Preload sensitivity determination of

Jarvik 2000 and Levitronix CentriMag ventricular assist devices in a dynamic

mock loop set-up

Marco Dat

BMTE 09.39

October 2009

Supervisors:

dr.ir. M.C.M. Rutten (Biomedical Engineering, Technical University of Eindhoven)

Prof. dr. mr. dr. B.A.J.M. de Mol (Academical Medical Center, Amsterdam)

E. Tüzün, MD. (Texas Heart Institute, Cardiovascular Surgery Lab)

1

Abstract With increasing length of waiting lists for heart transplantation, more research has to

be performed for other possibilities. One of these possibilities is a ventricular assist

device (VAD), which can be used to unload the ventricle by mechanically pumping

blood from the ventricle to the aorta. These VADs still need more improvement for

common use. This study is performed to study differences in pump characteristics of

the Jarvik 2000 and the Levitronix CentriMag. The VADs are placed in a mock loop

mimicking the systemic circulation. This mock-loop can produce physiological

pressures and flows. By changing preload and afterload, different scenarios can be

created.

Harmonic frequency analysis showed a more independent gain for the Jarvik 2000

LVAD for an increase in harmonic frequency. The Levitronix CentriMag had a lower

gain, for the lower frequency of 1 Hz, of 17.9 mmHgL-1

min-1

+/- 2.5 vs. 21.2 mmHgL-

1min

-1+/-1.6 (p<0.05) for the Jarvik 2000 LVAD, but had a higher gain for the higher

frequency of 3 Hz of 33.6 mmHgL-1

min-1

+/- 6.0 versus 28.7 mmHgL-1

min-1

+/-3.3 for

Jarvik 2000 LVAD. Jarvik also shows a pump speed dependency for the gain for the

first and second harmonic.

Unfortunately, not enough data was acquired to perform a comparison with the data

published by others.

2

3

Table of contents

Abstract ................................................................................................................... 1

Table of contents ..................................................................................................... 3

Chapter 1: Introduction.......................................................................................... 5

Chapter 2: Materials and Methods......................................................................... 7 2.1 Jarvik 2000 LVAD .......................................................................................... 7

2.2 Levitronix CentriMag LVAD........................................................................... 8

2.3 Mock-loop ....................................................................................................... 8

2.3.1 Mock-loop hardware ................................................................................. 8

2.3.2 Mock-loop software ................................................................................ 10

2.4 Simulations.................................................................................................... 11

2.5 Analysis......................................................................................................... 12

Chapter 3: Results ................................................................................................. 13

Chapter 4: Discussion............................................................................................ 20

Chapter 5: Conclusion .......................................................................................... 22

References.............................................................................................................. 23

4

5

Chapter 1: Introduction

Today, heart disease is a major cause of death. In the case of congestive heart failure,

if medical treatment is insufficient, a possible solution to help these patients is total

heart transplantation. However, there is a large waiting list for these patients. In the

United States alone, 3,529 people were on this list in 2003, 16% of which died before

they received a donor heart1. In some cases, where medical therapy is insufficient to

keep the patient alive until heart transplantation, a possible solution to lower mortality

before transplant, is using a Ventricular Assist Device (VAD) as a bridge-to-

transplant therapy. A VAD is supposed to unload the heart by mechanically pumping

blood from the ventricle to the pulmonary or systemic circulation.

This should improve the condition of the patient, so the patient has more time to wait

on a transplant2. A research on mortality among VAD-selected patients, performed in

2001 by Frazier et al3, shows that 29% of the VAD-treated patients (82/280) died

before receiving a transplant, compared with 67% of the control group (32/48) (P <

.001). The other 71% of the VAD–treated patients (198/280) survived: 67% (188/280)

ultimately received a heart transplant and 4% (10/280) had the device removed. One-

year post-transplant survival of VAD–treated patients was significantly better than

that of the control group (84% [158/188] vs. 63% [10/16]).

