Author + information
- Received January 6, 2014
- Revision received January 31, 2014
- Accepted February 3, 2014
- Published online August 1, 2014.
- Kavitha Muthiah, MBBS,
- Sunil Gupta, BMedSc,
- James Otton, MD,
- Desiree Robson, RN,
- Robyn Walker, RN,
- Andre Tay, RN,
- Peter Macdonald, MD,
- Anne Keogh, MD,
- Eugene Kotlyar, MD,
- Emily Granger, MBBS,
- Kumud Dhital, MBBCh, DPhil,
- Phillip Spratt, MBBS,
- Paul Jansz, MD and
- Christopher S. Hayward, MD∗ ()
- Heart Failure and Transplant Unit, St. Vincent’s Hospital and Victor Chang Cardiac Research Institute, Sydney, Australia
- ↵∗Reprint requests and correspondence:
Dr. Christopher S. Hayward, Heart Failure and Transplant Unit, St. Vincent’s Hospital, Victoria Street, Darlinghurst NSW 2010, Australia.
Objectives The aim of this study was to determine the contribution of pre-load and heart rate to pump flow in patients implanted with continuous-flow left ventricular assist devices (cfLVADs).
Background Although it is known that cfLVAD pump flow increases with exercise, it is unclear if this increment is driven by increased heart rate, augmented intrinsic ventricular contraction, or enhanced venous return.
Methods Two studies were performed in patients implanted with the HeartWare HVAD. In 11 patients, paced heart rate was increased to approximately 40 beats/min above baseline and then down to approximately 30 beats/min below baseline pacing rate (in pacemaker-dependent patients). Ten patients underwent tilt-table testing at 30°, 60°, and 80° passive head-up tilt for 3 min and then for a further 3 min after ankle flexion exercise. This regimen was repeated at 20° passive head-down tilt. Pump parameters, noninvasive hemodynamics, and 2-dimensional echocardiographic measures were recorded.
Results Heart rate alteration by pacing did not affect LVAD flows or LV dimensions. LVAD pump flow decreased from baseline 4.9 ± 0.6 l/min to approximately 4.5 ± 0.5 l/min at each level of head-up tilt (p < 0.0001 analysis of variance). With active ankle flexion, LVAD flow returned to baseline. There was no significant change in flow with a 20° head-down tilt with or without ankle flexion exercise. There were no suction events.
Conclusions Centrifugal cfLVAD flows are not significantly affected by changes in heart rate, but they change significantly with body position and passive filling. Previously demonstrated exercise-induced changes in pump flows may be related to altered loading conditions, rather than changes in heart rate.
The use of continuous-flow left ventricular assist devices (cfLVADs) in the management of end-stage heart failure is the standard of care in many tertiary centers globally, both as bridge to transplant and destination therapy (1–3). Refinement in pump design has led to more widespread use of the newer-generation axial and centrifugal continuous-flow pumps (2,4–6). Because these pumps continuously drain the LV, they are sensitive to changes in pre-load and excessive pump speeds can result in suction events (7).
Current cfLVADs are set at a fixed pump speed determined at implant to provide sufficient delivery of blood from the LV (8). At constant pump speed, pump flow is determined by pre-load and afterload, systemic venous return, and arterial pressure (9). It has been shown in several studies that LVAD pump flows with continuous-flow devices increase in response to exercise (10–14). It is unknown the extent to which the increase in flow observed with exercise is due to an increase in heart rate or to changes in pump-loading conditions determined by venous return.
To further investigate the separate contributions of heart rate and venous return, we examined estimated pump flow (l/min) and noninvasive hemodynamics at constant pump speed in stable cfLVAD patients under resting conditions. Although it has been hypothesized that increased heart rate (even in the setting of a closed aortic valve) may improve LV pre-load provided by the right ventricle (RV), whether this is relevant has not previously been examined. If it were found that increased heart rate enhanced LVAD pump flow, this would have major implications in ongoing management in this growing field.
Stable adult patients implanted with the continuous-flow HeartWare HVAD (HeartWare, Inc., Framingham, Massachusetts) were prospectively enrolled in 2 parallel studies. All patients were recruited from a single tertiary center (St. Vincent’s Hospital, Sydney, Australia). These patients had New York Heart Association functional class IV symptoms pre-implant and were in Interagency Registry for Mechanically Assisted Circulatory Support profile I or II before LVAD implant and were on inotropic support and/or intra-aortic balloon pump/venoarterial extracorporeal membrane oxygenation. Criteria for study participation included being ambulatory for at least 3 months post-LVAD implantation. Patients were excluded from enrollment if they had sepsis requiring intravenous antibiotics.
