bs_bs_banner

Copyright © 2014 International Center for Artificial Organs and Transplantation and Wiley Periodicals, Inc.

Thoughts and Progress Synchronized Pulsatile Speed Control of Turbodynamic Left Ventricular Assist Devices: Review and Prospects Raffael Amacher, Gregor Ochsner, and Marianne Schmid Daners Institute for Dynamic Systems and Control, Department of Mechanical and Process Engineering, ETH Zurich, Zurich, Switzerland Abstract: Turbodynamic blood pumps are used clinically as ventricular assist devices (VADs). They are mostly operated at a constant rotational speed, which results in a reduced pulsatility. Previous research has analyzed pulsing pump speeds (speed modulation) to alter the interaction between the cardiovascular system and the blood pump. In those studies, sine- or square-wave speed profiles that were synchronized to the natural cardiac cycle were analyzed in silico, in vitro and in vivo. The definitions of these profiles with respect to both timing and speed levels vary among different research groups. The current paper provides a definition of the timing of these speed profiles such that the resulting hemodynamic effects become comparable. The results published in the literature are summarized and compared using this definition. Further, applied to a turbodynamic VAD, a series of measurements is conducted on a hybrid mock circulation using a constant speed as well as different types of square-wave speed profiles and a sinewave speed profile. When a consistent definition of the timing of the speed profiles is used, the hemodynamic effects observed in previous work are in agreement with the measurement data obtained for the current paper. These findings allow the conclusion that the speed modulation of turbodynamic VADs represents a consistent tool to systematically change the ventricular load and the pulsatility in the arterial tree. The timing that yields the minimal left ventricular load also yields the minimal arterial pulse pressure.

Ventricular assist devices (VADs) are a type of blood pump that are increasingly important in managing patients with cardiovascular diseases (1). A number of different types of devices have been used clinically in the last few decades (2). The firstgeneration “pulsatile” VADs (pVADs) are either

doi:10.1111/aor.12253 Received July 2013; revised September 2013. Address correspondence and reprint requests to Raffael Amacher, ETH Zurich, ML K36.3, Sonneggstrasse 3, 8092 Zurich, Switzerland. E-mail: [email protected]

pneumatically or electrically actuated volume displacement pumps that automatically generate a pulsatile flow. Second- and third-generation VADs rely on the turbodynamic principles of a rotating impeller in the blood stream and are commonly operated at a constant rotational speed. Despite this constant speed, the flow through a turbodynamic VAD (tVAD) is slightly pulsatile due to the remaining cardiac function. The pulsatility of the flow depends on the pressure head (difference between the downstream and upstream pressures of the pump) and on the pump characteristics (3,4). However, this flow pulsatility is smaller than the pulsatility in physiological pulsatile flow (5). The constant-speed strategy for tVADs can lead to permanent closure of the aortic valve, which in turn can lead to aortic valve fusion (6–9). Further, this strategy leads to reduced pulsatility, which might be the cause for the higher rate of gastrointestinal bleeding events occurring with tVADs as compared with pVADs (10). Speed modulation has the potential to partially resolve these problems. Various studies have been performed in a setting without a beating heart to analyze the ability of speed modulation to restore pulsatility in the systemic circulation (11–16). In the clinically more relevant case when the natural heart is present, an active synchronization of the speed modulation profiles to the natural cardiac cycle is required to obtain stable hemodynamics. In this case, not only the systemic flow pulsatility and perfusion but also the load of the left ventricle (LV) is influenced by speed modulation, which in turn may have consequences for patients with curable heart disease. The timing of the action of a cardiac assist device with respect to the cardiac cycle was first analyzed when intra-aortic balloon pumps (IABPs) were established. The timing of inflation and deflation with respect to ventricular contraction was optimized to yield a higher systemic and coronary perfusion during diastole as well as to reduce the load of the LV during systole. In the context of IABP actuation, the term “counterpulsation” was established (17,18). The much more invasive pVADs are mostly operated asynchronously, meaning that the pump rate is different from the heart rate (19), which leads to chaotic variations in hemodynamic parameters.When used in the synchronized mode, for instance when realized by Artificial Organs 2014, 38(10):867–899

868

THOUGHTS AND PROGRESS a) Square wave speed profile

ϕp

Pump speed (rpm)

6000

ωH

5000 4000

ϕ

ωL

3000

b) Sine wave speed profile Pump speed (rpm)

6000 5000 4000

ϕ

ωA

3000

c) Electrocardiogram (−)

1

0.5

0 −25

0

25 50 75 Cardiac cycle (%)

100

125

FIG. 1. Definition of the pulsatile speed profiles and the phase shift φ with respect to the cardiac cycle. For illustrative purposes, φ is equal to 30% for both cases. The following cases are considered: a) tVAD with square-wave speed profile and b) tVAD with sine-wave speed profile. Panel c) shows an electrocardiogram; the R waves are the reference points for the definition of the phase shift φ. The parameter φp denotes the pulse width of the square-wave speed profile. The parameters ωL and ωH are the low and the high speed values in the square-wave speed profile, and ωA is the speed amplitude of the sine-wave speed profile. The dotted lines in panels a) and b) denote the mean speed.

