Secondary Logo

Journal Logo

Determinants of Improvement in Cardiopulmonary Exercise Testing After Left Ventricular Assist Device Implantation

Rosenbaum, Andrew N.*; Dunlay, Shannon M.*,†,‡; Pereira, Naveen L.*,†; Allison, Thomas G.*; Maltais, Simon§; Stulak, John M.§; Joyce, Lyle D.§; Kushwaha, Sudhir S.*,†

doi: 10.1097/MAT.0000000000000693
Adult Circulatory Support

Although improvement in cardiac output and hemodynamic parameters is routinely demonstrated in patients implanted with continuous-flow left ventricular assist devices (CF-LVADs), improvement in exercise capacity is inconsistently seen. Our purpose was to determine whether native cardiac factors, LVAD factors, or comorbid factors were associated with lack of improvement. Review of all patients undergoing preimplant cardiopulmonary exercise testing (CPET) and CPET on LVAD therapy at one institution was performed between 2007 and 2014 (n = 49). Comprehensive assessment of echocardiographic parameters, right heart catheterization data, medications, and comorbid illness was undertaken. There was no mean improvement in peak oxygen consumption (VO2; 11.8–12.4 ml/kg/min; p = 0.26), although exercise time (5.1 [46% predicted] to 5.8 min [56% predicted]; p = 0.02) and nadir of the ratio of minute ventilation to carbon dioxide production slope (VE/VCO2; 39–36; p = 0.001) improved. Factors most strongly associated with improvement in VO2 were Heartmate II pulsatility index (PI; R = 0.48; p = 0.001), power (R = −0.40; p = 0.009), pump flow (R = −0.40; p = 0.008), and pump speed (R = −0.32; p = 0.04). Peak heart rate (HR) was also associated with improvement in VO2 (R = 0.41; p = 0.004). Left ventricular ejection fraction (LVEF; R = 0.004; p = 0.77), right ventricular (RV) function (R = 0.22; p = 0.28), and aortic valve opening (R = 0.20; p = 0.57) were not associated with improvement in VO2. Our data suggest that less reliance on LVAD support was modestly associated with improvement in exercise capacity. Further studies should seek to determine the optimal level of device support prospectively in relation to exercise capacity.

From the *Department of Cardiovascular Diseases, Mayo Clinic, Rochester, Minnesota

William J von Liebig Center for Transplantation and Clinical Regeneration, Mayo Clinic, Rochester, Minnesota

Department of Health Sciences Research, Mayo Clinic, Rochester, Minnesota

§Department of Cardiovascular Surgery, Mayo Clinic, Rochester, Minnesota.

Submitted for consideration April 2017; accepted for publication in revised form September 2017.

Disclosures: Dr Kushwaha received a grant from Thoratec to study endothelial function. The other authors have no conflicts of interest to report.

Correspondence: Sudhir S. Kushwaha, Mayo Clinic Rochester, 200 1st St SW, Rochester, MN 55905. Email:

As the number of implanted left ventricular assist devices (LVADs), especially continuous-flow devices (CF-LVADs) approaches 13,000 globally in the interagency registry for mechanically assisted circulatory support (INTERMACS) registry for both destination therapy (DT) and bridge to transplant (BTT) indications, the necessity to understand the patient–device interaction grows.1 One of the most important patient-centered outcomes after LVAD implant is exercise capacity (peak oxygen consumption [VO2]), which is not only associated with decreased risk of adverse events in this population but also can reflect improved function and greater independence.2,3 Exercise capacity is assessed typically on cardiopulmonary exercise testing (CPET), where improvement in LVAD patients has been inconsistently demonstrated.4,5 Rigorous studies of cardiopulmonary fitness comparing patients with LVADs to patients with medically managed severe heart failure suggest that LVAD therapy may augment some parameters of exercise physiology,6–8 although studies of patients before and after LVAD demonstrated no augmentation in peak VO2 after implantation.5,9 This somewhat unexpected observation can be puzzling given the clear augmentation in cardiac index and improved filling pressures observed after LVAD placement, as well as perceived improvement in heart failure symptoms.