Research for using the VAD as a bridge-to-recovery is also promising. In that case,

the improvement of the cardiac function is sufficient during VAD support to allow

removal of the device, without replacing the native heart. Research on bridge-to-

recovery performed by Farrar et al4, shows survival of 20 of the 22 patients treated

with bridge-to-recovery VADs. Three patients required heart transplantation, 1 within

1 day after removal of the VAD, 2 at 12 and 13 months post-weaning, and 2 died at

2.5 and 6 months. The remaining 17 patients are alive with their native hearts after an

average of 3.2 years (range, 1.2-10 years). The survival of native hearts (transplant-

free survival) post-VAD support is 86% at 1 year and 77% at 5 years, which was not

significantly different (p = 0.94) from that of post-VAD transplanted patients, also at

86% and 77%, respectively. So, bridge-to-recovery is promising, but still needs a lot

of research for it to be clinical applicable.

Another possibility is destination therapy, in which the VAD is not explanted. In

research performed by Stevenson et al5, a six months survival of the population was

60% with LVAD compared to 39% without. At one year survival this was 49% for the

LVAD group compared to 24% without LVAD. By two years survival with LVADs

was 28% compared with 11% for current medical therapy. Mancini et al.6 have

recently demonstrated that LVAD patients can achieve a near-normal exercise

response, equivalent to that of patients with mild heart failure, and Dew et al.7 have

shown that patients with a LVAD enjoy a quality of life that is comparable to that of

transplant recipients.

Despite these promising researches, these VADs can still be improved to lower

morbidity and mortality. Performing research is difficult because most patients with a

VAD are in a poor condition, and are difficult to compare with each other because the

pumps are implanted in patients with different conditions. For these reasons, a mock-

loop, producing physiological pressures and flows8, is used in this research. Only the

systemic circulation is simulated because this is where most VADs are used for,

because usually the left ventricle fails before the right ventricle does, due to a higher

workload 9

. Using this mock-loop, the pump characteristics can be determined. In this

research the characteristics of the Jarvik 2000 Left Ventricle Assist Device (LVAD)

and Levitronix CentriMag LVAD are investigated. These are both continuous flow

6

VADs, which are pressure sensitive10

, this means that at a constant VAD speed, the

output varies depending on pressure differential (∆p) between pump inlet and outlet.

However, this sensitivity differs among different types of VADs. Because the ∆p

varies throughout the cardiac cycle, this pressure sensitivity determines the level of

unloading at different ∆p. When ∆p across the pump is high (end systole to early

diastole), the pump output will be reduced. If a VAD is very sensitive to an increased

∆p, it will produce less output at this moment in the cardiac cycle, the moment where

ventricular volume is low. With low ventricular volume the risk at intraventricular

suction is larger. So, high preload sensitivity gives a small output, possibly preventing

intraventricular suction.

The overall goal of this study is to investigate if there is a difference in pump

characteristics and if the preload sensitivity differs in the Jarvik 2000 LVAD and

Levitronix CentriMag LVAD in a dynamic mock-loop set up for different LVAD

speeds in different simulated conditions. Attention will also be paid to higher

harmonics of the cardiac cycle frequency.

7

Chapter 2: Materials and Methods

Two continuous flow LVADs were tested in this research. First the Jarvik 2000

LVAD will be discussed. After that, the Levitronix CentriMag LVAD will be

discussed. The pumps characteristics were investigated using a mock-loop, explained

after the LVAD discussion. Finally, protocols used and analyses performed are

discussed.

2.1 Jarvik 2000 LVAD

The Jarvik 2000 LVAD (Jarvik Heart, Inc., New York, US) (figure 4) is a continuous,

axial, valveless and intracorporeal flow device, first implanted in 200011, 12

. The major

difference with other LVADs is this LVAD is placed in the ventricular wall (figure

4a). This is possible because it is smaller than other LVADs. The advantages are less

infection, less discomfort by noise and weight12

. The pump speed can be easily

adjusted (5 speeds, analogue) by the patient to suit the patients level of activity. With

a small controller, the patient also has the possibility to be mobile. For this reason the

Jarvik 2000 can be used as destination therapy and bridge to recovery, instead of only

bridge to transplant. Another advantage of its small size is the possibility to treat

smaller adults and large children12

. The Jarvik 2000 can be powered for 8-10 hours on

a single rechargeable Lithium-ion battery pack that weighs less than one kilogram and

the battery can be recharged at any wall outlet12

. Jarvik 2000 can be implanted

without using the heart-lung machine13

, resulting in less risk.

