Continuous-flow left ventricular assist devices (LVAD) have recently been approved for both bridge to transplant and destination therapy in severe left ventricular (LV) failure. Device reliability, size, and simplicity have resulted in rapid dominance of this technology in LVAD with both axial flow and centrifugal flow pumps available.1–3 As these pumps can deliver up to 10 L/min with reduced pulsatility flow, there is a risk of ventricular suction in the absence of adequate LV preload. Pump speeds are often set within a small range with little adjustment of pump parameters postdischarge. We therefore examined the safe working range for pump speed in stable patients with a third-generation continuous-flow LVAD and examined the effect on pump flow rates, cardiac function, and blood pressure (BP).
Thirteen stable patients with VentrAssist continuous-flow LVAD were studied supine. Clinically stable outpatients between March 2008 and March 2009 with more than 2 months postpump insertion and willing to participate were studied. Patients were studied a median of 222 days postimplant (range, 99–841 days). Eleven patients were studied while listed for transplantation, with two patients implanted for destination therapy. The median age was 44 y (range, 14–72 y). There were 11 men and 2 women. The study was approved by the Human Research Ethics Committee of St Vincent's and Mater Health Services, Sydney, and the Alfred Hospital, Melbourne. The trial was registered with the Australia Clinical Trials Registry (ACTRN12608000106336).
Pump output and pump speed were recorded from the Ventrassist controller, and the study was completed with continuous transthoracic echocardiography (Acuson Cypress), noninvasive mean BP, and heart rate monitoring. BP was measured using Doppler flow at the wrist during sphygmomanometric cuff occlusion. While supine, subjects were studied at their preset baseline pump speed. The speed was then decreased from baseline by 50 revolutions per minute (RPM) increments until a prespecified stopping rule was reached or the speed reached 1,800 RPM. After a 5-minute reequilibration period at previous baseline, the speed was increased by 50 RPM every 2 minutes until a prespecified stopping rule was crossed or a maximum of 2,400 RPM was reached. Pump parameters, heart rate, echocardiographic dimensions, aortic valve status (open/closed), and degree of aortic and mitral regurgitation were recorded at the end of each 2-minute period. Pump flow was taken as that reported by the LVAD controller. We have previously reported simultaneous pump flow and thermodilution cardiac output and shown good correlation between the two (r = 0.92, p < 0.0001).4
Prespecified Stopping Rules
For low-pump speed, downward titration was stopped at 1,800 RPM or if there was an increase in LV dimension to >130% baseline end-diastolic dimension, or pump flow decreased to <80% of patient's baseline flow. Downward titration was also stopped with development of symptoms of breathlessness or chest discomfort. For high-pump speed, upward titration was stopped at 2,400 RPM or if LV end-diastolic dimension decreased to <75% baseline, or pump flow increased to >130% baseline. Upward titration was also stopped with the development of any sustained ventricular arrhythmia or sudden change in LV dimension (suckdown).
Results were further compared against contemporaneous invasive hemodynamic studies using Swan-Ganz catheterization via the right internal jugular artery. Systemic vascular resistance and pulmonary vascular resistance were calculated from systemic pressure gradient (mean arterial pressure − right atrial pressure) and transpulmonary gradient (mean pulmonary artery pressure − mean pulmonary capillary wedge pressure) divided by thermodilution derived cardiac output at rest. As in the study by Lee et al.,5 right ventricular (RV) impairment was based on global systolic function and assigned a numeric “score.” For the purposes of the current analysis, anything more than a score >1 (mild RV impairment from Table 1), was classified as impaired.
Repeated measures analysis of variance was used to analyze differences between achieved pump speeds and flow rates. Fisher's exact test was used to analyze difference in the proportion of patients in whom upward titration was stopped due to suckdown. Data are shown as mean ± standard deviation where normally distributed or median (range) in nonnormal distribution. A two-tailed p-value of <0.05 was taken as significant.