The mechanism for the lack of consistent improvement in CPET testing, specifically, peak VO2 after LVAD implant, has not been rigorously explored. We sought to expand on our previous observations regarding the effect of LVAD implant on exercise capacity and determine factors that may contribute to an improvement in exercise capacity post-implantantation.5

Back to Top | Article Outline

Materials and Methods

This study was a retrospective observational cohort study. We examined the factors accounting for differences in exercise capacity observed after LVAD implantation. We examined all patients implanted with contemporary CF-LVADs in the current era (2007–2014) at the Mayo Clinic in Rochester, Minnesota. Patients were implanted with either Heartmate II (HM2; Thoratec, Co, Pleasanton, CA) or HeartWare (HW; HeartWare International, Inc, Framingham, MA). Patients were included if they had preimplantation CPET and a postimplant CPET. Forty-nine patients met criteria for the study. This study was approved by the Mayo Clinic Institutional Review Board.

Back to Top | Article Outline

Cardiopulmonary Exercise Testing

CPET is routinely performed pre-LVAD as a part of the LVAD evaluation process in ambulatory patients. Since 2010, CPET is recommended in our program as a part of the first annual follow-up visit after LVAD. Cardiopulmonary exercise testing is performed via an accelerated Naughton protocol and performed as described previously.5 Peak VO2 and nadir minute ventilation to CO2 production (VE/VCO2) slope were measured using Medical Graphics metabolic cart (St. Paul, MN). Values were expressed in absolute terms and as a percentage of predicted. Percent change in VO2 was calculated by the following formula: VO2[LVAD] − VO2[pre]/VO2[pre]*100%. Patients were encouraged to exercise to full capacity, as judged by the Borg Rating of Perceived Exertion scale. When more than one CPET was performed pre-LVAD implantation, we included the one closest to the LVAD implantation date. Post-LVAD implantation, we included the CPET performed closest to 1 year after LVAD implantation.

Back to Top | Article Outline

Other testing

Pre-LVAD implantation, we included data obtained from the transthoracic echocardiogram performed closest to the date of CPET. Left ventricular end-diastolic diameter (LVEDD) and left ventricular ejection fraction (LVEF) were obtained via 2D measurement. Interventricular septum position and right ventricular (RV) function were obtained via 2D assessment. The latter was a semi-quantitative assessment incorporating lateral tricuspid annulus tissue Doppler velocity and tricuspid annular plane systolic excursion indices if available, but grade was ultimately at the discretion of the echocardiographer. Aortic valve opening was judged over five cardiac cycles, with and without color-flow imaging; results are reported based on frequency of opening. Six min walk tests (6MWT) were conducted in all patients before and after as able. We also included data from right heart catheterizations (RHC) performed pre-LVAD implantation (closest to the pre-implant CPET date) and post-LVAD implantation (closest to post-LVAD CPET). Pulmonary capillary wedge pressures (PCWP), pulmonary artery (PA) pressures, RV pressure and right atrial (RA) pressure, and cardiac output data (Fick or thermodilution) were obtained. Pulmonary vascular resistance (PVR) was calculated as PVR divided by body surface area. Left ventricular assist device parameters were assessed on the day of CPET (device-obtained speed [rpm], flow [L/min], Power [W], and pulsatility index [PI, HM2 only]).

Back to Top | Article Outline

Comorbid Illness

Comorbidities were assessed via chart review. Unless otherwise indicated, comorbid conditions were assigned based on clinical diagnosis. Type 2 diabetes mellitus (T2DM) was also assigned in the presence of an elevated A1c (>6.5%) or use of antihyperglycemic medications. Obstructive and restrictive lung disease diagnoses were also assigned based on results of pulmonary function tests when available. Chronic kidney disease was also assigned based on the presence of repeatedly demonstrated elevated creatinine with low estimated glomerular filtration rate (<60 ml/min/body surface area (BSA)). Medications were as documented at the visit the day of CPET.

Back to Top | Article Outline

Statistical Analysis

Data analysis was conducted using JMP version 10 (SAS Institute, Inc, Cary, NC). Differences in continuous variables pre- and post-implantation were compared using matched pair t-tests. Linear regression was used to evaluate the association between predictor variables with change in peak VO2 over time, as this variable was found to have a normal distribution. Candidate variables considered for the analysis were demographics (age, body mass index [BMI], implant intention), echocardiographic parameters (aortic valve opening, septal position, RV function, outflow cannula velocity, and LVEF%), RHC variables (PCWP, RA pressure, and cardiac index), LVAD parameters, certain comorbidities (atrial fibrillation, T2DM, lung disease, and chronic kidney disease), and medications (β-blockers and amiodarone). Univariate least squares regression analysis was performed, and strength of correlation and analysis of variance to test significance are given for categorical variables.

