Left Ventricular Assist Device Weaning: Hemodynamic Response and Relationship to Stroke Volume and Rate Reduction Protocols

Abstract

Clinical evidence of myocardial recovery in a small cohort of patients supported with a left ventricular assist device (LVAD) has been reported. Development of an optimal LVAD weaning protocol is needed for these patients to sustain recovery after device explant. In this study, we tested the hypothesis that LVAD stroke volume reduction produces a steady-state mechanical reloading of left ventricular (LV) pressures and volumes compared with LVAD rate reduction that results in transient mechanical reloading of the heart due to beat-to-beat variation in LV pressures and volumes. The relationship of LVAD flow to LVAD stroke volume and systolic interval over a range of LVAD rates (60, 80, 100, 120, and 140 bpm) was validated in a mock circulatory flow loop. In six acute experiments, calves were implanted with a pneumatic paracorporeal LVAD (PVAD, Thoratec, Pleasanton, CA). The PVAD was operated asynchronously in the auto volume mode (full decompression) for 30 minutes to establish a baseline control condition. The calf hearts were then mechanically reloaded by LVAD rate reduction (80, 60, and 40 bpm) or LVAD stroke volume reduction (100, 120, and 140 bpm) protocols consisting of 30 minutes of support at each LVAD beat rate. The order of weaning protocols was randomized with a 30-minute recovery period (LVAD volume mode to fully decompress heart allowing it to rest) between protocols to enable return to baseline control state. Aortic pressure and flow, LV pressure and volume, pulmonary artery flow, and LVAD flow waveforms were recorded for each test condition. The LVAD stroke volume reduction protocol produced steady-state mechanical reloading compared with VAD rate reduction that resulted in transient LV mechanical reloading. This distinction is due to differences in their temporal relationships between LVAD and LV filling and emptying cycles. The acute hemodynamic benefit of LVAD stroke volume reduction was greater reduction in LV end-diastolic pressure and increase in LV segmental shortening than LVAD rate reduction. The long-term effects of steady-state and transient LV mechanical reloading on myocardial structure and function toward achieving sustained myocardial recovery warrant further investigation.

Approximately 30,000 patients are listed worldwide for transplant every year, with only 3500 transplantations performed annually.¹ The limited number of donors for cardiac transplantation and the increased incidence and prevalence of heart failure in an aging and medically complex population has compelled the investigation of alternative treatment strategies. Left ventricular assist devices (LVADs) have been successfully used as a bridge to transplant when survival is threatened because of clinical deterioration before transplantation.² Using a ventricular assist device as a bridge to recovery is a more recently recognized potential application.³

Studies have shown that the profound ventricular pressure and volume unloading provided by LVAD support leads to reverse remodeling as evidenced by genetic, biochemical, histologic, and functional measures.^4–7 Normalization of the left ventricular (LV) pressure volume relationship with regression of LV hypertrophy has been reported even in severe heart failure.⁸ A decrease in myocardial fibrosis and reduction in myocardial matrix metalloproteinase expression may contribute to reverse remodeling.^9–11 Frazier and colleagues⁷ showed a significant improvement in the cardiothoracic ratio, left ventricular end-diastolic dimension, cardiac index, aortic pressure and ejection fraction, and a decrease in pulmonary capillary wedge pressure and pulmonary vascular resistance in 31 patients supported with an LVAD for more than 30 days. A marked reduction in myocytolysis occurred, and calcium uptake, calcium-binding rates, lipid levels, and plasma norepinephrine normalized. Potential mechanisms for the reversal of the remodeling process include decreased myocardial wall stretch, profound volume and pressure unloading, increased myocardial perfusion, neurohormonal normalization, and reduction in cytokine release¹² A noted and most impressive feature of the reported studies of LVAD support is that even the most diseased human hearts exhibit the capacity for profound phenotypic plasticity when undergoing ventricular unloading and neurohormonal stimulation.¹³

Central issues to a successful bridge to recovery include identifying which patients will benefit, which device to implant, the optimal timing and duration of support, identification of markers predictive of myocardial recovery sufficient for LVAD explantation, adjunct medical therapy, and the most beneficial method of reloading the heart during weaning from an LVAD.^14–17 We previously reported a method of weaning patients from LVAD support using the stroke volume reduction technique that allowed for steady-state mechanical reloading of the LV.¹⁸ This technique also allows for a longer period to monitor ventricular function and the opportunity to detect recurrent remodeling, thereby avoiding repeat LVAD implantation. The present study evaluated the weaning process in an animal model to compare the hemodynamic response and temporal relationship of mechanical reloading of the conventional LVAD rate reduction method with the newer method of LVAD stroke volume reduction.

