Continuous-flow left ventricular devices (cf-LVADs) are being used more frequently as bridge-to-transplantation and destination therapy because of their small size and reliability and have significantly increased survival and quality of life in patients with heart failure. However, current control systems operate at a fixed pump speed that cannot respond to changes in ventricular loading. As patients leave the hospital and return to their normal activities, ventricular loading can change because of changes in left ventricular (LV) contractility, heart rate, arterial pressure, and blood volume related to normal daily activities (e.g., sleep, exercise, positional changes, etc.). To reduce the risk of suction that can occur during periods of reduced preload, pump speed is set conservatively using echocardiography to assess ventricular unloading and ensuring that the aortic valve opens periodically.1 Consequently, there is limited increase in pump flow available to support increased demand, such as during exercise,2 and as such, patient exercise capacity is limited.3–7 A preload-sensitive control system that mimics the Frank-Starling mechanism is necessary to improve pump flow during exercise, limit suction events, and enhance quality of life.8
Control systems based on direct measurements of ventricular loading based on pressure or volume have been proposed9–13 but lack reliable sensors to implement. We have developed an inlet cannula tip for cf-LVADs with integrated electrodes for volume sensing based on conductance. The cannula tip was tested in vivo in conjunction with the HeartMate II axial-flow pump in an acute ovine model. This integrated conductance catheter tip provides a measure of volume unloading that will enable the development of a real-time control system for optimal management of cf-LVADs.
The cannula tip design is shown in Figure 1. The cannula body is constructed from polyetheretherketone (PEEK) with an outer diameter of 2.26 cm and an inner lumen diameter of 1.78 cm. Four 1 mm diameter platinum-iridium electrodes (90:10 Pt:Ir) are wrapped around groves in an outer shell to form the conductance catheter (Figure 1, A). The electrodes pass through the outer shell and are crimped to nickel alloy wires (MP35N) in a sealed cavity that is flush with the outer shell (Figure 1, A). The outer shell is bonded to the cannula body (Figure 1, B) that is machined and polished to form a seamless blood contacting inner surface. The MP35N wires pass through a space between the outer shell and cannula body and exit at the base of the cannula tip. The cavity is potted with epoxy to prevent liquid ingress. A flange slides over the assembled unit for attaching a felt sewing ring. The flange placement is adjustable to allow a certain degree of control over the insertion depth of the cannula tip into the left ventricle. The dimensions of the cannula tip are shown in Figure 1, C and a photograph of the finished device used in this study is shown in Figure 1, D.
Conductance is measured using the four platinum-iridium wires on the external shell of the cannula tip. A sinusoidal current excitation is applied across the outer pair of electrodes, and voltage is sensed across the inner pair of electrodes spaced 0.76 cm apart (Figure 1, C). The conductance is calculated from the ratio of the excitation current to the sensed voltage. The conductance algorithm is implemented in a microcontroller (PSoC5; Cypress Semiconductor Corporation, San Jose, CA) with configurable analog and digital blocks for generating the excitation current waveform (50 kHz sinusoid, 250 μA peak-to-peak amplitude). The sensed voltage is amplified with an instrumentation amplifier (AD8421, Analog Devices) before being sampled with the analog to digital converters integrated in the microcontroller. A quadrature-sampling algorithm is used whereby the voltage waveform is sampled synchronously with the excitation current waveform at phase positions of 0, π/2, π, and 3π/2 radians, at a rate of one sample per period. The phase indexed samples are averaged over multiple waveform periods within the sampling interval, and the amplitude of the voltage waveform is calculated from the phase samples at an overall sampling rate of 100 Hz.
The conductance catheter was tested in an acute ovine model (n = 3). The HeartMate II pump was used to provide circulatory support and unload the ventricle. A custom in-house motor controller was used to control pump speed.14 A cardiothoracic surgeon positioned the blood pump in the chest via a left thoracotomy. The outlet cannula was anastomosed to the proximal descending aorta and the cannula tip was placed in the ventricular apex. Left ventricular pressure was measured with a Millar intracardiac pressure transducer (Millar Instrument Inc., Houston, TX). Pump flow was measured using an ultrasonic flow transducer (Transonic Systems Inc., Ithaca, NY) in line with the cannula tip and blood pump. Echocardiography was used to assess cannula placement, aortic valve opening, and ventricular dimensions. A cardiologist experienced with LVAD patients administered the echocardiography measurements.