Back to Top | Article Outline


There were 49 patients with CPET performed both pre-implantation and while on LVAD therapy. In total, 82% were male, and etiology of cardiomyopathy was nonischemic in 29 (59%) and ischemic in 20 (41%). There were 43 (88%) HeartMate II devices and six (12%) HeartWare devices implanted. Average age at implantation was 63 ± 10 years. Left ventricular assist device was implanted as DT in 33 patients (67%), whereas 16 (33%) were implanted as BTT. The mode INTERMACS score was four (47%). Median time from CPET pre-implantation to LVAD implantation was 36 days (interquartile range [IQR], 17–87 days) and 350 days (IQR, 127–472 days) post-implantation. The median (IQR) time from echocardiogram to CPET was 3 days (1–20 days) pre-implantation and 0 days (0–1 days) post-implantation. In total, 45 patients (92%) had pre-implantation RHC, whereas 26 patients (53%) had post-implantation RHC. Among those with a RHC, the median (IQR) time from RHC to CPET was 16 days (6–55 days) pre-implantation and 161 days (19–317 days) post-implantation.

Baseline characteristics are displayed in Table 1. Briefly, mean peak VO2 was 11.8 ± 2.9 ml/kg/min (43 ± 10 % predicted). Mean exercise time was 5.1 ± 1.4 min. Patients achieved an average work load of 3.6 ± 1.0 metabolic equivalents (METS). Baseline LVEF was 21% ± 10% and LVEDD was 72 ± 11 mm. Right ventricular dysfunction pre-implantation was described by the following distribution: normal to mild: 17 (35%), mild/moderate to moderate: 18 (37%); and moderate/severe to severe: 14 (29%).

Table 1

Table 1

After LVAD implantation, characteristics at the time of CPET were described by the following: septal position was rightward in three patients (9%), neutral in 24 patients (69%), and leftward in eight patients (23%); mean systolic LVAD outflow cannula velocity was 12.7 ± 0.53; and aortic valve opening was none in 27 patients (55%), <50% of cycles in one patient (2%), >50% of cycles in 5 patients (10%), and all cycles in 16 patients (33%). Residual valvular disease is described in Table 1.

There was no improvement in peak VO2 (+0.58 ± 3.6 ml/kg/min; p = 0.26), although exercise time (5.1 [45% predicted] to 5.8 min [54% predicted]) and nadir of the ratio of minute ventilation to carbon dioxide production (VE/VCO2, 39.2 ± 6.5–36.0 ± 6.3) improved (p = 0.02 and p = 0.001, respectively). The median change in VO2 was 7.8% (IQR −18% to 30%; Figure 1). Similarly, the cohort experienced a significant improvement in 6MWT (344 ± 77–393 ± 81 m; p < 0.001). Markers of improved hemodynamics were present: LVEDD decreased from 72 ± 10 to 61 ± 12 mm (p < 0.001), mean PCWP decreased from 22 ± 7 to 12 ± 8 mm Hg (p < 0.001), mean PA pressure decreased from 35 ± 9 to 22 ± 8 mm Hg (p < 0.001), and PVR decreased from 4.4 ± 6.6 to 2.3 ± 1.0 (p = 0.007).