Figure 4: a) Jarvik 2000 implanted in the ventricle. 4b) components of Jarvik 2000

8

Figure 5: a) Levitronix console (back), motor, with on top, the rotary blood pump. B) Set-up to patient

2.2 Levitronix CentriMag LVAD

The Levitronix CentriMag (Waltham, Massachusetts, US) (figure 5) is a continuous-

flow, centrifugal-type rotary blood pump that is placed extracorporeal. The only

moving component within the pump is the impeller, which is magnetically levitated

and rotated in a contact-free manner. The pump can rotate up to 5500 rpm and can

provide flow rates of up to 9.9 liters per minute. The Levitronix CentriMag pump

causes very little damage to the blood because it does not contain any bearings or seal,

components that are known to cause hemolysis and promote thrombus formation9.

There are no possibilities for mobility for the patient because of the extracorporeal

placed pump (figure 5b) making it less usable for bridge to recovery and destination

therapy. Advantage of this pump involves early operative intervention and

implantation of biventricular support. This strategy avoids urgent placement of

expensive long-term ventricular assist devices in hemodynamically unstable patients

with multisystem organ failure whose neurological status is uncertain until end-organ

recovery and excellent hemodynamic stability is achieved with the relatively

inexpensive short-term CentriMag circulatory support system14

.

2.3 Mock-loop

A mock loop (HemoLab, Eindhoven, Netherlands) has been used to perform

measurements on the LVADs. This mock-loop was set up at the department of

cardiovascular research, Texas Heart Institute at St. Luke’s Episcopal Hospital,

Houston.

2.3.1 Mock-loop hardware

The mock-loop hardware (fig.1) representing the systemic circulation is build out of

different compartments. A cylinder is used to represent the left ventricle, with a

possibility to connect an LVAD at the bottom. Volume changes in the ventricle are

caused by displacement of fluid by the piston. The piston is driven by a servomotor,

which converts electricity into movement of the piston. A computer is connected with

the servomotor to establish movement (speed and amplitude) of the piston according

to the software.

The ventricle has an aortic and mitral valve, to displace fluid in one direction. A

Polyurethane tube, with a similar compliance as an aorta, is connected to the aortic

valve. The aorta is connected to a compliance chamber, representing the peripherals.

9

Preload and afterload can be adjusted with the adjustable resistances on the

compliance chamber. The compliance chamber is connected with a PVC tube,

representing the vena cava, returning fluid back to the atrium, which is connected to

the left ventricle, closing the loop. The closed loop is filled with a fluid, which should

represent blood. The used fluid was a solution of 35% glycerol in water at room

temperature, which had an approximate viscosity of 4 mPa·s15

, similar to blood

viscosity.

A T-connection, (fig. 2) was placed between aortic valve and aorta to connect LVAD

outflow to.

A data acquisition system was connected to probes to measure flows and pressures

Figure 1: components of mock-loop, without a LVAD connected

Figure 2: T-connection between aortic valve and aorta to connect outlet of LVAD

10

To control the servomotor, a connector box was needed to connect the computer with

a motion controller board.

Full details of the procedure of connecting the parts of this mock-loop can be found in

the Hemolab manual16

.

Flow was measured with a flow probe on the aorta, to measure total flow, a flow

probe was placed on the outlet tube of the LVAD to measure LVAD flow. Both

signals were sampled with a frequency of 1000 Hz. Total flow minus LVAD flow

gives aortic flow. Pressure is measured in the aorta with a pressure wire through a

small opening in the T-connection. Pressure in the ventricle was measured with a

pressure wire through a small opening in top of the ventricle.