After pump placement, the pump speed was set low (8,000–9,000 rpm) to allow for the animal to stabilize. After stabilization, the pump speed was incremented manually from a low pump speed that caused minimal LV unloading (7,000 rpm) to high pump speed that caused intermittent suction events (13,000 rpm). At each pump speed, the LV short-axis dimension near the tip of the cannula was measured using echocardiography at end diastole and end systole. Simultaneously, a 10 s conductance waveform was recorded. For each cardiac cycle in the waveform, the end-systolic and end-diastolic conductances were extracted from the minimum and maximum conductance values, respectively. The average and standard deviation of the end-systolic and end-diastolic conductances were averaged for all cycles in the 10 s waveform (typically 10–15). The speed ramp and volume measurement procedure were repeated three times in each study.
Photographs of the cannula tip placement from one of the in vivo studies are shown in Figure 2. The tip was placed centrally in the LV cavity, and adequate spacing can be seen around the tip to ensure proper conductance measurements (Figure 2, A). The wall thickness of the LV myocardium was 1.5, 1.7, and 1.0 cm measured on the septum, free wall, and apex, respectively. The tip of the cannula extended approximately halfway into the LV cavity, as shown in Figure 2, B. The clearance between the bottom electrode and the apical wall was 0.5 cm. The sewing ring was sutured to the epicardium at the apex. The placement was typical of the three in vivo studies. The measured hematocrits for the three animal studies were 21%, 24%, and 26%.
Representative waveforms for the in vivo study are shown in Figure 3 as the pump speed was increased from 8,000 to 13,000 rpm. During the cardiac cycle, conductance increased during diastole as the volume of the ventricle increased and then decreased during systole as the ventricle contracted and volume decreased. At 8,000 rpm (Figure 3, A), there was minimal circulatory support from the pump, as the mean pump flow was 2.5 lpm. The aortic valve opened on ever beat and there was substantial pulsatility in the arterial pressure. At 10,000 rpm (Figure 3, B), mean pump flow increased to 4.3 lpm and the aortic valve continued to open, although arterial pressure pulsatility was reduced. Conductance was reduced at both end systole and end diastole compared with at 8,000 rpm (Figure 3, A). At 12,000 rpm (Figure 3, C), mean pump flow was 6 lpm, and conductance decreased at end systole and end diastole compared with at the slower pump speeds (Figure 3, A and B). The aortic valve was not opening, but there was significant pulsatility in the pump flow indicating that there was sufficient residual volume in the ventricle. However, despite the pulsatile pump flow there was little pulsatility in the arterial pressure. At 13,000 rpm (Figure 3, D), the aortic valve remained closed and there were multiple suction events causing collapse of the ventricle. There was little pulsatility in the pump flow and abnormal rhythms were evident in the LV pressure and conductance waveforms. The overall pump flow was 6.1 lpm, which was not significantly different from the pump flow at 12,000 rpm (Figure 3, C). As seen in Figure 3, D, there was an underlying rhythm in the pressure and conductance waveforms that caused periodic suction events followed by recovery.
The end-diastolic and end-systolic short-axis dimensions measured using echocardiography are shown in Figure 4, A as a function of pump speed for the experiment shown in the Figure 3 waveforms. End-systolic and end-diastolic dimensions decreased with increasing pump speeds. In addition, stroke volume measured from the difference between the end-diastolic and end-systolic dimensions decreased with increasing pump speed. In Figure 4, B the mean end-systolic and end-diastolic conductance values are shown versus pump speed for the same experiment. Similar to the echocardiography dimensions, both end-systolic and end-diastolic conductances decreased with increasing pump speed, as well as the difference between end-systolic and end-diastolic conductance. As noted from the conductance waveform at 13,000 rpm in Figure 3, D, the standard deviation in the end-diastolic conductance was greater at higher pump speeds because of irregularities in the cardiac rhythm and intermittent suction events. The cyclical conductance waveform causes a large variation in the beat-to-beat end-diastolic conductance. However, there was little variation in the end-systolic waveform, which had a small standard deviation across all pump speeds. For all three studies, end-diastolic and end-systolic conductance decreased monotonically with pump speed.