Demographic details for the patients are shown in Table 1. All patients were fully anticoagulated (target international normalized ratio 2-3) and received antiplatelet therapy. Eight of 11 were treated with inhibition of renin-angiotensin-aldosterone blockade, 3/11 with beta-blockade, and 7/11 with ongoing diuretic therapy. The average implant duration was close to 1 y, and cause of cardiac dysfunction was evenly split between ischemic and nonischemic etiology. All patients had ongoing severe LV dysfunction, without aortic valve opening at rest.
The baseline pump speed was 2,073 ± 86 RPM (range 1,950–2,200 RPM) and resulted in an estimated flow rate of 5.59 ± 1.18 L/min at rest. Downward titration to 1,800 RPM was achieved in 69% (9/13), with the mean minimum pump speed achieved 1,835 ± 55 RPM. Titration was stopped due to a decrease in flow to <80% baseline in 3/13 patients or due to dyspnea in 1/13. The decrease in pump speed was associated with a significant decrease in estimated pump flows to 4.68 ± 0.99 L/min (p < 0.001).
Upward titration to the projected maximum of 2,400 RPM was achieved in 31% (4/13) with the mean maximum pump speed achieved 2,315 ± 66 RPM. Titration was stopped due to mean BP >105 mm Hg in 3/13, ventricular suction in 2/13, nonsustained ventricular tachycardia (VT) in 2/13, development of moderate-severe aortic regurgitation in 1/13, and a pump flow increase of >130% baseline flow in 1/12. The stopping rule of mean systemic pressure >105 mm Hg was developed in the course of the trial after discussion with coinvestigators. Aortic regurgitation was also not a prespecified outcome, but in the course of the pump titration it was clear that the regurgitation was speed-related and progressive with upward titration. The patient remained asymptomatic despite the increasing aortic regurgitation. The increase in pump speed was associated with a significant increase in estimated pump flows to 6.30 ± 1.30 L/min (p < 0.001). Changes in group pump flow and speed were significant (p < 0.0001 by analysis of variance, Figure 1).
Individual Changes in Pump Flow
As expected, there was a very strong direct linear relationship between change in pump speed and the change in pump flow rate (Figure 2). For each increase in pump speed of 50 RPM, there was a corresponding increase in flow of approximately 190 mL/min over the entire group. When the raw data for each subject is analyzed individually, however, it can be seen that the increase in flow due to a prescribed increase in pump speed is quite variable (Figure 3) with different slopes of the relationship between flow and change in pump speed. Furthermore, although the change in flow in response to speed change is fairly linear for each individual patient (mean r2 for individual correlation 0.92), the range of slopes (with flows converted to mL/min) for each patient ranged between 1.6 and 4.6 mL/min/RPM change. This corresponds to a range in flow change of 80–230 mL/min for a 50 RPM change in pump speed across the group.
When the individual's responses to speed change are compared against their corresponding systemic vascular resistance recorded at contemporaneous hemodynamic measurements, there is a striking differentiation between those with lesser change in flow compared with those with a greater change in flow for the same change in pump speed (Figure 4, upper left compared with lower right; r2 = 0.73, p < 0.005). There is no such relationship between the change in flow response to speed changes compared with pulmonary vascular resistance (r2 = 0.06, p = 0.53) in the same patients. There was no relationship between the flow change and the speed change as a function of baseline flow rate (r2 = 0.08, p = 0.45) or of preload (pulmonary capillary wedge pressure, PCWP, r2 = 0.19, p = 0.25).
In four patients, upward titration of the pump speed resulted in either ventricular suction or VT. In each case, transition between stable flow and suction/arrhythmia was rapid and necessitated immediate speed decrease. All patients returned immediately to normal without sequelae on resetting speed to baseline. Examination of this small cohort of patients suggested that the presence of RV dysfunction was a risk factor for sensitivity to increased pump speeds, as all patients who developed VT or ventricular suction at increased speeds had impaired RV function at rest (Figure 5), although this was not statistically significant (Fisher's Exact test p = 0.07).