The study was conducted according to the Declaration of Helsinki and Note for Guidance on Good Clinical Practice. The study was approved by the Human Research and Ethics Committee of St. Vincent’s Hospital (Sydney, Australia). All patients provided written informed consent.
Patients were studied supine in the heart and lung transplantation ambulatory care unit. Heart rate and baseline mean arterial blood pressure (MAP) were measured using transcutaneous arterial Doppler ultrasound (811B, Bosco Medical, Murarrie, Australia) and a cuff sphygmomanometer. LVAD pump flow (l/min), power (W), and speed (rpm) were recorded. Standard transthoracic echocardiography (Acuson 128XP/5 system, Siemens Medical Solutions, Bayswater, Australia) was performed and M-mode measurements taken from the parasternal long-axis view. LV end-diastolic and end-systolic dimensions, status of aortic valve opening, and degree of aortic and mitral regurgitation were recorded. RV function was independently assessed on resting transthoracic echocardiography.
Pacemaker/defibrillator device interrogation and adjustment were performed by a trained device technician. Pacing settings were then increased by 10 beats/min above the baseline heart rate and continued to be increased by 10 beats/min every 2 min up to 40 beats/min above the baseline heart rate. All noninvasive, pump hemodynamic and echocardiographic measurements were repeated at each interval. The presence of LV “suck down” defined by a fall in the LV end-diastolic dimension by 80% below baseline was an indication to discontinue. All measurements were then repeated at the baseline heart rate, with device setting returned to baseline.
In patients who were pacing dependent, pacing settings were adjusted to 10 beats/min below the baseline heart rate every 2 min down to 30 beats/min below the baseline or until the minimum pacing rate was achieved (40 beats/min). Noninvasive, pump hemodynamic and echocardiographic measurements were repeated. Pacing settings were then adjusted back to the baseline settings with all measurements repeated.
Patients were studied in the noninvasive testing laboratory of the coronary care unit; patients fasted for no more than 2 h prior to the procedure to avoid the confounding effects of relative dehydration and hypotension. Patients were then strapped onto a tilt-table supported around the waist and rested supine for 5 min prior to commencement of testing. Continuous electrocardiographic recordings of heart rate were obtained. Baseline MAP and LVAD pump flow, power, and speed were recorded. Standard transthoracic echocardiography was also performed, as for the pacing study.
The tilt-table was positioned at 30° head-up for 3 min. Patients were instructed to avoid any movement of the lower limbs to maximize venous pooling. Patients were then asked to perform continuous active ankle flexion for a further 3 min. This was performed in full dorsiflexion and plantarflexion, alternating between right and left, with weight fully supported by the abdominal belt. The aim of the ankle flexion was to activate the muscle pump, without inducing significant aerobic effort. All patients were able to complete the ankle flexion for 3 min without stopping. Studies were repeated at 60° and 80° head-up tilt. The LVAD pump controller was supported independently of the patient. Patients were then returned to supine position for 3 min followed by 30° head-down tilt testing, with further ankle flexion. All noninvasive, pump hemodynamic and echocardiographic measurements were repeated at each interval. A drop in MAP of >20 mm Hg, sustained ventricular tachycardia, or evidence of symptomatic “suction” events was absolute indication for study termination.
Pump flow versus continuous cardiac output estimation
The HVAD pump flow estimate was compared against invasive continuous cardiac output thermodilution estimate using a continuous cardiac output (CCO) Swan-Ganz catheter (Edwards Lifesciences, Irvine, California) in a separate cohort of 11 patients studied at rest, during increased pump speed and again at baseline. The aortic valve opened in 2 patients. The mean flow was 5.53 ± 1.98 l/min by CCO and 5.46 ± 1.12 l/min by HVAD pump flow estimate (p = 0.82).
All statistical analysis was carried out using EZ-Analyze software version 3.0 (Tim Poynton, Boston, Massachusetts). Data were visually checked for outlying values. Continuous normally distributed parameters are presented as mean ± SD and compared using the Student t test or paired t test. Measure of variance was analyzed using repeated-measures analysis of variance with post hoc Bonferroni correction. Values of p < 0.05 indicated statistical significance.
Eleven patients (8 men and 3 women; mean age of 52.6 ± 9.9 years) were enrolled to investigate the direct relationship between heart rate and estimated pump flow (Table 1). Six patients had pre-existing right heart failure with mild or greater impairment in resting RV systolic function on transthoracic echocardiography.