2.3.2 Mock-loop software

To control the displacement of the piston, it is connected to a computer with Hemolab

software. The software needs an input file (figure 3) with normalized piston

coordinates. 100 sample points led to a cycle length of 1 second. So the input file

contains the information about the total cycle length and the rate of volume change.

By setting the gain in the software, the amplitude of a cycle can be set. Maximum

gain was 15000 counts, which gave a stroke volume ~55 ml (measured with a

calibrated flow probe), at an aortic pressure of 120/80 mmHg. The servomotor was

unable to generate higher stroke volumes at a cycle rate of 60 cycles/minute.

The input file used in this research creates a volume change in the artificial ventricle

as in a normal left ventricle. Notice that this input file is inverted with volume change.

When displacement is zero, artificial ventricle volume is at its maximum.

The program also had possibilities to measure the signals of flow and pressure, but

these are not used. Instead a special data acquisition system is used to store all signals.

Figure 3: normalized signal for the displacement for the piston, as input for the servomotor.

11

2.4 Simulations

Similar simulations were performed for the Jarvik 2000 and the Levitronix CentriMag

with different settings for the mock-loop at different LVAD speeds.

For all the simulations a baseline was recorded, to simulate a situation before LVAD

implantation. This was done by clamping the outlet tube of the LVAD, which let to

zero flow behind the clamp. For the Jarvik 2000 simulations, all speeds, (zero (off) to

speed 5) were used for simulations. For the Levitronix CentriMag also, a baseline was

recorded, and simulations were performed at 0, 1000, 1500, 1800, 2000, 2200 and

2500 rpm. Higher rotation speeds are impossible due to mock loop limitations. Most

simulations were done at two amplitudes of displacement of the piston; 15.000 counts

(3.6 cm) and 10.000 counts (2.4 cm), with the same settings for peripheral resistance

as in the 15.000 counts (cts) simulations.

The pressure signals were filtered using a 4th order low pass digital Butterworth filter

with a cutoff frequency of 5 Hz. According to Milnor17

it is not necessary to include

higher frequencies. The flow signal is already filtered by the flow sensor transducer,

so no further filtering is needed.

Figure 6: measured pressures (top) and flows (bottom) in healthy set-up with a blood pressure of

120/80mmHg and a cardiac output of 3.4 liter/minute for the Jarvik 2000.

Several simulations were performed to simulate different scenarios in which blood

pressure (preload and afterload), stroke volume or heart rate differ.

The first simulation was a healthy blood pressure (120/80 mmHg) and an aortic flow

(mean 3.4 L/min, higher was mechanically impossible) at a cycle frequency of 1/60

Hz (figure 6). With these settings for the compliance chamber and the peripheral

resistance the same simulation was performed with lower amplitude of 10.000 cts,

resulting in a cardiac output of 2.5 L/min and lower aortic pressure of 80/40 mmHg.

This was done to simulate a lower cardiac output with the same parameters for

12

peripheral resistance. This way, the effect of an LVAD to low cardiac output could be

measured.

The same simulations were performed for other pressures, high blood pressure of

160/100 mmHg, medium high blood pressure 145/90 mmHg and low blood pressure

of 80/50 mmHg. All these simulations were done at a cycle frequency of 1/60 Hz for

15K cts, and with the same parameter setting for the peripheral resistances, at a piston

replacement of 10K cts.

Similar simulations were performed with a normal blood pressure of 120/80 mmHg at

a cycle frequency of 1/60 Hz set at 15K cts, but now at different displacements of the

piston. Piston displacements of 12K and 8K cts were used.

A simulation with a normal pressure of 120/80 mmHg is done with 12K cts

displacement of the piston, at a cycle frequency of 1/60 Hz, and a pressure of 120/80

for 8K cts. This should simulate a normal blood pressure at a lower stroke volume.

Simulations were also performed for a higher heart rate of 80 beats per minute.

Normal blood pressure was used and piston displacement was 15K cts. With the same

resistance settings, a simulation was also performed with a piston displacement of

10K cts.