In Figure 5, conductance measured at end diastole and end systole are shown compared with the LV short-axis dimensions for the three studies. In each plot, data are shown for all pump speeds from minimal support to suction. In all three studies, there is good linear agreement between conductance and short-axis dimension over the full range of ventricular unloading (R2 = 0.94, 0.83, and 0.83 for the data in Figure 5, A–C, respectively). The proportionality constant between short-axis dimension and conductance was 12, 10, and 21 mS/cm for the three studies shown in Figure 5, A–C, respectively. The size of the ventricle in the study shown in Figure 5, B was much smaller than the other two studies, with maximum unloading dimension of less than 3 cm compared with 4 cm. Correspondingly, the range of conductance values and ventricular dimensions measured over the pump speed ramp was compressed. The slope of the proportionality constant was much greater in the study shown in Figure 5, C compared with the other two studies. In addition, the standard deviation in the end-diastolic conductance measurements was larger because of greater beat-to-beat variation in the cardiac hemodynamic waveforms compared with the other studies.
The results of this study demonstrate the feasibility of integrating conductance catheter technology onto the inlet cannula tip of a rotary blood pump. The conductance measurements correlated well with LV size over the full range of pump speeds and thus could be used to provide a real-time measurement of LV volume unloading during LVAD support. Setting pump speed is critical for patient management during LVAD support. If the pump speed is too fast suction and ventricular collapse can occur, which can damage the myocardium and lead to arrhythmias. Conversely, if the pump speed is too slow the patient may not receive adequate support and be limited by their heart failure. Echocardiography is the standard-of-care to assess ventricular size and optimize pump speed. The American Society of Echocardiography guidelines recommend LVAD speed optimization every 6 months using a stepwise incremental adjustment to LVAD speed while assessing LV size from echocardiography.15 The measurements from our cannula tip could be used in lieu of echocardiography, which would simplify the procedure and allow for more frequent speed optimizations to be done at outpatient clinics or possibly even remotely.
In addition to assisting clinicians in adjusting pump support, the conductance catheter tip could be incorporated into an automatic control system to adjust pump speed based on LV size. A preload-sensitive control system for LVADs is necessary that can lower pump speed to prevent suction and increase pump speed during exercise to enhance quality of life.8 The lack of such a control system is a major limitation in the long-term care of LVAD patients.16 Ochsner et al.10,11 proposed such a control system that adjusts pump output based on end-diastolic volume, and demonstrated in vivo that the system could mimic the Frank-Starling mechanism of the heart.17 As end-diastolic volume increases because of an increase in filling pressure, the system responds by increasing pump flow. In periods of reduced filling, end-diastolic volume decreases and the system reduces speed and limit suction events. Their system relies on a volume sensor designed for short-term experimental use that is not suitable for long-term implantation. Our proposed catheter tip is designed for long-term implantation and would be a suitable device for implementing such a control system. Furthermore, the National Institutes of Health working group on advancing myocardial recovery with mechanical circulatory support recommends the development of objective measures of myocardial unloading and control strategies to enhance recovery,18 which our device would enable.
