis associated with cardiopulmonary

bypass during open heart surgery. J Cardiothorac Anesth 1989;3:181-88

2. De Somer F, François K, van Oeveren W, et al. Phosphorylcholine coating

of extracorporeal circuits provides natural protection against blood

activation by the material surface. Eur. J. of Cardiothoracic Surg (In press).

3. Wendel HP, Ziemer G. Coating-techniques to improve the

hemocompatibility of artificial devices used for extracorporeal circulation.

Eur J of Cardiothoracic Surg 1999;16:342-50.

4. Malinauskas RA, Sade RM, Dearing JP, Spinale FG, Crawford FA, von

Recum AF. Blood damaging effects in cardiotomy suction return. The

Journal of Extra-Corporeal technology 1988;20:41-6.

5. Fink PC, Römer M, Haeckel R, Fateh Moghadam A, et al. Measurement of

proteins with the Behring Nephelometer. A multicentre evaluation. J. Clin.

Chem. Clin. Biochem. 1989;27:261-76

6. Johnson AMA. A new international reference preparation for proteins in

human serum. Arch. Pathol. Lab. Med. 1993;117:29-31.

7. Lavee J, Naveh N, Dinbar I, Shinfield A, Goor DA. Prostacycline and

Prostaglandin E2 mediate reduction of increased mean arterial pressure

during cardiopulmonary bypass by aspiration of shed pulmonary venous

blood. J Thorac Cardiovasc Surg 1990;100:546-51.

Appendix 11

300

8. de Haan J, Boonstra PW, Monnink SHJ, Ebels T, van Oeveren W.

Retransfusion of Suctioned Blood During Cardiopulmonary Bypass Impairs

Hemostasis. Ann Thorac Surg 1995;59: 901-7.

9. Walpoth BH, Eggensperger N, Hauser SP, et al. Effects of unprocessed

and processed cardiopulmonary bypass blood retransfused into patients

after cardiac surgery. Int J Artif Organs 1999;22:210-16.

10. Hansbro SD, Sharpe DAC, Catchpole R, et al. Hemolysis during

cardiopulmonary bypass: an in vivo comparison of standard roller pumps,

nonocclusive roller pumps and centrifugal pumps. Perfusion 1999;14:3-10.

11. Boisclair MD, Lane DA, Philippou H, Sheikh S, Hunt B. Thrombin

production, inactivation and expression during open heart surgery

measured by assays for activation fragments including a new ELISA for

prothrombin fragment F1+2. Thrombosis and Haemostasis

1993;70(2):253-58.

12. Boisclair MD, Lane DA, Philippou H, et al. Mechanisms of thrombin

generation during surgery and cardiopulmonary bypass. Blood

1993;82:3350-57.

13. Morris KN, Kinross FM, Stirling GR. Hemolysis of blood in the pericardium:

the major source of plasma hemoglobin during total body perfusion. J

Thoracic and Cardiovas Surg. 1965;49:250-58

14. Niles SD, Sutton RG, Ploessl J, Pennell B. Correlation of ACT as

measured with three commercially available devices with circulating

heparin level during cardiac surgery. J Extra Corpor Technol.

1995;27:197-200.

Appendix 11

301

15. Despotis GJ, Joist JH, Hogue CW Jr, Alsoufiev A, Joiner-Maier D, Santoro

SA, Spitznagel E, Weitz JI, Goodnough LT. More effective suppression of

hemostatic system activation in patients undergoing cardiac surgery by

heparin dosing based on heparin blood concentrations rather than ACT.

Thromb Haemost. 1996;76:902-8.

16. Knudsen L, Hasenkam JM, Kure HH, Hughes P, Bellaiche L, Ahlburg P,

Djurhuus C. Monitoring thrombin generation with prothrombin fragment 1.2

assay during cardiopulmonary bypass surgery. Thromb Res. 1996; 84: 45-

54.

17. Zwaal RFA, Hemker HC. Blood Cell Membranes and Haemostasis.

Haemostasis 1982;11:12-39.

18. Yianni YP. Biocompatible surfaces based upon biomembrane mimicry. In:

Quinn PJ and Cherry RJ, editors Structural and dynamic properties of

lipids and membranes. London: Portland Press Ltd, 1992: 182-217.

19. Tuman KJ, McCarthy RJ, O’Connor CJ, McCarthy WE, Ivankovitch AD.

Aspirin does not increase allogeneic blood transfusions in reoperative

coronary artery surgery. Anesth Analg 1996;83(6):1178-84

20. Brooker RF, Brown WR, Moody DM, et al. Cardiotomy Suction: A Major

Source of Brain Lipid Emboli During Cardiopulmonary Bypass. Ann Thorac

Surg 1998;65:1651-55.