2.5 Analysis

The acquired data were analyzed with Matlab R2008b (The MathWorks, Inc., Natick,

USA). Figure 7 to 10 were created for explorative data analysis, to see the effect of

the LVADs on aortic and ventricle pressure and flow. For the blood pressure signals a

4th order low pass digital Butterworth filter with a cutoff frequency of 5 Hz17

. is used,

as mentioned before. The flow is filtered by the data acquisition system.

To explore the transfer function from pressure to flow, the Fourier coefficients of the

∆p signal and pump flow are calculated. First the mean of the total signal is subtracted

from the signal to eliminate a peak at the frequency of 0 Hz. The harmonic frequency

coefficients cn are calculated in (1)

∑ ∆=−

tetxn

fxc

nftj

p

Hn

H **2*)(**2)( π

(1)

with n the order of harmonic, fH is the heart frequency (set to 1Hz), np the number of

periods in the given signal and x(t) the input signal. The desired transfer function is

the coefficients of the pressure difference divided by the flow coefficients.

Pressure sensitivity plots are created by plotting the coefficients of the pressure

difference versus the coefficients of the pump flow. By drawing a line between

measurements with the same pump speed, for different experiment settings the flow-

preload relationship, described by Khalil15

can be compared to the found flow-preload

relationships.

13

Chapter 3: Results

For all experiments for both pumps the pressures (figure 7, 8) and flows (figure 9, 10)

are measured. Figure 7 to 10 only show pressures and flows for experiment 1. Other

experiments have the similar shape on a different range in pressure and flow.

Figure 7: Pressures measured in experiment 1 with Jarvik 2000 LVAD. Baseline pressure is 120/80

mmHg. All LVAD-speed are shown. Top graph shows ventricular pressures, middle graph shows

aortic pressure and bottom graph shows pressure difference over the pump.

Figure 8: Pressures measured in experiment 1 with Levitronix CentriMag. Baseline pressure is 120/80

mmHg. All LVAD-speed are shown. Top graph shows ventricular pressures, middle graph shows

aortic pressure and bottom graph shows pressure difference over the pump.

14

Figure 9: Flows measured in experiment 1 with Jarvik. Baseline pressure is 120/80 mmHg. All LVAD-

speed are shown. Top graph shows aortic flows, middle graph shows pump flows and bottom graph

shows flow through aortic valve.

Figure 10: Flows measured in experiment 1 with Levitronix CentriMag. Baseline pressure is 120/80

mmHg. All LVAD-speed are shown. Top graph shows aortic flows, middle graph shows pump flows

and bottom graph shows flow through aortic valve.

It seems there is more difference among pump speeds between with the Levitronix

CentriMag LVAD, but this is due to the larger relative difference in rpm (1000, 1500,

1800, 2000, 2200, 2500 rpm) than with the Jarvik 2000 LVAD(8000, 9000, 10.000,

11.000, 12.000 rpm). Furthermore there is no real difference shown in the raw data of

all experiments.

15

For all experiments performed, Fourier analysis is done. Figure 11a to 11e show the

frequency response for the 5 pump speeds of the Jarvik. Figure 12a to 12f show the

frequency response for the Levitronix CentriMag. Mean and standard deviation of the

different cycles during an experiment are shown. All pump speeds for the Jarvik

show a constant response for different experiments for the harmonic frequencies

between 1 and 3 Hz. For the Levitronix CentriMag only higher pump speeds have

constant response between 1 and 3 Hz. For all speeds, for both pumps, there is

dispersion for a frequency of 0 Hz and 4 Hz.

Figure 11: Jarvik harmonic frequency response for different pump speeds (1 to 5)

16

Figure 12: Levitronix CentriMag harmonic frequency response for different pump speeds (1000, 1500,

1800, 2000, 2200, 2500 rpm)

A large spread at a frequency of 0 Hz is shown. This is the mean response in

mmHg/(L/min) and it is expected there is a difference between different mock-loop

settings. At 1000 rpm, the Levitronix CentriMag shows a large deviation, compared

with the other pump speeds. So the frequency response for this speed for all different

experiments is plotted in graph 13.