Echocardiographic and Hemodynamic Findings
Although increased pump flow was associated with the expected decrease in the pulsatility index (r2 = 0.44, p < 0.0001; Figure 6a), there was no significant change in left ventricular end-diastolic (or end-systolic, data not shown) dimensions in response to the flow change over the short study period (r2 = 0.02, p = 0.08; Figure 6b). The change in pulsatility index at the maximal achieved pump speed (immediately prior to ventricular suction or VT in those cases with events) is shown in Figure 7. It can be seen that there tended to be a larger decrease in pulsatility index in those who developed ventricular suction compared to the rest of the group, and in one of the two who developed VT. The intergroup differences were not significant, most likely related to the small patient numbers. Increased pump flow was associated with a small increase in measured mean arterial pressure (r2 = 0.11, p < 0.001; Figure 6c) and very little change in heart rate (r2 = 0.06; Figure 6d).
We have shown that pump output changes in proportion to pump speed within a relatively small speed range. There are however significant interindividual differences in the response to pump speed changes. Some of the variation in this response may be due to differences in systemic vascular resistance, consistent with the known afterload sensitivity of continuous-flow pumps. In view of the relatively narrow range of speeds over which changes in flow occur, alterations in pump speed should not be made without close monitoring, particularly in the presence of concomitant RV impairment. Short-term monitoring of left ventricular dimensions to predict responses to changes in pump speeds and flows would appear to be unhelpful.
With increasing use of continuous-flow LVADs in bridge to transplant,2,6 as well as in destination therapy1 for severe heart failure, increased understanding of the interaction between ventricular filling and pump function is needed. However, there is limited literature concerning acute changes in cardiac function with adjustment in pump speeds. This is the first report characterizing changes in pump flow, pulsatility parameters, and cardiac function in patients across the entire working pump speed range.
VentrAssist is a third-generation centrifugal continuous-flow LVAD, with proven efficacy in bridge to transplant support.2 It is able to provide circulatory support between 4 and 9 L/min over a narrow pump speed range, usually between 2,000 and 2,200 RPM.7 Assessment of the appropriate speed is made according to the indirect parameters of mean arterial pressure and the pulsatility of the outflow generated by the pump. This pulsatility index is recorded and displayed along with pump output on the patient controller. Although the pulsatility index tended to be lower at maximal achieved speed in those with ventricular suction, only one of the patients with VT had an obviously lower pulsatility index compared with the individual's baseline. In the other patient, the pulsatility index was actually higher (without any increase in pump flow) and suggests that pulsatility index alone may not be a not sufficient marker of impending hemodynamic collapse. Algorithms to minimize risk of suckdown have been proposed8 to adjust pump speed9; however, these algorithms have not been incorporated in the majority of pumps. Animal models and computer simulations may involve excessive time intervals (circulatory collapse of 20 seconds) for human relevance.10 Although the mean relative changes in pump flow from baseline (an increase of 0.71 L/min and a decrease of 0.91 L/min) are not large, only nine patients tolerated minimum pump speed and four patients tolerated maximum pump speed. Only two patients (15%) tolerated the full range of pump speed adjustment from 1,800 to 2,400 RPM.
In the clinical setting, the pump speed is commonly set at an arbitrary speed immediately on insertion and pump speed increased marginally once hemodynamic stability is achieved. Early perioperative RV failure is a major complication associated with LVAD insertion.11,12 This limits the degree to which the pump speed can be increased in the early phase. Although pulsatile pumps are able to increase output through both an increase in cycle rate as well as by variation in displacement in response to changes in preload,13 continuous-flow devices increase output in response to increased preload or decreased afterload.14 Although we have demonstrated that pump speed does affect pump output in the short term, RV dysfunction and excessive increases in mean arterial pressure limit the extent to which routine adjustments should be made for that reason alone.