Appendix 11

302

Table 1: Demographic data and operative details of the studied groups a

Control (n=10)

(Group C)

Suction (n=10)

(Group S)

Coated (n=10)

(Group S+P)

Clinical

parameter

Mean ± SD Mean ± SD Mean ± SD

Age (years) 62 ± 8 67 ± 11 64 ± 10

Female/male 4/6 1/9 0/10

Weight (kg) 81 ± 14 77 ± 9 83 ± 13

Number of distal

anastomoses

3.9 ± 0.9 3.5 ± 0.7 3.8 ± 0.6

Extracorporeal

time (min)

91 ± 26 80 ± 15 78 ± 16

Aortic cross

clamp time (min)

50 ± 16 39 ± 7 43 ± 10

a All parameters were analyzed by unpaired Student’s t-test and showed no

significant difference between the three study groups.

SD = Standard deviation

Appendix 11

303

Table 2 Blood loss during the first 4 postoperative hours.

< 250 mL > 250 mL Total

Group C

Group S

Group S+P

Total

4

5

9

18

6

5

1

12

10

10

10 *

30

* p=0.05

Appendix 11

304

Figure 1. LDH and β-TB levels. Pre, Blood samples taken after induction; PO1

and PO2, blood samples taken on postoperative days 1 and 2; XC, blood

samples taken after release of the aortic crossclamp; post, blood samples

taken 20 minutes after CPB.

Prae PO1 PO20

200

400

600

800

1000

1200

U/L

LDH

Group CGroup SGroup S+P

C vs S: p=0.005C vs S+P: p=0.01

Prae XC post PO2-50

0

50

100

150

200

% d

iffer

ence

from

bas

elin

e

β-Thromboglobulin

S+P vs S: p=0.004S+P vs C: p=0.016

Appendix 11

305

Figure 2. TAT and prothrombin 1+2 levels. Pre, Blood samples taken after

induction; PO1 and PO2, blood samples taken on postoperative days 1 and 2;

XC, blood samples taken after release of the aortic crossclamp; post, blood

samples taken 20 minutes after CPB.

prae 15 XC end post PO1 PO2-100

0

100

200

300

400

500

600

700

800

900

1000

1100

1200

1300

1400

% d

iffer

ence

from

bas

elin

e

TAT

Group CGroup SGroup S+P

C vs S: p=0.001C vs S+P: p=0.001

prae 15 XC end post PO1 PO2-50

0

50

100

150

200

250

300

350

400

450

% d

iffer

ence

from

bas

elin

e

Prothrombin 1+2

C vs S: p=0.001C vs S+P: p=0.001

Appendix 11

306

Figure 3. Correlation between TAT and F1+2 in group C.

0 50 100 150 200 250 300 350TAT (nmol/L)

0

2

4

6

8

10

PF 1

+2 (µ

g/L)

Correlation TAT - PF 1+2 in group C

r=0.916p<0.001

Appendix 11

307

Figure 4. Free plasma hemoglobin and haptoglobin levels. Pre, Blood

samples taken after induction; PO1 and PO2, blood samples taken on

postoperative days 1 and 2; XC, blood samples taken after release of the

aortic crossclamp; post, blood samples taken 20 minutes after CPB.

Prae 15 XC end post PO10

20

40

60

mg/

100

mL

Free plasma haemoglobin

Group CGroup SGroup S+P

Prae 15 XC end post PO1-60

-40

-20

0

% d

iffer

ence

from

bas

elin

e

Haptoglobin

C vs S: p=0.001C vs S+P: p=0.001

Appendix 11

308

Figure 5. Blood loss.

0 - 4 4 - 8 8 - 12 Total Blood LossHours postoperatively

0

200

400

600

800

Bloo

d Lo

ss (m

L)

Group CGroup SGroup S+P

* p=0.05

Nederlandse samenvatting

309

Nederlandse samenvatting

Sinds het eerste gebruik van de hart-longmachine voor totale

cardiopulmonale bypass op 5 april 1951, is veel veranderd. Waar de eerste

twee patiënten de procedure niet overleefden, is vandaag de mortaliteit

veroorzaakt door de cardiopulmonale bypass zo goed als onbestaande. De

enorme circuits met een bellen- of film-oxygenator, welke verschillende liters

vulvloeistof vereisten, werden vervangen door kleine membraanoxygenatoren

ingebouwd in circuits die slechts gebruik maken van een paar honderd

milliliters vulvloeistof.