R wave detection (20), “copulsation” denotes the case when the starts of ejection of the ventricle and the pVAD occur at the same time. Counterpulsation is realized by applying a time delay of approximately half a cardiac cycle to the start of ejection of the pVAD. The ejection time delay can be set between zero and a full cardiac cycle, and the effect of such variations has been analyzed (21,22). Varying the ejection delay leads to a corresponding variation in hemodynamic parameters. In the case of tVADs with speed modulation, varying definitions of the phase shift have been used in previous publications. Figure 1 illustrates the definitions of the phase shift φ and the pulse width φp for a tVAD for the cases of square-wave and sine-wave Artif Organs, Vol. 38, No. 10, 2014

speed modulation, which are used consistently in the current paper. These two types of waveforms are described most often in the literature. The reference point is the R wave of the electrocardiogram (ECG). For a phase shift of 0%, the sine wave attains its maximum value and the square wave is at the center of its high-speed pulse exactly when the R wave occurs. This definition was chosen because a firstorder harmonic approximation of a square wave results in a sine wave with the same phase shift as the sine-wave speed profile shown. Therefore, the same phase shift is thought to have a comparable effect on hemodynamics, independently of the waveform. Two main problems need to be solved to implement speed modulation in vivo. First, a robust method for synchronizing the speed profile to the sinus rhythm of the native heart must be available. One way of solving this problem is by measuring the ECG and applying R wave detection, followed by filtering arrhythmic heart beats and irregular R wave detections as described in (23). Second, a speed modulation waveform must be selected and parameterized. In the current paper, the previous research on speed modulation of tVADs is summarized in the Literature Overview section. In the Methods section, a series of in vitro measurements is described, where a constant speed and various cases of speed modulation are tested with a tVAD in a hybrid mock circulation. The Results section provides the results of these experiments. In the Discussion section, the results of previous research and of the new experiments are compared and discussed. LITERATURE OVERVIEW Table 1 summarizes previous studies where either sine-wave or square-wave speed modulation was applied in various settings. Every study is assigned a number (S1–S8) to identify it in the text. The type of the study (in silico, in vitro, or in vivo) and the blood pump used are indicated. The phase shifts applied in these studies are converted to the definition shown in Fig. 1 and are also provided in Table 1, along with the pulse width used in the case of square-wave speed modulation. In most studies, only a limited number of phase shifts were analyzed (S1–S6); two research groups (S7 and S8) analyzed the whole phase shift range between 0% and 100%. In the following sections, the methods for choosing the speed profiles as well as the hemodynamic findings published in studies S1–S8 are summarized. This summary is divided into four groups of studies in which similar phase shift settings were applied.

THOUGHTS AND PROGRESS

869

TABLE 1. Literature overview of synchronized speed modulation using tVADs ID

Reference #

Speed profile

Study type

Blood pump

S1

(24,25)

Square wave

In silico

Hemopump

S2 S3

(26) (27–29)

Square wave Square wave

In vivo (goats) In vivo (pigs)

Evaheart Evaheart

S4

(30)

Square wave

In vivo (sheep)

CentriMag

S5

(31)

Square wave

In vivo (sheep)

CentriMag

In silico In silico In silico/in vitro

HeartMate II Incor Medos Micro-diagonal pump

Sine wave S6 S7 S8

(32) (33,34) (35,36)

Sine wave Sine wave Sine wave

Phase shifts

Pulse widths

44% 48% 42% 42% 16% 16% 66% 16% 66% 25% 75% 25% 75% 75% 0%–100% 0%–100%

72% 80% 68% 76% 33% 33% 66% 33% 66% 50% 50% N/A N/A N/A N/A N/A

Identifiers (S1–S8) are given to identify the studies in the text. The phase shift and pulse width values correspond to the definition shown in Figure 1 and are recalculated based on the information provided in the respective publications.

S1 He et al. (24,25) performed an in silico study using square-wave speed profiles. Seven different speed levels in the range of 17 000 rpm to 26 000 rpm, with intervals of 1500 rpm, were considered. Simulations were performed with varying speeds, phase shifts, and pulse widths. An objective function based on physiological values was designed, including LV stroke volume, mean left atrial pressure, minimum (diastolic) aortic pressure, and mean pump speed, and it was evaluated for all simulation conditions. Two sets of weighting parameters for this objective function were defined, one emphasizing stroke volume and one emphasizing minimum aortic pressure. Defined using two different levels of ventricular ischemia, the cardiovascular system model was set to two different conditions, such that a total of four solutions were obtained. For all of these cases, the best square-wave speed profile was characterized by a phase shift φ between 42% and 48% and a pulse width φp between 68% and 80%. The low speed was equal for all solutions, while the high speed varied. When a higher degree of heart failure was simulated, an increased high pump speed resulted from the optimization procedure. S2–S4 Umeki et al. and Ando et al. analyzed square-wave speed modulation in goats (26) and pigs (27–29). The low speed was set to approximately 700–1000 rpm. The high speed was manually adapted such that a desired bypass ratio (ratio between the flow through the blood pump and the total cardiac output) was reached. In different experiments, different mean speeds and speed amplitudes were required to reach