As first described by Baan et al.,19 the measured conductance is the parallel sum of the conductance caused by the volume of blood in the ventricle and the conductance caused by the myocardial muscle wall. During the cardiac cycle, the muscle conductance remains relatively constant, and therefore, the measured conductance is directly proportional to changes in the blood volume. Unlike conventional conductance catheters with multiple electrodes spanning from the apex to the base of the ventricle, the conductance catheter implemented on the cannula tip only has one pair of sensing electrodes. Therefore, a true measure of volume is not possible. Instead, the conductance catheter tip provides a measurement based on the localized volume. However, as shown in Figure 5, the measured conductance in our studies was linearly proportional to short-axis dimension, which was unexpected. For a small conductance catheter segment, volume can be approximated by cylinder with fixed length equal to the electrode spacing, and the measured conductance is proportional to the median cross sectional area at the cannula tip location.20 In the current studies, cross sectional area was estimated from the dimension measurement using a circle approximation, and a similar goodness of fit was found comparing conductance versus the approximated area compared with the single-axis dimension (R2 = 0.93, 0.85, and 0.84 using approximated area for the data sets in Figure 5, A–C, respectively). More experimental data and using 2D echocardiography for a more accurate area measurement may provide a better correlation for the conductance measurements. It could also be possible that the significantly larger diameter of our conductance catheter tip compared with conventional devices invalidates the cylindrical volume estimation. Especially during systole as the volume collapses around the tip, the cross sectional area may be better represented as a thin annulus with area proportional to the circumference.
Current standard-of-care using echocardiography for LVAD patients typically provides a single short-axis dimension. Long-axis dimensions are generally unattainable or only attainable at skewed angles because of the presence of the LVAD cannula at the apex. The short-axis dimensions have proved to be sufficient for determining LV unloading and optimizing pump speed. In the same manner, the linear relationship between conductance and short-axis dimension validates the use of the cannula tip as a surrogate for the current practice of using echocardiography for device management.
For a conventional conductance catheter, a proportionality constant and the parallel muscle conductance value must be determined for accurate volume measurements. This is typically done either by injection with a small bolus of hypertonic saline19 or using multiple frequencies to separate muscle from blood based on complex conductance.21 In our current application, the conductance catheter measurements could be calibrated using echocardiography at the time of LVAD implantation and periodically during routine checkup. This routine calibration would account for variations in tip position and angle, as well as for changes in blood conductivity because of hematocrit. Over a hematocrit range of 30–50%, conductivity can vary almost twofold: from 5 to 9 mS/cm.22
The results presented in this the study were from acute testing of the device in an anesthetized animal. The long-term accuracy of the device is unknown and chronic animal testing will be required for validation. Possible sources of error include changes in conductivity because of hematocrit fluctuations,22 changes in tip position during postural changes in an awake animal, and tissue ingrowth and encapsulation of the electrodes. For the acute testing, the device was plumbed in series with a blood pump; however, for chronic testing, the device will need to be incorporated directly into the blood pump design. The tip is designed with a seamless blood contacting surface to minimize the risk of thrombosis. Platinum-iridium electrodes are used because of the success in long-term implantation with pacemakers and defibrillators. Although the current prototype is hard wired to external electronics, space within the tip cavity has been allocated for integrating the electronics and minimizing connections to existing pump controllers. Despite these precautions, the conductance cannula tip may increase the risk of thromboembolic events.
We have designed a cannula tip with integrated conductance electrodes for use with continuous-flow LVADs. In vivo conductance measurements during pump support correlated with ventricular size measured using echocardiography. The volume sensing cannula tip can provide a real-time measurement of ventricular unloading, which will enable the development of a preload-sensitive control system for blood pumps that can respond effectively to changes in patient physiology.
1. Myers TJ, Bolmers M, Gregoric ID, Kar B, Frazier OH. Assessment of arterial blood pressure during support with an axial flow left ventricular assist device
. J Heart Lung Transplant 2009.28: 423–427.
2. Akimoto T, Yamazaki K, Litwak P, et al. Rotary blood pump
flow spontaneously increases during exercise under constant pump speed: results of a chronic study. Artif Organs 1999.23: 797–801.
3. Jacquet L, Vancaenegem O, Pasquet A, et al. Exercise capacity in patients supported with rotary blood pumps is improved by a spontaneous increase of pump flow at constant pump speed and by a rise in native cardiac output. Artif Organs 2011.35: 682–690.