Naarmate meer en meer procedures werden uitgevoerd nam de kennis en de

langetermijnopvolging van patiëntjes met congenitale hartgebreken toe.

Gebaseerd op deze nieuwe inzichten worden tegenwoordig meer en meer

kinderen zeer vroeg, met name in de eerste dagen of weken van hun leven,

chirurgisch behandeld daar dit leidt tot een betere langetermijnoverleving.

Nochthans wordt de chirurg geconfronteerd met tal van technische

beperkingen wanneer hij een vroeggeborene van pakweg 2 kilo op

cardiopulmonale bypass dient te plaatsen. Derhalve is verder onderzoek

nodig om schade veroorzaakt door de cardiopulmonale bypass aan de vaak

nog immature organen, tot een absoluut minimum te beperken.

Een eerste probleem is de vasculaire toegang. De anatomisch kleine

bloedvaten van een kind dienen te worden gecanuleerd zonder deze te

obstrueren of de wanden te beschadigen. Wat is het beste ontwerp om dit te

realiseren? Hoe kan men bereiken dat alle organen worden bevloeid, dat het

hart niet wordt blootgesteld aan een bijkomende nabelasting en dat de

Nederlandse samenvatting

310

volledige veneuze terugvloei naar de cardiopulmonale bypass wordt

afgevoerd? Appendix 1 belicht de beperkingen en voordelen van vacuum

geassisteerde veneuze terugvloei (VAVD) wanneer deze wordt gebruikt bij

kinderen. VAVD maakt het mogelijk de veneuze terugvloei met ongeveer 10%

te verhogen door middel van een groter drukverval. Daarnaast laat het

gebruik van VAVD ook toe kleinere canules te gebruiken, waardoor het

bloedvat minder geobstrueerd wordt en de vaatwand minder beschadigd. Een

groter operatieveld en minder terugvloei uit collaterale bloedvaten kan worden

bekomen door de combinatie van kleinere canules en VAVD.

Het ontwerp van de canule is zeer belangrijk voor een optimale re-infusie van

bloed in de aorta. Appendix 2 legt de relatie uit tussen canule-ontwerp en het

onstaan van jets, terwijl appendix 3 de beperkingen van bestaande

pediatrische canules belicht. Zo werden grote verschillen in druk-debiet

relaties aangetoond veroorzaakt door afwijkingen in binnendiameter en het

ontwerp van de canule.

De oxygenator blijft een probleem apart in het pediatrische cardiopulmonale

bypass circuit door zijn vulvolume, het relatief groot oppervlak aan

lichaamsvreemd materiaal en de niet altijd optimale vloeistofdynamica. Deze

problemen worden deels veroorzaakt door het feit dat de meeste, zo niet alle,

pediatrische oxygenatoren in feite verkleinde volwassen oxygenatoren zijn,

die niet aangepast zijn aan de noden van neonatale procedures. Appendix 4

somt de voordelen op van een oxygenator die specifiek voor de neonatale

cardiopulmonale bypass werd ontworpen. Het gebruik van een dergelijke

oxygenator maakt het mogelijk kleinere circuits samen te stellen, waardoor er

minder bloedverdunning optreed. Appendix 5 toont de klinische impact van

Nederlandse samenvatting

311

een dergelijke neonatale oxygenator in combinatie met een klein circuit op het

verbruik van bloedproducten. Om een optimaal massatransport en een goede

hemocompatibiliteit te bekomen is de vloeistofdynamica in een oxygenator

vitaal. Appendix 6 stelt een nieuwe techniek voor om de druk-debiet relatie

van oxygenatoren met een verschillend ontwerp te vergelijken. Deze aanpak

maakt het mogelijk om objectieve beslissingen te nemen, wanneer men

verschillende producten vergelijkt. De impact van de nieuwe ELF

membraanoxygenatoren op bloedelementen werd bestudeerd in appendix 7.

Het gebruik van een arteriële filter in een pediatrisch circuit kan in vraag

gesteld worden, daar dit enkel het vulvolume zal vergroten zonder dat hierbij

de veiligheid wordt verhoogd. Appendix 8 suggereert dat de hollevezelbundel

van het membraancompartiment een aanvaardbaar alternatief zou kunnen

zijn voor een arteriële filter daar deze een dieptefilter is voor partikels en in

staat is gasembolen te verwijderen. Terzelfdertijd, zal dit alternatief het

vulvolume van het totale circuit reduceren zonder impact op de veiligheid.