the desired bypass ratio. In terms of timing, two cases were analyzed (“co-pulse mode”: φ = 16%, φp = 33%; “counterpulse mode”: φp = 66%, φp = 66%). Pirbodaghi et al. (30) used a fixed mean speed of 2000 rpm. Different speed amplitudes of up to 1000 rpm were used to create different scenarios. The same two phase shift settings were analyzed by Pirbodaghi et al. (30), where the case with the lower phase shift value (φ = 16%) was termed “high systole” and the case with the higher phase shift value (φ = 66%) was termed “low systole.” In those studies, the duration of systole was assumed to be equal to 33% of the duration of one cardiac cycle, which explains the selection of the value of the pulse width φp. In both (26) and (30), an increase of the pulse pressure compared with the constant-speed case was observed at a phase shift of φ = 16%, and it reached almost the same value as when the pump was clamped. Energy equivalent pressure decreased almost to the mean arterial pressure when a constant pump speed was applied, but it increased to physiological values at a phase shift of φ = 16% (26). The mean coronary flow and the diastolic coronary flow at a phase shift of φ = 66% both increased compared with the constant-speed case, while at a phase shift of φ = 16%, a decrease in the diastolic coronary flow and an increase in the systolic coronary flow were observed. In the latter situation, the mean coronary flow remained unchanged (27). In (30), the highest mean coronary flow was observed at a phase shift of φ = 66%. LV end-diastolic volume was observed to decrease at φ = 66% and increase at φ = 16% when compared with the constant-speed case (28). In (30) end-diastolic volume was observed to decrease to Artif Organs, Vol. 38, No. 10, 2014

870

THOUGHTS AND PROGRESS

96% of the constant-speed case at φ = 66% and increase to 107% of the constant-speed case at φ = 16%. The LV oxygen consumption (MVO2) data from (29) and the stroke work data from (30) are both related to the workload of the native ventricle, and they show a strong agreement. Compared with the constant speed case, φ = 16% resulted in an MVO2 of approximately 112% and a stroke work of 104%, while a φ = 66% resulted in an MVO2 of approximately 87% and a stroke work of 72%. It is also worth noting that the MVO2 at constant speed reached around 80% of the baseline value and that the same results were obtained for bypass ratios of 100% and 50%. S5 and S6 Pirbodaghi et al. (31) analyzed two different phase shifts (φ = 25% and φ = 75%), where the effect of four different types of waveforms (including sinewave and square-wave speed modulation with a pulse width of φp = 50%) were compared. The mean speed was kept at 2000 rpm, and the speed amplitude was varied. The minimum stroke work (74% of the value obtained at constant speed) occurred at φ = 75%, and the maximum stroke work (96% of the value obtained at constant speed) was obtained at φ = 25%. In this study, the effect of applying different waveforms was shown to have a negligible effect on the hemodynamics. Differences in calculated hemodynamic parameters were mostly due to the different phase shift settings. Cox et al. (32) performed a computer simulation study, where varying levels of mean pump speeds (7000 rpm to 11 000 rpm in steps of 1000 rpm) were used to compare the constant-speed case with a sinewave speed modulation case with φ = 75%. This mode was called “counterpulsation mode.” The maximum speed was held constant at 14 000 rpm; thus, the speed amplitude varied between 7000 rpm and 3000 rpm. Cox et al. concluded that speed modulation of the pump at a fixed φ = 75% does not restore the pulsatility that is reduced in the constant speed case. On the other hand, compared with the constant speed case, the cardiac output and the coronary flow were increased, and the stroke work and the heart rate, in response to a programmed reflex mechanism, were decreased. S7–S8 Shi et al. (33,34) performed an in silico study using synchronized sine-wave speed modulation. A mean pump speed of 4000 rpm was chosen, such that a reasonable total cardiac output resulted when no speed modulation was applied. The pulsation ratio Artif Organs, Vol. 38, No. 10, 2014

(ratio between speed amplitudes and mean speed) was varied between 0% and 100%. For a pulsation ratio of 100%, the minimum speed is equal to 0 rpm. A parameter study was performed where the full range of phase shifts φ (0–100%) was also analyzed. An objective function based on several hemodynamic variables, including cardiac output, minimum value of the pump flow, pulse pressure, mean left atrial pressure, mean LV pressure, energy equivalent pressure, mean LV volume, and LV stroke volume was defined. It was found that a phase shift of φ = 81% yielded the best performance with respect to the chosen objective function. The outcome of such a study strongly depends on the structure and parameterization of the objective function. Vandenberghe et al. (35,36) analyzed synchronized sine-wave speed modulation both in silico and in vitro, where the full range of phase shifts φ (0–100%) was considered. A mean speed of 4000 rpm and a speed amplitude of 2000 rpm were chosen for all cases analyzed. The mean arterial pressure, stroke volume, pressure–volume area, and the end-diastolic volume revealed a sinusoidal pattern when plotted against the phase shift. Vandenberghe et al. concluded that a counterpulsation setting (such that the highest pump speed occurs during diastole) yields the best unloading, which corresponds to a phase shift of approximately φ = 80%. At this phase shift, the mean arterial pressure attained its maximum value, while the stroke volume and the end-diastolic volume attained their minimum value (36). The other extreme values occurred at a phase shift of approximately φ = 30%. METHODS In order to illustrate the effect of a cardiaccyclesynchronized tVAD with sine-wave and squarewave speed modulation, a series of measurements was conducted on the hybrid mock circulation described in (37). This test bench was based on a hardware-in-the-loop concept, where a numerical model of the human circulation (38) is simulated in software and interacts with a real blood pump using two pressure-controlled reservoirs and one flow probe. The pressure reference signals are supplied by the numerical circulation model and tracked by proportional–integral–derivative controllers in real time. The reservoir upstream of the VAD represents the LV, while the reservoir downstream of the VAD represents the aorta. The flow probe measures the flow through the VAD and provides this value to the numerical circulation model, such that a real-time interaction between the blood pump and the numeri-