This study examined short-term responses to sudden changes in pump speed. More gradual changes in pump speed may be able to be tolerated but are not tested by the current protocol. There were no significant changes in left ventricular dimensions in the current study, again most likely related to the short-term period. A further limitation is that the VentrAssist pump is no longer produced; however, device-independent outcomes (such as ventricular arrhythmias and left ventricular suction) remain relevant across all continuous-flow pumps. A further limitation is that the sample rate of the pump flow parameters is only 0.92 Hz, lower than the typical cardiac cycle. Intrabeat hemodynamics are thus not able to be assessed with the current recording parameters. In this outpatient group, the pump pressure gradients (invasive inlet vs. outlet pressures) were not available. Varying pressure gradient within the cardiac cycle and nonsteady state conditions may have contributed to some of the scatter in the echocardiographic and hemodynamic parameters. It has, however, been previously shown ex vivo that estimated VentrAssist pump flows are stable across a broad operating speed range for a stable hematocrit, as is the case in this study.15
Pump flow associated with pump speed changes vary between individuals, which may be related to afterload. Pump speed changes should not be made without close monitoring due to variation in response between patients, particularly in the presence of concomitant RV impairment.
The study was financially supported in part by VentraCor Ltd. Drs. Woodard and Ayre were previous employees of VentraCor.
1. Slaughter MS, Rogers JG, Milano CA, et al
: Advanced heart failure treated with continuous-flow left ventricular assist device. N Engl J Med
361: 2241–2251, 2009.
2. Esmore D, Kaye D, Spratt P, et al
: A prospective, multicenter trial of the VentrAssist left ventricular assist device for bridge to transplant: safety and efficacy. J Heart Lung Transplant
27: 579–588, 2008.
3. Kirklin JK, Naftel DC, Kormos RL, et al
: Second INTERMACS annual report: more than 1,000 primary left ventricular assist device implants. J Heart Lung Transplant
29: 1–10, 2010.
4. Hayward CS, Salamonsen R, Keogh A, et al
: Impact of LVAD pump speed on exercise parameters and invasive hemodynamics. Presented at International Society of Rotary Blood Pumps
, October 2010, Berlin.
5. Lee S, Kamdar F, Madlon-Kay R, et al
: Effects of the HeartMate II continuous-flow left ventricular assist device on right ventricular function. J Heart Lung Transplant
29: 209–215, 2010.
6. Pagani FD, Miller LW, Russell SD, et al
: Extended mechanical circulatory support with a continuous-flow rotary left ventricular assist device. J Am Coll Cardiol
54: 312–321, 2009.
7. Esmore DS, Kaye D, Salamonsen R, et al
: Initial clinical experience with the VentrAssist left ventricular assist device: the pilot trial. J Heart Lung Transplant
27: 479–485, 2008.
8. Takami Y, Otsuka G, Mueller J, et al
: Flow characteristics and required control algorithm of an implantable centrifugal left ventricular assist device. Heart Vessels
12: 92–97, 1997.
9. Choi S, Boston JR, Antaki JF: Hemodynamic controller for left ventricular assist device based on pulsatility ratio. Artif Organs
31: 114–125, 2007.
10. Reesink K, Dekker A, Van der Nagel T, et al
: Suction due to left ventricular assist: implications for device control and management. Artif Organs
11. Ochiai Y, McCarthy PM, Smedira NG, et al
: Predictors of severe right ventricular failure after implantable left ventricular assist device insertion: analysis of 245 patients. Circulation
106(12 Suppl 1): I198–I202, 2002.
12. Matthews JC, Koelling TM, Pagani FD, Aaronson KD: The right ventricular failure risk score a pre-operative tool for assessing the risk of right ventricular failure in left ventricular assist device candidates. J Am Coll Cardiol
51: 2163–2172, 2008.
13. Maybaum S, Williams M, Barbone A, et al
: Assessment of synchrony relationships between the native left ventricle and the HeartMate left ventricular assist device. J Heart Lung Transplant
21: 509–515, 2002.
14. 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
23: 797–801, 1999.
Copyright © 2011 by the American Society for Artificial Internal Organs
15. Lim E, Karantonis DM, Reizes JA, et al
: Noninvasive average flow and differential pressure estimation for an implantable rotary blood pump using dimensional analysis. IEEE Trans Biomed Eng
55: 2094–2101, 2008.