Materials and Methods

Experimental Design

The hemodynamic response and temporal relationship of mechanical reloading by LVAD rate reduction and LVAD stroke volume reduction protocols was investigated in vitro and in vivo. First, LVAD flow, stroke volume, and systolic interval were recorded over the beat rate range for rate reduction (80, 60, and 40 bpm) and stroke volume reduction (100, 120, and 140 bpm) protocols to validate the relationship between LVAD beat rate and flow. Next, the LVAD weaning protocols were conducted in calves with normal heart function to characterize the LV pressure-volume relationship and investigate the hemodynamic response to each technique. In six acute experiments, calves were implanted with a pneumatic paracorporeal LVAD (PVAD, Thoratec, Pleasanton, CA). The PVAD was operated asynchronously in the auto volume mode (full decompression) for thirty minutes to establish a baseline control condition. The calf hearts were then mechanically reloaded by LVAD rate reduction (80, 60, and 40 bpm) or LVAD stroke volume reduction (100, 120, and 140 bpm) protocols consisting of 30 minutes support at each LVAD beat rate. The order of weaning protocols was randomized with a 30-minute recovery period (LVAD volume mode to fully decompress heart allowing it to rest) between protocols to enable return to baseline control state. Aortic pressure and flow, LV pressure and volume, pulmonary artery flow, and LVAD flow waveforms were recorded for each test condition. The recorded waveforms were then used to characterize the LV pressure-volume relationship, temporal relationship of LVAD and LV filling and emptying sequences, and determine the hemodynamic response of each weaning protocol over the respective range of LVAD beat rates.

The study was approved by the University of Illinois Animal Care Committee and all animals received humane care in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publication 85-23, revised 1985).

Mock Circulatory Flow Loop

The LVAD rate reduction and LVAD stroke volume reduction protocols were tested in a mock circulatory flow loop to validate the relationship of LVAD stroke volume and systolic time period to LVAD flow rate over a range of LVAD beat rates comparable to both weaning protocols. A Donovan-type mock loop, consisting of two reservoirs with air chambers that represent the venous and aortic compliance, and a variable clamp to adjust the systemic resistance, was used with a saline-glycerine blood analog solution at a viscosity of 4 cP as the working fluid. The PVAD (Thoratec Corporation) was inserted in the mock circulatory flow loop with apical inflow cannulation from the venous inlet reservoir and outflow cannulation to the mock aortic outlet chamber. Flow was measured with an ultrasound flow probe on the VAD outflow cannula connected to a Transonic flowmeter (Ithaca, NY). Inlet and outlet pressures were measured with Gould pressure transducers. Data were digitized and analyzed on a personal computer. The VAD stroke volume was calculated by integration of the pulsatile flow signal. Arterial blood pressure was adjusted to a mean value of 100 mm Hg, with an inlet pressure of 10 mm Hg. The LVAD flow rate and stroke volume were recorded over a range of LVAD beat rates (60, 80, 100, 120, and 140 bpm) while maintaining the LVAD systolic period at 300 milliseconds. This sequence of recordings enabled validation of LVAD mean flow rates as a function of LVAD beat rate and stroke volume for both weaning protocols.

Acute Animal Study

The PVAD (Thoratec Corporation) was implanted through a left thoracotomy with left ventricular apex to descending aorta cannulation in six healthy Holstein calves (Table 1). Animals were premedicated with xylazine 0.1–0.2 mg/kg IM, followed by thiopental sodium 10 mg/kg IV and isoflurane anesthesia (1–2%) via mechanical ventilation. Blood gases were maintained within normal limits. A stomach tube was inserted to drain the rumen and prevent bloating during the procedure. Pancuronium 2–8 mg IV bolus was given to effect. A lidocaine infusion at 2 mg/min was started before incision and maintained for the duration of the procedure. Noradrenaline acid tartrate (Levophed) 8 mg/250 cc D5W was titrated to maintain mean arterial pressure < 100 mm Hg and > 70 mm Hg (range 4–16 μg/min).