THOUGHTS AND PROGRESS

871

TABLE 2. Overview of the speed profiles used in the hybrid mock circulation Mode

Pulse width φp (%)

Mean speed ωc (rpm)

Amplitude ωA (rpm)

N/A 30 50 70 N/A

4000 4000 4000 4000 4000

N/A 1000 1000 1000 1000

Constant speed Square wave Square wave Square wave Sine wave

The phase shift is varied between 0% and 95% in increments of 5%.

cal circulation model is established. For all investigations, the numerical circulation model was set to simulate a pathological situation with an increased heart rate of 90 bpm and a decreased contractility of 34% as described in (38). The tVAD (Deltastream DP2, Medos Medizintechnik AG, Stolberg, Germany) implemented in the hybrid mock circulation was actuated by a servoamplifier (Accelus, Copley Controls Corp., Canton, MA, USA) with encoder feedback to yield speed control with a high bandwidth. Five different reference speed settings were defined for the tVAD (constant speed, square waves with a pulse width of φp = {30%, 50%, 70%} and a sine wave), where the mean speed was always equal to 4000 rpm. The amplitude of the sine wave was equal to 1000 rpm, and the low (ωL) and high (ωH) speeds of the square wave signals were calculated as

ω L = ω c − 2ω A ϕ p

(1)

ω H = ω c + 2ω A (1 − ϕ p )

(2)

These speeds thus depend on the pulse width φp that is chosen. This implementation ensures that the difference between these two speeds is equal to the speed amplitude ωA multiplied by a factor of two, and the mean speed is equal to the prescribed value ωc = 4000 rpm. For all configurations except for the constant-speed case, the phase shift φ was changed between 0% and 95% in increments of 5%. The frequency for all waveforms was dictated by the simulated heart rate. Table 2 lists all configurations used for the measurements. Each setting was applied for 20 s to allow the system to reach steady state. The data obtained from one such steady-state heart beat for each setting were then analyzed to calculate the following parameters: total cardiac output (defined as the mean flow through the aortic valve plus the mean flow through the pump), LV stroke work, LV end-diastolic volume, mean arterial pressure, mean flow through the aortic valve, arterial pulse pressure, mean flow through the pump, surplus hemodynamic pressure, mean backflow to the LV through the pump, bypass ratio,

and pulsatility index (difference between the maximum and the minimum flow in the aorta, divided by its mean value). Surplus hemodynamic pressure (SHP) is calculated as the difference between the energy equivalent pressure (EEP) and the mean arterial pressure (MAP):

∫ SHP =

t2

t1

pao ⋅ (qav + qbp ) dt

( t2 − t1 ) ⋅ TCO

− MAP = EEP − MAP (3)

where TCO is total cardiac output, pao is the aortic pressure, qav is the flow through the aortic valve, and qbp is the flow through the blood pump; t1 and t2 denote the onsets of two consecutive cardiac cycles. The commonly used term “surplus hemodynamic energy” is replaced by the term “surplus hemodynamic pressure” because this index is defined in units of pressure and not in units of energy. RESULTS Figure 2 shows the following indices plotted versus phase shift: total cardiac output, stroke work, enddiastolic volume, mean arterial pressure, mean flow through the aortic valve, arterial pulse pressure, mean flow through the pump, surplus hemodynamic pressure, and mean backflow to the LV through the pump. As a reference, the gray line in each panel indicates the value that is achieved when a constant pump speed of 4000 rpm is applied. The effect of phase shift on these variables is qualitatively equal for all cases analyzed. The maximum and minimum LV stroke work is attained at phase shifts of approximately 30% and 80%, respectively. For the square-wave speed profiles, the following tendency can be seen: for the case with lower pulse width (φp = 30%), the effect on the hemodynamic values is reduced at high phase shifts and increased at low phase shifts, while the opposite is true for the case with higher pulse width (φp = 70%). It is possible to control most of the parameters to reach values above or below the constant speed parameter value by selecting a specific phase shift. The ranges of the parameters calculated are comparable for the different speed modulation Artif Organs, Vol. 38, No. 10, 2014

872

THOUGHTS AND PROGRESS

0.35

4.8

SW (J)

TCO (L/min)

5

4.6

0.3 0.25 0.2

4.4

DISCUSSION

0.15

4.2

c)

d) 100 MAP (mm Hg)

180 EDV (mL)

attained at a constant tVAD speed (PI ≈ 1.4). For phase shifts around φ = 30%, the maximum value was attained (PI ≈ 3.5).

b)

a)

160 140

98 96 94 92

e)

f) 30 PP (mm Hg)

AoF (L/min)

0.8 0.6 0.4 0.2

25 20 15 10

0

g)

h) SHP (mm Hg)

PF (L/min)

5 4.5 4

6 4 2 0 0

20

40 60 Phase shift (%)

80

100

i) BF (L/min)