Figure 1

Figure 1

Factors most associated with improvement in VO2 were HM2 PI (R = 0.48; p = 0.003), HM2 power (R = −0.40; p = 0.009), HM2 pump flow (R = −0.40; p = 0.008), and pump speed (R = −0.32; p = 0.04). Figure 2, A–D displays the relationship between LVAD parameters and exercise capacity. Peak heart rate (HR) was also associated with improvement in VO2 (R = 0.41; p = 0.004). Neither LVEF (R = 0.004; p = 0.74) nor change in LVEDD (R < 0.001; p = 0.98) were associated with improvement in VO2. Echocardiographic assessment of RV function showed a trend toward association but was not significant (R = 0.22; p = 0.28). pre-implantation RV function by pulmonary artery pulsatility index (PAPi) and by RV stroke work index (RVSWI) was not associated with change in VO2. Aortic valve opening was not related to exercise capacity change (p = 0.57), and the relationship did not improve when adjusting for patients in whom the aortic valve had been oversewn at the time of LVAD implantation (n = 4). Septal position was not associated with change in VO2 (p = 0.11). Additionally, presence of clinically significant aortic and mitral regurgitation was not associated with exercise capacity (p = 0.82 and p = 0.05, respectively). Changes in VO2 were not associated with RHC data (PCWP, RA pressure, and cardiac output or index) closest to time of CPET post-implantation or comorbid illnesses. These results are displayed in Table 2. Medications at the time of CPET were not associated as well (data not shown). Similar to peak VO2, percent improvement in exercise time was significantly associated with PI (R = 0.47; p = 0.002); however, the associations with pump flow (R = −0.17; p = 0.25), pump speed (R = −0.26; p = 0.09), and power (R = 0.3; p = 0.06) were not significant.

Table 2

Table 2

Figure 2

Figure 2

Back to Top | Article Outline


The principal finding in our study was a modest but significant association between peak VO2 and variables corresponding to a lack of support by the LVAD (lower flow, power, and speed), as well as higher pulsatility index and peak HR during CPET. We also found that successful unloading of the ventricle, including improved RHC-derived filling pressures and neutral ventricular septal position based on echocardiogram while on LVAD therapy, was not associated with improvement in peak VO2. Additionally, we found a modest but significant association between age and peak VO2.

Improvement in peak oxygen consumption in LVAD patients is inconsistently demonstrated. One study found that VO2 is higher in patients on LVAD therapy compared with patients with advanced heart failure on medical therapy; however, these patients were not matched control patients.7 Patients on LVAD support may be able to improve their peak VO2, partly through exercise training, although our study did not assess serial CPET.3,10 However, other studies assessing VO2 improvement after LVAD implantation have found no significant difference, including over time.9

Data on native heart function or pump support with regard to exercise capacity have been variable. In one study, peak VO2 correlated with ejection fraction, especially at lower levels of LVAD support levels11; however, other studies have found no correlation between LVEF and peak VO2.10,12 Physiologically, native heart function may be an important determinant of exercise capacity as axial-flow pumps may contribute approximately 1 L (900 ml/min6 or 1,300 ml/min13) at maximal exercise, but native cardiac output can contribute an additional 3 L (3.0 L/min6 or 3.1 L/min13) even in patients with CF-LVADs. The relative pump contribution mostly reflects an increase in preload with exercise and possibly a greater proportion of systole relative to diastole, but unlike the native heart, there are no direct inotropic or chronotropic effects.13–15

With regard to pump support, some data have shown that higher peak VO2 was achieved at higher pump speeds (9,000 vs. 6,000 rpm),11 although 6,000 rpm is not a pump speed that is used clinically in HM2 patients. Another small study showed that an increase in HM2 speed from a mean 9,357 to 10,843 rpm was associated with improvement in VO2.12 However, in an invasive study of cardiac output and peripheral blood flow, an augmentation in pump speed did not improve exercise tolerance.16 Because pump speed was not actively manipulated in our study, we cannot draw conclusions about real-time changes in LVAD support as they relate to exercise capacity.

Optimization of ambulatory LVAD patients is challenging because of the need to balance septal position, aortic valve opening, mean arterial pressure, and estimated filling pressures. Even if these factors are optimized at rest, they may change dynamically. For example, we were unable to establish that aortic valve opening was a predictor of improved exercise capacity, despite the postulate that native cardiac function may be a primary driver of exercise capacity. Although not inconsistent with previous observations regarding aortic valve opening,17 we also did not dynamically assess aortic valve opening, which may be a more relevant end-point.18 As a result of limitations in optimization at rest, there may be utility in evaluating LVAD patients dynamically on a routine basis to improve functional capacity.

Our data provide an early indication that a relative lack of pump support may be associated with better exercise capacity although prior data have provided a physiologic rationale. One corollary relates to the utility of neurohormonal blockade in patients on LVAD therapy. There are data to suggest improvement in myocardial recovery with continued institution of neurohormonal therapies in advanced heart failure patients,19 including β-blockade.18 Nevertheless, a focus on instituting neurohormonal therapies, which do not produce excessive negative inotropy and chronotropy, may be a strategy for further study to improve exercise capacity. Additionally, assessment of response to dynamic physiology in the form of exercise is difficult to evaluate on resting hemodynamic evaluations. Dynamic evaluation of cardiac output reserve and filling pressures in patients supported on LVAD therapy may be necessary for optimization and characterize the mechanisms of suboptimal response to exercise better.