Effect of pacing
The median pump speed setting of the studied patients was 2,600 rpm. All 11 patients were successfully paced up to 40 beats/min above the baseline heart rate. Atrioventricular (AV) pacing was performed in the 9 patients who had right atrial leads to maintain AV synchrony. Five of the 11 patients studied were pacing dependent, allowing the pacing rate to be decreased to 30 beats/min below the baseline or until the minimum pacing rate was achieved. One patient was in atrial fibrillation. All patients tolerated the changes in paced ventricular rates, and there were no short or long-term adverse events.
Changes in pump flow recordings, MAP, and echocardiographic measurements with the changes in paced ventricular rates are provided in Tables 2 and 3⇓. Significant increments (32% increase) and decrements (29% decrease) in heart rate were achieved at both maximum and minimum paced ventricular rate compared with baseline: 86.5 ± 17.3 beats/min versus 126.8 ± 17.6 beats/min (p < 0.001; baseline vs. maximum paced ventricular rate, n = 11); 76.0 ± 11.4 beats/min versus 53.6 ± 17.3 beats/min (p = 0.04; baseline vs. minimum paced ventricular rate, n = 5). Despite this, there was no significant change in pump flows (baseline 5.21 ± 1.3 l/min; maximum heart rate 5.17 ± 1.2 l/min; p = 1.00). There was also no significant change in flow rates at each 10-beats/min increment interval. Decreasing heart rate through reduction in the pacing rate did not change pump flow (baseline 4.6 ± 0.6 vs. minimum paced rate 4.5 ± 0.5 l/min; p = 0.74). There was no significant change in flow rate at each 10-beats/min decrement interval. MAP, LV dimensions, and aortic valve opening status were not affected by heart rate (Tables 2 and 3).
We compared changes in pump flow between those with open and closed aortic valves. There were insignificant differences in pump flow at maximum and minimum paced ventricular rates in both groups, closed AV (baseline 4.5 ± 0.4 l/min, maximum heart rate 4.5 ± 0.4 l/min, p = 1.00; baseline 4.3 ± 0.4 l/min, minimum heart rate 4.6 ± 0.6 l/min, p = 0.43) and open AV (baseline 6.2 ± 1.4, maximum heart rate 6.7 ± 0.6 l/min, p = 0.55). In 1 patient, the aortic valve status was noted to be open at baseline and only open intermittently at the increased pacing rate. This patient’s underlying rhythm was atrial fibrillation, and the device was an implantable defibrillator. The patient was not pacemaker dependent. Even at the higher pacing rate, it was noted that the patient still had intermittent intrinsic conduction due to atrial fibrillation. It is likely that the variable filling time contributed to the intermittent valve opening. A second patient went from “open” aortic valve status at baseline to “closed” at maximal paced ventricular rate (120/min). There was no change in noninvasive MAP in that patient (74 mm Hg at both heart rates). The pump flow decreased slightly with the aortic valve opening (4.6 l/min down to 4.1 l/min), consistent with parallel flow through both the aortic valve and the pump. It is likely that this patient’s very poor intrinsic ventricular function (estimated ejection fraction of 10% at rest despite LVAD unloading) relied on an adequate filling time to allow significant contraction. There was also evidence of mild RV impairment which may have further limited LV filling at the higher heart rate.
There was also no difference in pump flows, MAP, and LV dimensions in those who were AV paced compared with those who were V paced. We examined the impact of resting RV impairment on the response to pacing at either increased or reduced ventricular rates and did not observe any difference in response (Fig. 1). Although the baseline pump flows were slightly lower in those with normal resting RV function compared with impaired RV function (4.4 ± 0.4 l/min vs. 5.7 ± 1.3 l/min), this was not significant (p > 0.05) and was likely related to chance. The apparent difference in pump flows was decreased after adjustment for body size.
Ten patients (9 men and 1 women, mean age of 56.1 ± 11.1 years) were enrolled in the substudy investigating the effect of passive venous return on estimated pump flow (Table 4). Six patients had evidence of mild or greater resting RV impairment on transthoracic echocardiography. Two patients who had completed the head-up tilt part of the study did not complete head-down tilt due to a history of intracerebral bleeding events in the prior 12 months.