Table 1: Baseline Hemodynamic Data for Healthy Holstein Calves (n = 6) Used in the LVAD Weaning Protocol Study

Continuous hemodynamic monitoring, including an indwelling arterial line, central venous pressure catheter and left ventricular end-diastolic pressure (LVEDP), was performed. A 5-Fr high-fidelity pressure catheter (Edwards Life Sciences, Irvine, CA) was used for measuring the LVEDP. Flow probes (Transonic Systems) were placed around the main pulmonary artery and the outflow graft of the LVAD. Surface electrodes were placed for a three-lead ECG. Piezoelectric crystals (64 MHz, Sonometrics Corporation, London, Canada) were placed epicardially on the base, apex, anterior and posterior walls of the left ventricle and endocardially in the anterior wall as illustrated in Figure 1. Once all monitors were in place, baseline hemodynamics were recorded. All transducers were signal conditioned, analog-to-digital (A/D) converted at a sampling rate of 400 Hz and stored using a medical data acquisition system (Sonometrics Corporation). Measurements made by the piezoelectric crystals are defined in Table 2.

Figure 1.: Illustration of placement of 2-mm, 64-MHz piezoelectric crystals for sonomicrometry measurements (Sonometrics Corporation) of left ventricle (LV) dimensions. Crystals were placed epicardially on the anterior wall for segmental shortening (T2-T3 pair) and endocardially (T1) for wall thickness (T2-T1 pair). Crystals placed epicardially on the posterial (T4), base (T5), and apex (T6) were used to derive LV volume from short (T2-T4) and long (T5-T6) axis dimensions using a cylindrical model approximation.

Table 2: Summary of Left Ventricular Dimension Calculations with Corresponding piezoelectric crystal pairs, as illustrated in Figure 1

Before implant, nonheparinized blood was collected and used to preclot the arterial outflow graft. Calves were than anticoagulated with heparin (300 units/kg IV). Activated clotting times were routinely monitored (maintained at level > 480 seconds). A 14-mm arterial outflow graft was then anastomosed to the descending aorta by end-to-side anastomosis. The left ventricular apex was incised and the apical inflow cannula inserted and secured with eight pledgetted double-armed sutures. The LVAD was then primed with volume, connected to the inflow and outflow cannulae, and de-aired. Once all connections were completed, and before turning on LVAD, baseline hemodynamics (post-LVAD cannulation) were recorded. All LVADs were inserted without using cardiopulmonary bypass.

The LVAD was then set to asynchronous volume mode (full decompression) for 1 hour, enabling the heart to achieve a mechanically unloaded resting state. For all test conditions, mean right atrial pressure of 15–20 mm Hg, mean arterial pressure ≥ 65 mm Hg, and hematocrit ≥ 29% was maintained. In all animals, vasopressors and inotropic support were administered to maintain arterial pressure and right ventricular output. The LVAD rate reduction or stroke volume reduction protocol was then initiated. For the rate reduction protocol, the LVAD was set to asynchronous mode at sequentially decreasing LVAD rates of 80 bpm, 60 bpm, and then 40 bpm. For the stroke volume reduction protocol, the LVAD was set to asynchronous mode at sequentially increasing LVAD rates of 100, 120, and 140 bpm. At each rate, the LVAD operated for 30 minutes while maintaining a 300-m/s systolic period. The order of weaning protocol was randomly selected. Between weaning protocols, the LVAD was programmed to asynchronous volume mode for 30 minutes, enabling the heart to recover back to original full decompression (rest) state. For all baseline and LVAD rates, hemodynamic measurements were recorded during the last minute of each 30-minute test condition.

Hemodynamic Analysis

Hemodynamic waveform recordings were used to calculate landmark cardiovascular parameters using Sonometrics software. Specifically, mean arterial pressure, mean right atrial pressure, left ventricular end-diastolic diameter (LVEDD), left ventricular end diastolic pressure (LVEDP), pulmonary artery flow rate, LVAD flow rate, LVAD beat rate, LVAD stroke volume, and heart rate were calculated on a beat-to-beat basis with all beats averaged to obtain a single representative value. The short (anterior-posterior) and long (base-apex) axes piezoelectric crystals were used to derive a continuous LV volume waveform. The LV volume was approximated using a cylindrical geometric model defined by the equation LVV = PI × Long Axis × Short Axis × Short Axis / Four–Wall Volume. The resulting LV volume and recorded LV pressure waveforms were used to generate LV pressure-volume (PV) loops using m-files developed in Matlab (MathWorks, Natick, MA).