0 Constant speed Square wave, ϕp = 30% Square wave, ϕp = 50% Square wave, ϕp = 70% Sine wave

−0.1 −0.2 −0.3 −0.4 0

20

40 60 Phase shift (%)

80

100

FIG. 2. The variations of selected hemodynamic parameters with varying phase shift φ. The thick gray line shows the values resulting from applying a constant speed. TCO, total cardiac output; SW, left ventricular stroke work; EDV, left ventricular end-diastolic volume; MAP, mean arterial pressure; AoF, mean flow through the aortic valve; PP, systemic arterial pulse pressure; PF, mean flow through pump; SHP, surplus hemodynamic pressure; BF, mean backflow to left ventricle through pump.

waveforms that were applied, presumably due to the identical mean speed and the constant conditions of the simulated circulatory system. The bypass ratios of all the cases analyzed had a mean value of 94% and a standard deviation of 4%.The flow through the aortic valve was characterized by a median value of 0.27 L/ min, a minimum value of 0 L/min, and a maximum value of 0.8 L/min. The aortic valve opened in all cases except for the square wave with a pulse width of φp = 70% at the phase shifts between φ = 80% and φ = 100%. The maximum flow through the aortic valve occurred roughly between the phase shifts of 50% and 60%. The pulsatility index revealed a pattern similar to those of surplus hemodynamic pressure and pulse pressure. At phase shifts around φ = 80%, it was approximately equal to the value Artif Organs, Vol. 38, No. 10, 2014

The current paper discusses the cardiaccyclesynchronized speed modulation of tVADs. It focuses on a definition of phase shift that leads to comparable hemodynamic effects regardless of the speed modulation waveform that is applied. An overview of the literature using speed modulation of tVADs is given, where the phase shifts applied in the respective studies are recalculated according to the definition established in the current paper to ensure comparability of the results. Further, the results of an in vitro study are presented, where measurements using a tVAD with constant speed, sine-wave speed modulation, and square-wave speed modulation at three different pulse widths were conducted. The independent variable was the phase shift, which was varied between 0% and 95% in 5% increments. Results of previous research, described in the Literature Overview section, qualitatively match the results presented in the Results section with respect to the phase shift when the phase shift is defined according to the definition shown in Figure 1. This agrees with the finding in (31), where the authors concluded that no major differences in the hemodynamics occur when using sine waves, square waves, or sawtooth-shaped or triangle-shaped waveforms. The results in studies S1–S8 were obtained with six different tVADs. The studies include in silico, in vitro, and in vivo experiments; in the latter, goats, pigs, and sheep were used. The attainable maximum and minimum values of hemodynamic variables over the full range of phase shifts vary with the state of the cardiovascular system, as well as with the mean speed and the speed amplitude. The latter two were not varied in the new series of experiments. As demonstrated in (34), a higher speed amplitude can be expected to increase the range of the parameters when the phase shift is varied. When using a tVAD with either sine-wave or square-wave speed modulation, there is a tradeoff that is inherent to the system. Either the unloading of the ventricle can be improved until maximum unloading is achieved at a phase shift of approximately 80% (upstream benefit), expressed by a further reduction of the end-diastolic volume, the stroke work, and consequently myocardial wall stress and external work, compared with the constant speed case; or the hemodynamic conditions in the aorta can be altered toward physiologic pulse pressure and

THOUGHTS AND PROGRESS surplus hemodynamic pressure at a phase shift of approximately 30% (downstream benefit), which could help to prevent vascular dysfunction (39,40). The investigations of Ando et al. (26) suggest that the pulsatility in terms of pulse pressure and energy equivalent pressure can be restored to values comparable to physiologic ones by speed modulation. An intermediate phase shift between 50% and 60% has been shown to result in maximal flow through the aortic valve, which might help to reduce aortic valve dysfunction. The approach chosen does not allow both upstream and downstream values to be optimized at the same time. Most likely, myocardial recovery is favored with a more pronounced unloading of the LV (41–43). Therefore, patients could potentially benefit from speed modulation with a phase shift of 80% in the early postoperative phase and a gradual reloading of the ventricle by changing phase shift as a weaning procedure. The use of tVADs induces blood damage due to the contact of blood with an artificial surface and the high shear stress levels inside the pump (44,45). This shear stress can lead to hemolysis and platelet activation. There are differences in the amount of blood damage to be expected from using different clinically used VADs (46), and the design of the pump geometry and the impeller has an influence on the blood damage levels (47,48). It has been shown that higher pump speeds and lower pump flows increase the modified index of hemolysis (49) under stationary conditions, meaning that the tVADs are analyzed at a constant rotational speed and at constant pressure heads. The influence of speed modulation on hemolysis has been analyzed by Tayama et al. (50,51). The outcome of these studies suggests that dynamic impeller speed variation increases the rate of hemolysis compared with constant-speed operation. Further research is necessary to further analyze the effect of speed modulation on blood damage in a physiologic environment. Clearly, counterpulsation (high pump speed during ventricular diastole) leads to higher coronary flow due to the increased diastolic systemic arterial pressure in silico (24,32,33,35). In vivo experiments reveal somewhat contradictory results. Ando et al. (27) observed that diastolic coronary flow could be significantly increased using a phase shift of 66%. In contrast, the study of Pirbodaghi et al. (31) reported that only minor variations in coronary flow were obtained when phase shifts of 25% and 75% were applied. However, an earlier in vivo study using a pVAD suggested that the coronary circulation has its own autoregulation mechanism that relates coronary flow to the oxygen demand of the ventricle (52). This