Het controleren van het inflammatoire antwoord is een van de belangrijkste

doelstellingen van het pediatrische team. Het behandelen van alle

lichaamsvreemde oppervlakken met een coating die het buitenmembraan van

de rode bloedcel nabootst, leidt tot een vermindering van de complement

activatie en een betere bescherming van de bloedplaatjesfunctie. Dit wordt

beschreven in appendix 9 en bevestigd in de kliniek in appendix 10.

Ongelukkigerwijs kan deze coating het inflammatoire antwoord niet volledig

blokkeren en dit kan wellicht verklaard worden door de bevindingen in

Nederlandse samenvatting

312

appendix 11, die aantonen dat bloed afkomstig uit ruimtes die niet niet bedekt

zijn met endotheel, zoals het pericard en de pleuraholtes, de stolling activeert.

Secundair zal dit leiden tot activatie van de complementcascade en tot een

verhoogde permeabiliteit van de vaatwand.

Klinische implicaties en toekomstige ontwikkelingen

Meer en meer pasgeborenen met een congenitaal hartgebrek worden

geopereerd tijdens de eerste dagen of weken van hun leven. Als gevolg

hiervan is het lichaamsgewicht vaak erg laag en zijn de anatomische

structuren klein. Het opstarten van de cardiopulmonale bypass onder

dergelijke omstandigheden vraagt specifieke canules met een minimale

afwijking van de vooropgestelde binnendiameter. Om onder alle

omstandigheden een optimale veneuze terugvloei en arteriële re-infusie te

realiseren dienen meer ontwerpen en kleinere diameters te worden

ontwikkeld. Druk-debiet diagrams vertrekkende van visceuze vloeistoffen

zoals water-glycerine zouden deze nieuwe ontwerpen dienen te vergezellen.

Vacuum geassisteerde veneuze terugvloei in combinatie met specifieke

veneuze canules zal het totale vulvolume van het cardiopulmonale circuit

verder verkleinen, en even belangrijk het “dood volume” in zuigerlijnen

verminderen. Als gevolg hiervan zal het bloed blootgesteld worden aan een

kleinere hoeveelheid lichaamsvreemd materiaal en zal er minder verdunning

optreden van de stollingseiwitten en de bloedelementen. Door deze minder

uitgesproken verdunning zal het gebruik van bloedproducten afnemen

Nederlandse samenvatting

313

waardoor verhinderd wordt dat het patiëntje aan meerdere bloeddonoren

wordt blootgesteld.

De behandeling van alle lichaamsvreemd materiaal met een

hemocompatibele coating zal het inflammatoire antwoord beter controleren.

Nieuwe ontwikkelingen dienen te gebeuren in

1. membraantechnologie: microporeuze versus diffusiemembranen

2. oppervlaktebehandeling van alle lichaamsvreemde materiaal

3. Integratie van de componenten en verdere miniaturisatie van de

cardiopulmonale bypass om het vulvolume en de hoeveelheid

lichaamsvreemdmateriaal verder te verminderen

4. Vloeistofmechanica van het volledige cardiopulmonale bypass circuit

gecombineerd met extensieve modellering van de vloeistofmechanica van

elke afzonderlijke component.

5. Cannules en de fysische en biologische aspecten van vasculaire toegang

in het algemeen.

6. De selectieve behandeling van geactiveerd bloed.

Nederlandse samenvatting

314

Dankwoord

315

Dankwoord

Wanneer ik, met dit werk in de hand, terugblik op de voorbije jaren komen

herinneringen aan enkele griekse sagen in me op. Net als de personages in

deze sagen, kende ik in de periode waarin dit werk tot stand kwam,

momenten van zowel kommer en kwel, als van intens geluk. Op mijn weg

kwam ik vele boeiende mensen tegen, welke allen een stempel op mij hebben

gedrukt. Elk van hun ben ik dan ook bijzonder dankbaar voor hun bijdrage

aan mijn werk. Dit eindresultaat kon immers enkel tot stand komen dankzij

hun hulp, ervaring en wijze raad.

Graag wil ik enkele personen in het bijzonder vernoemen. Mijn heel speciale

dank gaat uit naar mijn beide promotores. Professor doctor T. Ebels leerde ik

waarderen als een heel beminnelijk man, met een open en analytische geest.