Figure 14 shows the mean and standard deviation of the different experiments for

each pump speed. For this figure the measurements with 1000 and 1500 rpm on the

Levitronix CentriMag are left out because these pump speeds have very low outflow

compared to the Jarvik 2000. Also these pump speeds show more disturbances than

the other pump speeds. Because pump speed 1 and 4 (Jarvik) have similar flow output

as 1800 and 2200 rpm (Levitronix CentriMag), respectively, these are represented as

thick lines in figure 14, to be better comparable.

This figure also shows that the harmonic frequencies of 1, 2 and 3 Hz are constant,

and the harmonic frequencies of 0 and 4 Hz are not.

17

Figure 13: Frequency response for the Levitronix CentriMag at 1000 rpm, which shows high responses

at 3 Hz for 3 experiments.

Figure 14: Mean and standard deviation for all experiments performed. The top graph is the frequency

response of the Jarvik. The bottom graph is the frequency response for the Levitronix CentriMag.

18

An ANOVA test is performed to compare the pump speed 1 of the Jarvik 2000 LVAD

with the 1800 rpm Levitronix CentriMag measurement because these have a similar

flow output. Only for the second harmonic frequency a statistically significant

difference can be found, Jarvik 2000 has a mean of 19.2 mmHgL-1

min-1

+/- 1.5 vs.

25.0 mmHgL-1

min-1

+/-10.9 (p<0.05). An ANOVA test has also been performed for

pump speed 4 of Jarvik 2000 LVAD with 2200 rpm of the Levitronix CentriMag.

ANOVA shows a statistically significant difference for the first harmonic frequency

of 21.2 mmHgL-1

min-1

+/-1.6 for Jarvik 2000 LVAD versus 17.9 mmHgL-1

min-1

+/-

2.5 for the Levitronix CentriMag (p<0.05). ANOVA also shows a statistically

significant difference for the third harmonic frequency of 28.7 mmHgL-1

min-1

+/-3.3

for Jarvik 2000 LVAD versus 33.6 mmHgL-1

min-1

+/- 6.0 for the Levitronix

CentriMag (p<0.05).

Figure 15 shows the 0th

order harmonic frequency response of four experiments with

different preload and afterload settings. There is a visible deviation at normal pressure

at the Jarvik LVAD (top two graphs). Also in the Levitronix CentriMag, at 10k

displacement (bottom left) the normal pressure measurement is shifted. The

Levitronix CentriMag at 15k displacement (bottom right) shows large deviation on

the middle two experiments, as the circles are not found near the fit-line.

Figure 15: |H| of LVAD flow vs. |H| of pressure difference over the LVAD at 0 Hz. The top graphs are

Jarvik 2000 LVAD, versus the bottoms graphs of Levitronix Centrimag. The left graphs are with a

piston movement of 10K cts. The right graphs are with 15K cts. In the graphs 2 measurements with

high pressures, one with normal and one with low pressure are shown. All measured LVAD speeds are

shown as circles. A line is fitted through these circles. Remaining circles are baseline measurement

(LVAD clamped) and no pump activation (LVAD unclamped).

19

Figure 16: Flow-preload relationships for the Jarvik 2000 (J2000) and HeartMate II (HMII) left

ventricular assist devices (LVADs) at a fixed afterload of 80 mm Hg measured by Khalil11. The lowest

pump speeds shown for these two LVADs are 8000 and 7000 rpm, respectively. The increment in the

pump speed between curves is 1000 and 200 rpm, respectively.

If a line would be fitted through circles of different experiments for the same LVAD

speed of figure 15, this could be compared with the work Khalil et al performed15

,

shown in figure 16. Preload in this graph is actually ∆p over the pump with a constant

afterload. This is why in this graph x-axis should be inverted for easy comparison

with graph 15. Khalil performed this test in a static mock-loop for a Jarvik 2000 and a

Heartmate II. However, there are not enough data points in figure 15 to give an

accurate fit through these points, especially, with the measurements for the Levitronix

CentriMag.