Statistical Analysis

A student’s paired t test (ANALYSE-IT, Leeds, England) was used to discern differences in landmark hemodynamic parameters between LVAD rate reduction and LVAD stroke volume reduction weaning protocols for LVAD beat rates with comparable LVAD flow rates. Statistical significance was defined define as p < 0.05 with 95% confidence.

Results

The mock circulatory flow loop study and the acute animal model demonstrated that weaning from LVAD support (mean flow) can be achieved by either reducing LVAD beat rate or stroke volume (Table 3). The acute animal study demonstrated differences in hemodynamic response and the temporal relationship between native LV and LVAD fill and empty cycles. First, with increasing LVAD beat rate (stroke volume reduction), greater segmental shortening and reduction in LVEDP occurred given comparable LVAD flow rates (Table 4). For example, given a comparable VAD flow of 2.6 l/min for rates of 40 bpm (rate reduction) or 120 bpm (stroke volume reduction), the higher LVAD beat rate resulted in greater segmental shortening (20% vs 12%, p < 0.001) and lower LVEDP (17 vs 21 mm Hg, p < 0.001). There were no discernable differences in wall stress, wall thickness, LVEDD, or other landmark hemodynamic parameters. Second, the effect of reducing LVAD beat rate is random beat-to-beat variation in LVAD and native LV fill and empty pressures and volumes creating a transient temporal relationship. By comparison, increasing LVAD beat rate to reduce LVAD stroke volume creates a steady-state temporal relationship, where the native LV and LVAD maintain more consistent fill and empty cycles. This distinction can be clearly observed in the recorded hemodynamic waveforms (Figure 2). To more clearly illustrate the effect of LVAD beat rate on the left ventricular pressure-volume relationship, PV loops for control (no LVAD), LVAD 40 bpm, and LVAD 120 bpm were created using a computer simulation model (Figure 3).

Table 3: Summary of LVAD Flow Rate (VAD Q) in Relation to LVAD Beat Rate, Systolic Interval, and Stroke Volume for Mock Circulatory Flow Loop and Acute Animal Studies

Table 4: Summary of Differences in Key Landmark Hemodynamic Parameters between LVAD Rate Reduction and LVAD Stroke Volume Reduction Weaning Protocols for the Acute Animal Study (n = 6)

Figure 2.: Upper panel: Electrocardiogram (ECG), arterial pressure (ART), and left ventricular pressure (LVP) waveforms recorded during equal LVAD flow rates of 2.4 l/min with (A) LVAD rate reduction (40 bpm); lower panel: LVAD stroke volume reduction (120 bpm). A transient temporal relationship between native LV and LVAD fill and empty cycles can be seen in the ART and LVP tracings with LVAD rate reduction (upper panel). There are fully loaded beats, fully unloaded beats, and partially loaded beats that equate to sudden increases in LVEDP on a random and unexpected beat-to-beat basis. In contrast, a more consistent steady-state is achieved with LVAD stroke volume reduction (lower panel) as evidenced by a more stable end-diastolic LVP and mean ART.

Figure 3.: Illustration of left ventricular pressure-volume (PV) loops for LVAD rate reduction (40 bpm) and LVAD stroke volume reduction (120 bpm) created using a computer simulation model. The LVAD stroke volume reduction provides more consistent end-systolic and end-diastolic volumes and pressures compared with LVAD rate reduction. Steady-state mechanical reloading of the left ventricle by LVAD stroke volume reduction may be more favorable in the myocardial recovery process toward achieving a sustained recovery of cardiac function.

Discussion

After myocardial recovery, mechanical reloading of the heart by reducing LVAD beat rate may appear to be a logical approach. By contrast, reloading the heart by sequentially increasing LVAD beat rate may seem counterintuitive. However, the mock circulatory flow loop study demonstrated that LVAD flow rate can be reduced by either approach. Importantly, the acute animal study revealed key differences in the acute hemodynamic response and temporal relationship with each weaning protocol, which may have significant effects on the ability to ultimately remove the LVAD for recovery. The LVAD stroke volume reduction protocol (increasing LVAD beat rate) provides a more consistent steady-state for native LV and LVAD fill and empty cycles compared with the LVAD beat-reduction protocol, which creates a more random or chaotic, transient state.