Hij was steeds beschikbaar om grote en kleine problemen van de baan te

helpen alsook om advies te geven bij de verschillende ontwikkelingsfases van

mijn thesis. Zowel hijzelf, als de faculteit Medische wetenschappen van de

Rijksuniversiteit Groningen zullen steeds een aparte plaats in mijn hart

bekleden. Professor doctor G. Van Nooten ben ik zeer erkentelijk voor de

mogelijkheden die hij aanbracht. Dank zij hem heb ik dit werk aangevat en

door zijn enthousiasme, wetenschappelijke kennis en kritische bemerkingen

heb ik het kunnen beëindigen. Ik zie hem dan ook als mijn professionele

mentor en hoop nog vele jaren met hem te mogen samenwerken.

Als clinicus ontbrak het mij, zeker in de beginfase, aan inzicht in de

theoretische achtergronden van de biomedische ingenieurstechnieken. Dank

zij mijn co-promotor Professor doctor P. Verdonck en zijn uitzonderlijke groep

Dankwoord

316

werd dit gemis snel verholpen. Peter Dierickx en Dirk De Wachter leerden mij

het nut van mathematische modellering inzien als hulpmiddel voor het

begrijpen van complexe problemen en waren steeds bereid om mij met woord

en daad bij te staan.

Professor doctor Taeymans ben ik erkentelijk voor de originele manier waarop

hij de min- en plus-punten van statische analyses aanbracht.

Mijn collega’s, Dirk, Martin, Daniel, Patrick en Kurt, ben ik dankbaar voor het

klankbord dat zij mij boden gedurende de vele discussies, waardoor mijn

gedachten zich konden ordenen en ik misstappen tot een minimum kon

beperken. Ook hun bereidheid om een deel van mijn klinische taken over te

nemen wanneer ik met experimenten bezig was heb ik ten zeerste weten te

waarderen.

Mijn paranimfen, Dirk en Ton, waren het peper en het zout. Zij zorgden voor

de broodnodige structuur, terwijl hun optimisme en humor de meest efficiënte

medicamenten waren tegen zwartgalligheid.

Naast hen die ik speciaal vernoemd heb, zijn er nog vele anderen wie ik

erkentelijkheid verschuldigd ben en die ik bij deze hartelijk wil danken.

Ondanks al deze hulp denk ik niet dat ik dit werk had kunnen voltooien zonder

de liefde en morele steun van Caroline en Casper. Ik dank hen voor het

begrip en de zelfopoffering die zij zo lange tijd hebben opgebracht en ik wil dit

werk dan ook heel speciaal aan hen opdragen.

Als laatste, maar zeker niet als minste wil ik ook mijn ouders vermelden. Dank

zij hun opvoeding en de mogelijkheden die zij me gegeven hebben ben ik tot

dit punt geraakt.

Curriculum vitae

317

Curriculum vitae

De schrijver van dit proefschrift werd geboren op 10 mei 1960 te Aalst, België.

In 1978 haalde hij het getuigschrift van Hoger Secundair Onderwijs, richting

Latijn-wetenschappen, evenals het bekwaamheidsdiploma dat toegang

verleent tot het Hoger Onderwijs aan het Sint Jozefcollege te Aalst.

Aansluitend, volgde hij de opleiding tot gegradueerde verpleegkundige aan

het Sint Augustinusinstituut, eveneens te Aalst, alwaar hij afstudeerde in

1981. In 1990 beëindigde hij het postgraduaats Opleidingsprogramma voor

perfusionist aan de Universitaire Ziekenhuizen Leuven en bekwam hij het

getuigschrift “Erkend Klinisch Perfusionist” (EKP) van de Belgische

Vereniging voor Extracorporale Technologie (BelSECT). Dit werd aangevuld

met een European Certificate in Cardiovascular Perfusion in 1995, uitgereikt

door de European Board of Cardiovascular Perfusion (ECCP).

Na een kort verblijf als verantwoordelijke voor peritoneaal dialyse in het OLV

ziekenhuis te Aalst, start hij in 1981, in hetzelfde ziekenhuis, zijn loopbaan als

perfusionist. In 1988 wordt hij ad interim hoofdperfusionist in het Academisch

Ziekenhuis van de Vrije Universiteit Brussel (VUB). Sinds 1989 tot heden is hij

hoofdperfusionist in het Universitair Ziekenhuis Gent.

Hiernaast fungeert hij als gastdocent in de Katholieke Hogeschool Sint

Lieven, in de “Postgraduaatsopleiding tot perfusionist” van de Universitaire

Ziekenhuizen Leuven en in het Institute for Biomedical Technology (IBITECH)

van de Universiteit Gent.