20

Chapter 4: Discussion

Explorative data analysis of figure 7 to 10 shows an artifact of the mock loop. It

shows that ventricular pressure arises with higher pump speeds. This is not expected

in an assisted ventricle as it is meant to unload the ventricle. Normally higher aortic

pressure is corrected by the baroreflex receptors, resulting in lower cardiac output.

However, the ventricle used in the mock-loop cannot adjust itself as a real ventricle

does by a decrease in contraction and heart frequency. Because of the higher pump

speed, more pressure is created in the aorta, behind the aortic valves, resulting in more

pressure building up in the ventricle until the aortic valves open.

With the Levitronix CentriMag there can be seen a backflow in some experiments

with, but only with, lower pump speed. In this case pump speed is so low, that

pressure difference over the pump is too high to create a positive flow through the

pump. However, these pump speeds are not clinically used. This is also the

explanation why data shown in figure 12a and 12 b is left out of figure 13.

The responses for a frequency of 4 Hz (fig 14) have a large spread of mean response.

These responses also have a large standard deviation, so this is caused by noise on the

signal.

A fit in figure 15 is not applied because fitting a line using only two points, does not

give the significance to compare it with the work of Khalil.

Conclusions which could be drawn are the more constant gain, independent of

harmonic frequency of the Jarvik 2000 LVAD versus the Levitronix CentriMag in

which the gain increases with harmonic frequency, statistically shown for the higher

output flow. So with a similar output flow, the harmonic frequencies are more equally

represented in the pressure difference over the pump. At the other hand, higher

harmonic (3 Hz) frequency are better represented in the Levitronix CentriMag

pressure difference, but the lower frequency (1 Hz) is less represented.

The standard deviations are all in the same range, although, Levitronix CentriMag has

some higher values occasionally, which are negligible. So both are reasonable similar

dependent on the experiment performed, meaning both LVADs are similar sensitive

to preload and afterload of the LVAD. The values are shown in table 1 and 2.

Table 1: gains [mmHgL-1min-1] of Jarvik 2000 LVAD at different harmonic frequencies

Jarvik 1Hz 2Hz 3Hz

mean std mean std mean std

low flow 17.4 6.5 19.1 1.4 27.8 3.1 high flow 21.2 1.64 24.0 2.0 28.7 3.3

p-value 0.04 <0.005 0.45

Table 1: gains [mmHgL-1min-1] of Levitronix CentriMag LVAD at different harmonic frequencies

Levitronix 1Hz 2Hz 3Hz

mean std mean std mean std

low flow 18.4 7.3 25.1 10.8 31.5 9.0

high flow 17.9 2.5 27.0 4.0 3.0 6.0

p-value 0.85 0.61 0.57

These values also show that there is a difference between low and high pump flow for

the first and second harmonic frequency for the Jarvik 2000 LVAD. So this means

21

that there is a pump speed dependency for the gain, so a relative higher pressure

difference is created with increasing pump flow.

Unfortunately figure 16, produced by Khalil et al, cannot be reproduced because of

lack of sufficient data (figure 15). Figure 15 shows a strange deviation for the normal

pressure, which is the first experiment performed. No explanation can be given for

this. It probably is caused by an error in the set up of the mock loop for the first

measurement.

.

22

Chapter 5: Conclusion

This study has been performed to study differences in pump characteristics of the

Jarvik 2000 LVAD and the Levitronix CentriMag.

One must notice first that an artifact is created using the Mock-loop. The ventricle

pressure rises with higher pump speeds, which normally would be corrected by the

baroreflex in the human body.

Harmonic frequency analysis showed a more independent gain for the Jarvik 2000

LVAD for an increase in harmonic frequency. The Levitronix CentriMag had a lower

gain for the lower frequency of 1 Hz than the Jarvik 2000 LVAD, but had a higher

gain for the higher frequency of 3 Hz. Jarvik also shows a pump speed dependency

for the gain for the first and second harmonic.