873

could not be confirmed in a later study (53), where the use of a tVAD did not show any influence of bypass flow and thus left ventricular load on coronary flow. Coronary perfusion under VAD assistance thus represents a topic requiring further research to determine, and if so, to what extent speed modulation has any significant influence. Control systems described in the literature that automatically adapt the speed of tVADs do not include cardiac-cyclesynchronized speed modulation. The task of finding an appropriate mean speed for a tVAD is not discussed in the current paper and is challenging in itself (54–56). To date, no automatic control algorithm has been described in the literature that automatically adapts the parameters defining pulsatile speed profiles. Further, there is an inherent limitation to the tracking performance of the reference speed. Due to the rotor inertia and the maximum power the motor can generate, which lead to a limitation in the derivative of pump speed with respect to time, square-wave speed profiles cannot be tracked perfectly. Speed control algorithms that deliver a low bandwidth diminish differences between square-wave and sine-wave speed profiles. Further issues include the fact that varying cannula lengths have an influence on the fluid inertia, as well as the fact that different nonlinear pump characteristics (pressure head-vs.-flow diagrams) can lead to varying hemodynamic results. These effects would have to be considered in the process of designing a speed modulation strategy for a specific blood pump. In silico optimizations generate outcomes that depend on the objective function chosen. The optimization study conducted in (24,25) yielded square waves with phase shifts approximately between 40% and 50%. Thus, a comparison of those results with those shown in Fig. 2 allows the conclusion that the chosen objective function results in an increased pulse pressure while keeping the end-diastolic value and stroke work at levels similar to those obtained with a constant speed. The objective function chosen in (33,34) yields a phase shift of φ = 81%, such that this objective function focuses on ventricular unloading because end-diastolic volume and stroke work attain their minimum at that phase shift value, whereas pulse pressure is approximately equal to that obtained when operating the pump at a constant speed. Presumably, the quintessential speed profile that leads to optimal physiologic performance is a nontrivial waveform unlike a sine wave or a square wave. Further research is necessary to find speed profiles that optimize the dynamic interaction between the tVAD and the cardiovascular system. Such speed Artif Organs, Vol. 38, No. 10, 2014

874

THOUGHTS AND PROGRESS

profiles depend on the type of pump used and on the ever-varying state of the cardiovascular system. In the case of in silico studies, mathematical optimization and optimal control theory represent promising approaches to find the best possible speed profile for a specific tVAD in a specific cardiovascular environment (57). The selection of a speed profile for modulating the action of a tVAD is not the only possible solution. From a conceptual point of view, it would also be possible to use a strategy that directly applies pump flow modulation (58). In that case, a reliable estimate or a measurement of this flow would be necessary. Speed modulation is easier to realize because speed control is usually available in clinically used tVADs to keep the pump speed at a preset constant value. Although the availability of flow measurement data with a sufficient sampling frequency and long-term accuracy with a compact sensor or an estimator is a goal, it is not a clinical reality as yet. However, the approach of flow modulation would be preferable because it is “closer” to the cardiovascular system and the hemodynamic effects occurring with the use of flow modulation are easier to predict. CONCLUSION The current paper has demonstrated that the various proposed strategies for synchronized speed modulation of tVADs lead to comparable hemodynamic effects when a consistent definition of the phase shift is used. Speed modulation should therefore be considered for clinical practice to either increase pulsatility in the systemic arterial circulation (downstream benefit) or to enhance LV unloading (upstream benefit). Adjustment of the phase shift of cardiac-cyclesynchronized speed modulation leads to control over the tradeoff between these two benefits. Future research should include the optimization of the speed modulation profiles and the analysis of potential long-term effects and benefits of this approach. REFERENCES 1. Kirklin JK, Naftel DC, Kormos RL, et al. Fifth INTERMACS annual report: risk factor analysis from more than 6000 mechanical circulatory support patients. J Heart Lung Transpl 2013;32:141–56. 2. Timms D. A review of clinical ventricular assist devices. Med Eng Phys 2011;33:1041–7. 3. Yamazaki K, Saito S, Kihara S, Tagusari O, Kurosawa H. Completely pulsatile high flow circulatory support with a constantspeed centrifugal blood pump: mechanisms and early clinical observations. Gen Thorac Cardiovasc Surg 2007;55:158–62. 4. Gaddum NR, Fraser JF, Timms DL. Increasing the transmitted flow pulse in a rotary left ventricular assist device. Artif Organs 2012;36:859–67.