Effect of venous return
The aortic valve remained closed in 8 patients and was intermittently open in 2 patients; this did not change during the study. Pump flow decreased significantly from baseline (4.9 ± 0.6 l/min) with passive head-up tilt (30°/60°/80° 4.5 ± 0.5/4.5 ± 0.4/4.5 ± 0.5 l/min, respectively; p < 0.0001 analysis of variance). With 3 min of continuous active ankle flexion, LVAD flow returned to baseline values (4.9 ± 0.4/5.0 l/min ± 0.4/5.3 ± 0.7 l/min, respectively) (Fig. 2). With continuous active ankle flexion, pump flows returned to baseline values (flow increase p < 0.05 for each angle) (Fig. 2). There was no significant change in flow from baseline (5.2 ± 0.7 l/min) with a 20° head-down tilt with and without ankle flexion exercise (5.3 ± 0.8 l/min; p = 0.45) (Fig. 2). Heart rate did not change significantly with the passive 30°, 60° head-up tilt, 20° head-down tilt from baseline but increased significantly with 80° head-up tilt (baseline 78.2 ± 21.2 beats/min vs 81.7 ± 24.2 beats/min at 80°; p = 0.02) (Fig. 3). There was a slight rise in MAP in comparison with baseline only with 80° head-up tilt with ankle flexion (79.8 ± 8.6 mm Hg vs. 88.0 ± 13 mm Hg; p = 0.01), but this was not significant on post hoc Bonferroni correction (p = 0.22) (Fig. 4). The degree of aortic and mitral regurgitation did not change significantly through all study stages. LV dimensions did not change significantly, and there were no suction events.
The impact of resting RV function on responses to passive tilt was also examined (Fig. 5). There was a tendency toward more prominent decreases in pump flows with head-up tilt and more prominent increases in flows with head-down tilt in patients with impaired resting RV function. Due to small patient numbers in the subgroup analyses, these changes were not statistically significant (p = 0.06 and p = 0.15 for changes in flow in response to maximum head-up and head-down tilt, respectively).
Previous studies have shown that exercise is associated with increased centrifugal continuous-flow left ventricular assist flow output despite unchanged pump speeds. Whether this is due to the associated changes in loading or to the associated increase in heart rate has not been previously tested. We report 2 clinical studies to observe the effects of alterations in ventricular pre-load and heart rate on cfLVAD parameters and physiological responses. We found that pacing-induced changes in heart rate did not impact cfLVAD function or noninvasive hemodynamics. Changes in passive venous return were, however, associated with changes in pump flow.
A number of studies have shown that estimated pump flows increase during exercise despite constant pump speed in patients with cfLVADs (9,10,13). A confounding factor in all of these studies is the significant increase in heart rate and blood pressure during exercise. In the paper by Brassard et al. (10), heart rate increased by 42% with exercise and was associated with a 49% increase in cardiac output. We previously showed a relative increase of 14% in flows during supine bicycle exercise in our cohort of LVAD patients supported with centrifugal pumps, in association with marked increases in heart rate (41%) (15). We also demonstrated the changes in pump hemodynamics relating to level of activity, with determinants of flow change relating to sleep (or a supine state), energy expenditure, and skin temperature with changes in heart rate occurring in parallel to activity (16).
It has been accepted that increased heart rate is the main driver for the increase in flow in these settings, largely based on the clear statement from Akimoto et al. (13) that “mean pump flow during exercise under constant pump speed was caused by an increase in heart rate.” It is important to recognize, however, that neither venous return nor atrial pressure was measured in the Akimoto et al. (13) seminal paper. Because both heart rate and venous return increase with exercise, the current study was undertaken to examine these aspects separately (outside the context of exercise) because the collinearity of the 2 variables makes interpretation difficult. The current study challenges the heart rate–dependent dogma of pump flow and exercise and emphasizes venous return rather than heart rate as an important determinant of pump flow. Indeed, as seen in Figure 3, at the maximum passive head-up tilt (80°), there was a marginal increase in heart rate compared with baseline; however, this was associated with a very significant reduction in pump flow compared with baseline (Fig. 2). In the head-down cohort, there was no augmentation of pump flow by ankle flexion and no effect of a slight increase in heart rate on pump flows (Figs. 2 and 3).
It is known that patients with end-stage cardiomyopathy have activation of the sympathetic nervous system and resultant beta-adrenoceptor down-regulation (17–19). This contributes to ongoing chronotropic incompetence and abnormal heart rate recovery (20–22). Even in LVAD patients, where systemic blood flow is returned, Dimopoulos et al. (23) demonstrated the persistence of cardiac autonomic dysregulation post-implantation of cfLVADs. Given the current heart rate findings from this study, any concern about ongoing chronotropic incompetence in the setting of cfLVADs can be allayed. This is also reassuring when considering the maximization of beta-blocker therapy post-LVAD, recognized to help with the myocardial reverse-remodeling process in the setting of mechanical unloading (24).