Unfortunately, not enough data was recorded to perform a comparison with Khalil et

al.15

. This is a good point of interest for future research. More measurements with

smaller steps in pump speed can be done for a better comparison. Also more

measurements have to be performed with smaller steps in preload and afterload for

more accurate conclusions.

23

References

1. 2004 Annual Report of the U.S. Organ Procurement and Transplantation Network and the

Scientific Registry of Transplant Recipients: Transplant Data 1994-2003. Department of

Health and Human Services, Health Resources and Services Administration, Healthcare

Systems Bureau, Division of Transplantation, Rockville, MD; United Network for Organ

Sharing, Richmond, VA; University Renal Research and Education Association, Ann Arbor,

MI.

2. Gemmato CJ, Forrester MD, Myers TJ, Frazier OH, Cooley DA. Thirty-Five Years of

Mechanical Circulatory Support at the Texas Heart Institute. Texas Heart Institute Journal.

2005; 32:168-177

3. Frazier, OH et al., Multicenter clinical evaluation of the HeartMate vented electric left

ventricular assist system in patients awaiting heart transplantation, J Thorac Cardiovasc Surg

2001;122:1186-95

4. Farrar D.J. et al., Long-term follow-up of Thoratec ventricular assist device bridge-to-

recovery patients successfully removed from support after recovery of ventricular function. J

Heart Lung Transplant. 2002 May;21(5):516-21

5. Stevenson LW, Miller LW, Desvigne-Nickens Pet al. for the REMATCH investigators. Left

ventricular assist device as destination for patients undergoing intravenous inotropic therapy.

A subset analysis from REMATCH (Randomized Evaluation of Mechanical Assistance in

Treatment of Chronic Heart Failure). Circulation, 2004; 110: 975–981.

6. Mancini D, Goldsmith R, Levin H et al. Comparison of exercise performance in patients with

chronic severe heart failure versus left ventricular assist devices. Circulation, 1998; 98: 1178–

1183.

7. Dew MA, Kormos RL, Winowich S et al. Quality of life outcomes in left ventricular assist

system inpatients and outpatients. ASAIO J, 1999; 45: 218–225.

8. Cox, L., Loerakker, S., Modeling Unloading of the Left Ventricle by the Levitronix

CentriMag LVAS using a Cardiovascular Simulator

9. Frazier OH, Myers TJ. Left Ventricular Assist System as a Bridge to Myocardial Recovery.

Ann Thorac Surg. 1999; 68:734-41.

10. Khalil, H.A. et al., Preload sensitivity of the Jarvik 2000 and Heartmate II Left Ventricular

Assist Devices. ASAIO Journal 2008; 245-248

11. Westaby, S. et al., First permanent implant of the Jarvik 2000 Heart, The Lancet, Volume 356,

Issue 9233, Pages 900 - 903, 9 September 2000

12. Frazier, O.H., et al, Research and Development of an Implantable, Axial-Flow Left

Ventricular Assist Device: The Jarvik 2000 Heart, Volume 71 (2001),journal Annals of

Thoracic Surgery (pp. S125-S132).

13. Frazier, O.H., Implantation of the Jarvik 2000 left ventricular assist device without the use of

cardiopulmonary bypass. Annals of Thoracic Surgery, 2003 (75:1028-30).

14. Ranjit, J. et al., Experience with the Levitronix CentriMag circulatory support system as a

bridge to decision in patients with refractory acute cardiogenic shock and multisystem organ

failure, J Thorac Cardiovasc Surg 2007;134:351-358

15. Khalil, H.A. et al., Continuous Flow Total Artificial Heart: Modeling and Feedback Control in

a Mock Circulatory System, ASAIO Journal 2008

16. Operation and Installation Manual Cardiovascular Simulator. HemoLab Cardiovascular

Engineering, Eindhoven, The Netherlands.

17. Milnor WR Hemodynamics. 1988. Williams & Wilkins, Baltimore