Artif Organs, Vol. 38, No. 10, 2014

5. Soucy KG, Koenig SC, Giridharan GA, Sobieski MA, Slaughter MS. Defining pulsatility during continuous-flow ventricular assist device support. J Heart Lung Transpl 2013;32: 581–7. 6. Rose AG, Park SJ, Bank AJ, Miller LW. Partial aortic valve fusion induced by left ventricular assist device. Ann Thorac Surg 2000;70:1270–4. 7. Banchs JE, Dawn B, Abdel-Latif A, et al. Acquired aortic cusp fusion after chronic left ventricular assist device support. J Am Soc Echocardiogr 2006;19:1401.e1–1401.e3. 8. John R, Mantz K, Eckman P, Rose A, May-Newman K. Aortic valve pathophysiology during left ventricular assist device support. J Heart Lung Transpl 2010;29:1321–9. 9. May-Newman K, Enriquez-Almaguer L, Posuwattanakul P, Dembitsky W. Biomechanics of the aortic valve in the continuous flow VAD-assisted heart. ASAIO J 2010;56:301–8. 10. Crow S, John R, Boyle A, et al. Gastrointestinal bleeding rates in recipients of nonpulsatile and pulsatile left ventricular assist devices. J Thorac Cardiovasc Surg 2009;137:208–15. 11. Miller GE, Etter BD, Dorsi JM. A multiple disk centrifugal pump as a blood flow device. IEEE Trans Biomed Eng 1990;37:157–63. 12. Bearnson GB, Olsen DB, Khanwilkar PS, Long JW, Allaire PE, Maslen EH. Pulsatile operation of a centrifugal ventricular assist device with magnetic bearings. ASAIO J 1996;42: M620–4. 13. Göbel C, Arvand A, Eilers R, et al. Development of the MEDOS/HIA DeltaStream extracorporeal rotary blood pump. Artif Organs 2001;25:358–65. 14. Bourque K, Dague C, Farrar D, et al. In vivo assessment of a rotary left ventricular assist device-induced artificial pulse in the proximal and distal aorta. Artif Organs 2006;30:638–42. 15. Shiose A, Nowak K, Horvath DJ, Massiello AL, Golding LAR, Fukamachi K. Speed modulation of the continuous-flow total artificial heart to simulate a physiologic arterial pressure waveform. ASAIO J 2010;56:403–9. 16. Khalil HA, Kerr DT, Schusterman IIMA, Cohn WE, Frazier OH, Radovancevic B. Induced pulsation of a continuous-flow total artificial heart in a mock circulatory system. J Heart Lung Transpl 2010;29:568–73. 17. Nanas JN, Moulopoulos SD. Counterpulsation: historical background, technical improvements, hemodynamic and metabolic effects. Cardiology 1994;84:156–67. 18. Moulopoulos SD. Intra-aortic balloon counterpulsation 50 years later—initial conception and consequent ideas. Artif Organs 2011;35:843–8. 19. Maybaum S, Williams M, Barbone A, Levin H, Oz M, Mancini D. Assessment of synchrony relationships between the native left ventricle and the HeartMate left ventricular assist device. J Heart Lung Transpl 2002;21:509–15. 20. Farrar DJ, Compton PG, Lawson JH, Hershon JJ, Hill JD. Control modes of a clinical ventricular assist device. IEEE Eng Med Biol Mag 1986;5:19–25. 21. Heredero A, Perez-Caballero R, Otero J, et al. Synchrony relationships between the left ventricle and a left ventricular assist device: an experimental study in pigs. Int J Artif Organs 2012;35:272–8. 22. Amacher R, Weber A, Brinks H, et al. Control of ventricular unloading using an electrocardiogram-synchronized Thoratec paracorporeal ventricular assist device. J Thorac Cardiovasc Surg 2013;146:710–7. 23. Amacher R, Ochsner G, Ferreira A, Vandenberghe S, Schmid Daners M. A robust reference signal generator for synchronized ventricular assist devices. IEEE Trans Biomed Eng 2013;60:2174–83. 24. He P, Bai J, Xia DD. Optimum control of the Hemopump as a left-ventricular assist device. Med Biol Eng Comput 2005;43: 136–41. 25. He P, Bai J, Qiao H. A simulation study of hemodynamic benefits and optimal control of axial flow pump-based left ventricular assist device. In: Biomechanical Systems

bs_bs_banner

THOUGHTS AND PROGRESS

26. 27. 28.

29.

30. 31. 32.

33. 34.

35. 36.

37. 38.

39.

40.

41. 42. 43.

44. 45.