The LVAD stroke volume reduction technique also results in gradual reloading of the ventricle as shown by changes in the LVEDP, which should allow gradual increases in wall stress, wall tension, and work load by the native ventricle. The concept of increasing the rate and decreasing the percent systole was previously tested in controlled trials using a mock circulatory loop.¹⁸ Estimates of the LVAD stroke volume and output from the mock loop were remarkably similar to the in vivo results in this study (Table 4). The mock loop data and subsequently the present animal model demonstrate that by increasing the rate and percent systole, a less abrupt and a more physiologically stable transition to the native ventricle occurs compared with the conventional LVAD rate reduction method.

The methods vary for weaning an LVAD once a patient has met specified criteria for adequate myocardial recovery. The conventional method generally involves an incremental rate reduction to a minimum rate of 40 bpm (2.6 l/min) in an asynchronous mode over a brief period. With rate reduction, the minimum rate is generally 40 bpm because of concerns with thrombus formation within the pump. Also, as shown in the hemodynamic tracings (Figure 2), there are fully loaded beats, fully unloaded beats, and partially loaded beats that equate to sudden increases in LVEDP on a random and unexpected basis. Rapid weaning may contribute to fewer patients tolerating device removal. For example, Mancini and colleagues⁶ report that a retrospective analysis identified only a few patients (5%) in whom the LVAD was successfully explanted. Subsequently, a prospective analysis using hemodynamic, echocardiographic, and exercise measurements at rest and during ventricular loading also revealed only a few patients with adequate myocardial recovery to permit explantation. However, they note that the rapid weaning (down titration of the LVAD rate by 10 cycles per minute with hemodynamic and echocardiographic measurements every 10 minutes until the patient developed symptoms or a rate of 20 cycles per minute was achieved) may not have allowed the neurohormonal system and heart sufficient time to readjust to the reduction in cardiac output or to the increased workload. They also note that the duration of support may have been too short (86–100 days) and that their patients did not receive the benefit of maximal medical therapy, which might have improved explantation rates.⁶

In contrast, Muller and colleagues¹⁰ studied 17 patients in NYHA class IV with nonischemic idiopathic dilated cardiomyopathy who were implanted with an LVAD. Five patients (29%) with significant cardiac recovery were weaned after 160 to 794 days and were device-free at 51 to 592 days. Weaning was performed in the asynchronous mode for 3 consecutive weeks to test the stability of the recovered cardiac function. Anti–β1-adrenoceptor autoantibody disappeared gradually without increase after weaning, and cardiac function and volume density of fibrosis remained normal. Similarly, Hetzer and colleagues¹⁹ report attempted weaning of a LVAD in 28 of 95 patients with nonischemic, idiopathic dilated cardiomyopathy meeting the criteria of improved cardiac performance. Their weaning method was fixed-rate reduction to a minimum rate of 60 bpm within a 2-week period. Successful removal was achieved in 16 of 28 (57%) patients (follow-up 1 month to 5.5 years).

In another method of weaning, Frazier and colleagues²⁰ describe a protocol using dobutamine stress testing and intermittent hand pumping in four cases of successful LVAD explantation. This method involves relatively rapid ventricular reloading using rate reduction and hemodynamic monitoring over a short period of time, but included intensive medical management during LVAD support. A favorable dobutamine stress response has been previously shown to identify successful LVAD explantation cases.²¹

Although not used currently in weaning patients on LVADs, a method has been reported by Chen and colleagues²² that measures left ventricular end-systolic elastance noninvasively. They developed a noninvasive method using arm-cuff blood pressure, echo Doppler cardiography, and the electrocardiogram for assessing systolic function, therapeutic response, and ventricular-arterial interaction. It is possible that this may provide useful clinical information on LV recovery in these patients.

The number of patients who can be successfully weaned from an LVAD with durable ventricular recovery is unknown. To date there has been no uniform or agreed upon method for weaning an LVAD or reliable parameters that predict outcome after weaning and device removal. The best method of weaning patients from LVAD support is not well defined.^14,16 Clearly, LVAD bridge to recovery is complex, but is possible in certain patients. The weaning process is just one of many factors that may relate to the ability to discontinue LVAD support. The present study demonstrates hemodynamic and temporal relationship advantages of the LVAD stroke volume reduction method of weaning. Additional studies in a chronic animal model of heart failure are needed to determine the effects of weaning strategies on myocardial structure and function for successful device removal and sustained recovery of cardiac function.