4. Jakovljevic DG, George RS, Donovan G, et al. Comparison of cardiac power output and exercise performance in patients with left ventricular assist devices, explanted (recovered) patients, and those with moderate to severe heart failure
. Am J Cardiol 2010.105: 1780–1785.
5. Jung MH, Gustafsson F. Exercise in heart failure
patients supported with a left ventricular assist device
. J Heart Lung Transplant 2015.34: 489–496.
6. Kugler C, Malehsa D, Tegtbur U, et al. Health-related quality of life and exercise tolerance in recipients of heart transplants and left ventricular assist devices: a prospective, comparative study. J Heart Lung Transplant 2011.30: 204–210.
7. Leibner ES, Cysyk J, Eleuteri K, El-Banayosy A, Boehmer JP, Pae WE. Changes in the functional status measures of heart failure
patients with mechanical assist devices. ASAIO J 2013.59: 117–122.
8. Salamonsen RF, Mason DG, Ayre PJ. Response of rotary blood pumps to changes in preload and afterload at a fixed speed setting are unphysiological when compared with the natural heart. Artif Organs 2011.35: E47–E53.
9. Bullister E, Reich S, Sluetz J. Physiologic control algorithms for rotary blood pumps using pressure sensor input. Artif Organs 2002.26: 931–938.
10. Ochsner G, Wilhelm MJ, Amacher R, et al. In vivo
evaluation of physiologic control algorithms for left ventricular assist devices based on left ventricular volume or pressure. ASAIO J 2017.63: 568–577.
11. Ochsner G, Amacher R, Wilhelm MJ, et al. A physiological controller for turbodynamic ventricular assist devices based on a measurement of the left ventricular volume. Artif Organs 2014.38: 527–538.
12. Petrou A, Ochsner G, Amacher R, et al. A physiological controller for turbodynamic ventricular assist devices based on left ventricular systolic pressure. Artif Organs 2016.40: 842–855.
13. Stevens MC, Gaddum NR, Pearcy M, et al. Frank-starling control of a left ventricular assist device
. Conf Proc IEEE Eng Med Biol Soc 2011.2011: 1335–1338.
14. Cysyk J, Jhun CS, Newswanger R, Weiss W, Rosenberg G. Rotary blood pump
control using integrated inlet pressure sensor. Conf Proc IEEE Eng Med Biol Soc 2011.2011: 373–376.
15. Stainback RF, Estep JD, Agler DA, et al.; American Society of Echocardiography: Echocardiography in the management of patients with left ventricular assist devices: recommendations from the American Society of Echocardiography. J Am Soc Echocardiogr 2015.28: 853–909.
16. Schmid Daners M, Kaufmann F, Amacher R, et al. Left ventricular assist devices: challenges toward sustaining long-term patient care. Ann Biomed Eng 2017.45: 1836–1851.
17. Berne RM. Physiology, 2004, p. 5th ed. St. Louis, Mosby, 1014.
18. Drakos SG, Pagani FD, Lundberg MS, Baldwin TJ. Advancing the science of myocardial recovery with mechanical circulatory support: a working group of the National, Heart, Lung, and Blood Institute. J Thorac Cardiovasc Surg 2017.154: 165–170.
19. Baan J, van der Velde ET, de Bruin HG, et al. Continuous measurement of left ventricular volume in animals and humans by conductance catheter
. Circulation 1984.70: 812–823.
20. Baan J, Jong TT, Kerkhof PL, et al. Continuous stroke volume and cardiac output from intra-ventricular dimensions obtained with impedance catheter. Cardiovasc Res 1981.15: 328–334.
21. Wei CL, Valvano JW, Feldman MD, Nahrendorf M, Peshock R, Pearce JA. Volume catheter parallel conductance varies between end-systole and end-diastole. IEEE Trans Biomed Eng 2007.54: 1480–1489.
22. Visser KR. Electric conductivity of stationary and flowing human blood at low frequencies. Med Biol Eng Comput 1992.30: 636–640.