Technology, vol. 2, Cardiovascular Systems. New York: World Scientific, 2007;1–28. Ando M, Nishimura T, Takewa Y, et al. Electrocardiogramsynchronized rotational speed change mode in rotary pumps could improve pulsatility. Artif Organs 2011;35:941–7. Ando M, Takewa Y, Nishimura T, et al. A novel counterpulsation mode of rotary left ventricular assist devices can enhance myocardial perfusion. J Artif Organs 2011;14:185–91. Umeki A, Nishimura T, Ando M, et al. Alteration of LV enddiastolic volume by controlling the power of the continuousflow LVAD, so it is synchronized with cardiac beat: development of a native heart load control system (NHLCS). J Artif Organs 2012;15:128–33. Umeki A, Nishimura T, Takewa Y, et al. Change in myocardial oxygen consumption employing continuous-flow LVAD with cardiac beat synchronizing system, in acute ischemic heart failure models. J Artif Organs 2013;16:119–28. Pirbodaghi T, Weber A, Axiak S, Carrel T, Vandenberghe S. Asymmetric speed modulation of a rotary blood pump affects ventricular unloading. Eur J Cardio-Thorac 2013;43:383–8. Pirbodaghi T, Axiak S, Weber A, Gempp T, Vandenberghe S. Pulsatile control of rotary blood pumps: does the modulation waveform matter? J Thorac Cardiovasc Surg 2012;144:970–7. Cox LGE, Loerakker S, Rutten MCM, de Mol BAJM, van de Vosse FN. A mathematical model to evaluate control strategies for mechanical circulatory support. Artif Organs 2009;33: 593–603. Shi Y, Lawford PV, Hose DR. Numerical modeling of hemodynamics with pulsatile impeller pump support. Ann Biomed Eng 2010;38:2621–34. Shi Y, Brown AG, Lawford PV, Arndt A, Nuesser P, Hose DR. Computational modelling and evaluation of cardiovascular response under pulsatile impeller pump support. Interface Focus 2011;1:320–37. Vandenberghe S, Segers P, Meyns B, Verdonck P. Unloading effect of a rotary blood pump assessed by mathematical modeling. Artif Organs 2003;27:1094–101. Vandenberghe S, Segers P, Antaki JF, Meyns B, Verdonck PR. Hemodynamic modes of ventricular assist with a rotary blood pump: continuous, pulsatile, and failure. ASAIO J 2005;51: 711–8. Ochsner G, Amacher R, Amstutz A, et al. A novel interface for hybrid mock circulations to evaluate ventricular assist devices. IEEE Trans Biomed Eng 2013;60:507–16. Colacino FM, Moscato F, Piedimonte F, Arabia M, Danieli GA. Left ventricle load impedance control by apical VAD can help heart recovery and patient perfusion: a numerical study. ASAIO J 2007;53:263–77. Nishimura T, Tatsumi E, Takaichi S, et al. Prolonged nonpulsatile left heart bypass with reduced systemic pulse pressure causes morphological changes in the aortic wall. Artif Organs 1998;22:405–10. Westaby S, Bertoni GB, Clelland C, Nishinaka T, Frazier OH. Circulatory support with attenuated pulse pressure alters human aortic wall morphology. J Thorac Cardiovasc Surg 2007;133:575–6. Burkhoff D, Klotz S, Mancini DM. LVAD-induced reverse remodeling: basic and clinical implications for myocardial recovery. J Card Fail 2006;12:227–39. Drakos SG, Terrovitis JV, Anastasiou-Nana MI, Nanas JN. Reverse remodeling during long-term mechanical unloading of the left ventricle. J Mol Cell Cardiol 2007;43:231–42. Krabatsch T, Schweiger M, Dandel M, et al. Is bridge to recovery more likely with pulsatile left ventricular assist devices than with nonpulsatile-flow systems? Ann Thorac Surg 2011; 91:1335–40. Goubergrits L, Affeld K. Numerical estimation of blood damage in artificial organs. Artif Organs 2004;28:499–507. Bluestein D, Chandran KB, Manning KB. Towards nonthrombogenic performance of blood recirculating devices. Ann Biomed Eng 2010;38:1236–56.

875

46. Heilmann C, Geisen U, Benk C, et al. Haemolysis in patients with ventricular assist devices: major differences between systems. Eur J Cardio-Thorac 2009;36:580–4. 47. Wang SS, Chu SH, Chou NK, Qian KX. The pulsatile impeller pump for left ventricular assist. Artif Organs 1996;20:1310–3. 48. Qian KX. Pulsatile impeller heart: a viable alternative to a problematic diaphragm heart. Med Eng Phys 1996;18:57–66. 49. Arvand A, Hormes M, Reul H. A validated computational fluid dynamics model to estimate hemolysis in a rotary blood pump. Artif Organs 2005;29:531–40. 50. Tayama E, Niimi Y, Takami Y, et al. Hemolysis test of a centrifugal pump in a pulsatile mode: the effect of pulse rate and RPM variance. Artif Organs 1997;21:1284–7. 51. Tayama E, Nakazawa T, Takami Y, et al. The hemolysis test of the Gyro C1E3 pump in pulsatile mode. Artif Organs 1997;21: 675–9. 52. Noda H, Takano H, Taenaka Y, et al. Regulation of coronary circulation during left ventricular assist. Trans Am Soc Artif Intern Organs 1989;35:445–7. 53. Voitl P, Vollkron M, Bergmeister H, Wieselthaler G, Schima H. Coronary hemodynamics and myocardial oxygen consumption during support with rotary blood pumps. Artif Organs 2009;33:77–80. 54. Antaki JF, Boston JR, Simaan MA. Control of heart assist devices. In: Proceedings of the 42nd IEEE Conference on Decision and Control, Maui, HI. New York: Institute of Electrical and Electronics Engineers, 2003;4084–9. 55. Schima H, Vollkron M, Jantsch U, et al. First clinical experience with an automatic control system for rotary blood pumps during ergometry and right-heart catheterization. J Heart Lung Transpl 2006;25:167–73. 56. Wu Y, Allaire PE, Tao G, Olsen D. Modeling, estimation, and control of human circulatory system with a left ventricular assist device. IEEE Trans Control Syst Technol 2007;15:754–67. 57. Bryson AE, Ho YC. Applied Optimal Control. New York: Taylor & Francis, 1975. 58. Ising M, Warren S, Sobieski MA, Slaughter MS, Koenig SC, Giridharan GA. Flow modulation algorithms for continuous flow left ventricular assist devices to increase vascular pulsatility: a computer simulation study. Cardiovasc Eng Technol 2011;2:90–100.

Artif Organs, Vol. 38, No. 10, 2014