Our study showed a significant change in centrifugal pump flow at fixed speed with changes in body position due to passive venous return. The significant fall in pump flows at 30° head-up tilt was maintained at the higher tilt angles with no further decrement. Flows returned to baseline with the relatively minor exercise of continuous ankle flexion. This study suggested that the majority of venous pooling occurs at only 30° in passive head-up tilt. More pronounced venous pooling may have occurred had the tilt been maintained longer. Measured LV and RV dimensions remained constant, and there was no evidence of LV suction due to the LVAD at higher tilt angles. Although decreased venous return may have released some pericardial constraint due to the enlarged heart size, the fact that pump flows still fell at the greatest head-up tilt suggests that this is not a major factor in determining pump flows.
Centrifugal cfLVADs are sensitive to changes in afterload, with even modest changes in systemic vascular resistance impacting the magnitude of pump flow. Our patient cohort maintained constant MAP. Heart rate variation was only significant at the 80° head-up tilt, possibly due the interplay of feedback mechanisms from peripheral baroreceptors in the setting of decreased pump flow. Furthermore, the status of the aortic valve opening did not change during each stage of tilt-table testing, further eliminating the confounding effect of ejection of blood flow through the aortic valve. In light of this, it is not expected that any changes in contractility due to the low-level exercise of ankle flexion would have had any significant effect on the augmentation of flow seen and that the return of pump flows to baseline was due to the effects of the muscle pump restoring venous return. A further postulated mechanism for pump flow responsiveness independent of exercise is the role of increased intrinsic RV and LV contraction in “priming” the LV and the pump pre-load, respectively. It is possible that in the setting of impaired RV function, pump flows may respond differently to changes in heart rate or venous return. On review of our studied patients, we found that 6 patients in both the pacing study and tilt-table study had impaired RV function. As shown in the subgroup analyses, results were not difference in the impaired RV group compared with those with normal RV function. It is likely that, in the setting of normalized pulmonary vascular resistance (in the setting of long-term stable LVAD patients), RV function may not contribute significantly to pump flow at rest. In our study, there did seem to be a tendency to increased sensitivity to passive venous return in the impaired RV function group, but this will need to be confirmed in a larger cohort.
Our study was a noninvasive study; hence, measurement of central hemodynamics was not performed. Catheter measurements of central venous pressures may provide a better reflection of pre-load, especially in studying the effect of body position on flow dynamics. Pump flow measurements were estimated automatically based on LVAD power consumption, impeller speed, and the patients’ hematocrit. The intensity of ankle flexion exercise was not objectively measured. Patients were given verbal encouragement to keep ankle exercise going for the entire 3 min. Subgroup analyses according to RV function resulted in small patient groups for study. The effects of RV function on filling and heart rate responses in cfLVAD patients may need to be confirmed in a larger study.
Centrifugal cfLVAD flow was not affected by heart rate variation but was importantly affected by changes in body position and passive venous return, when pump speed remained constant. These findings support the concept of changes in pre-load as a contributor to exercise-related changes in pump output.
Dr. Hayward has received research support from HeartWare, Inc. Dr. Dhital has served on advisory boards for Bayer. Dr. Spratt owns stock in HeartWare, Inc. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- continuous-flow left ventricular assist device
- mean arterial blood pressure
- right ventricle/ventricular
- Received January 6, 2014.
- Revision received January 31, 2014.
- Accepted February 3, 2014.
- American College of Cardiology Foundation
- Strueber M.,
- O’Driscoll G.,
- Jansz P.,
- Khaghani A.,
- Levy W.C.,
- Wieselthaler G.M.
- Aaronson K.D.,
- Slaughter M.S.,
- Miller L.W.,
- et al.
- Park S.J.,
- Milano C.A.,
- Tatooles A.J.,
- et al.
- Pagani F.D.,
- Miller L.W.,
- Russell S.D.,
- et al.
- Hayward C.S.,
- Swartz M.
- Brassard P.,
- Jensen A.S.,
- Nordsborg N.,
- et al.
- Andersen M.,
- Gustafsson F.,
- Madsen P.L.,
- et al.
- Muthiah K.,
- Robson D.,
- Pritchard R.,
- et al.
- Imai K.,
- Sato H.,
- Hori M.,
- et al.
- Colucci W.S.,
- Ribeiro J.P.,
- Rocco M.B.,
- et al.
- Drakos S.G.,
- Wever-Pinzon O.,
- Selzman C.H.,
- et al.