Back to Top | Article Outline


The current study is limited most by the retrospective nature, including the inability to manipulate LVAD settings in real time to affect changes in cardiopulmonary parameters described herein and, particularly, to assess these dynamically. However, we believe that the data still reflect real-life scenarios where LVAD optimization has already occurred and the balance of left ventricular unloading, ventricular septal position, mean arterial pressure, and symptomatic improvement has been achieved. Our main conclusion relates to controller-derived variables, which are inherently less well validated and therefore may be less relevant than traditional hemodynamic variables. Furthermore, pump flow is derived from power and speed and thus would correlate with putative variables to the same degree. Right heart catheterization was less commonly obtained near to the date of CPET, and data were only available in approximately half the study cohort, reducing the external generalizability of the RHC data. Our echocardiographic evaluation of RV function was qualitative as we could not consistently obtain tricuspid valve lateral annulus velocities, which have been shown to be a prognostic RV marker on echocardiography.17 Our mode INTERMACS score was four, which was likely a function of ascertainment bias related to an ambulatory population; however, our results are oriented toward management of ambulatory patients who may be persistently symptomatic. Finally, we are unable to determine to what degree patients participated in cardiac rehabilitation after implant as the majority (82%) were referred for rehabilitation closer to home.

Back to Top | Article Outline


Our data suggest that lack of reliance on LVAD support based on LVAD parameters and native heart rate may explain a proportion of the variability in exercise capacity. Further studies should prospectively evaluate level of device support and medication therapy in relation to exercise capacity, potentially improving exertional capabilities of patients on LVAD support.

Back to Top | Article Outline


1. Kirklin JK, Naftel DC, Pagani FD, et al. Sixth INTERMACS annual report: A 10,000-patient database. J Hear Lung Transplant 2014.33: 555–564.
2. Francis DP, Shamim W, Davies LC, et al. Cardiopulmonary exercise testing for prognosis in chronic heart failure: Continuous and independent prognostic value from VE/VCO(2)slope and peak VO(2). Eur Heart J 2000.21: 154–161.
3. Jonge N De, Kirkels H, Lahpor JR, et al. Exercise performance in patients with end-stage heart failure after implantation of a left ventricular assist device and after heart transplantation: An outlook for permanent assisting? J Am Coll Cardiol 2001.37: 7–12.
4. Compostella L, Russo N, Setzu T, Compostella C, Bellotto F. Exercise performance of chronic heart failure patients in the early period of support by an axial-flow left ventricular assist device as destination therapy. Artif Organs 2014.38: 366–373.
5. Dunlay SM, Allison TG, Pereira NL. Changes in cardiopulmonary exercise testing parameters following continuous flow left ventricular assist device implantation and heart transplantation. J Card Fail 2014.20: 548–554.
6. 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.
7. Mancini D, Goldsmith R, Levin H, et al. Comparison of exercise performance in patients with chronic severe heart failure versus left ventricular assist devices. Circulation 1998.98: 1178–1183.
8. Jakovljevic DG, George RS, Nunan D, et al. The impact of acute reduction of continuous-flow left ventricular assist device support on cardiac and exercise performance. Heart 2010.96: 1390–1395.
9. 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.
10. Maybaum S, Mancini D, Xydas S, et al; LVAD Working Group: Cardiac improvement during mechanical circulatory support: A prospective multicenter study of the LVAD Working Group. Circulation 2007.115: 2497–2505.
11. Noor MR, Bowles C, Banner NR. Relationship between pump speed and exercise capacity during HeartMate II left ventricular assist device support: Influence of residual left ventricular function. Eur J Heart Fail 2012.14: 613–620.
12. Jung MH, Hansen PB, Sander K, et al. Effect of increasing pump speed during exercise on peak oxygen uptake in heart failure patients supported with a continuous-flow left ventricular assist device. A double-blind randomized study. Eur J Heart Fail 2014.16: 403–408.
13. Martina J, de Jonge N, Rutten M, et al. Exercise hemodynamics during extended continuous flow left ventricular assist device support: The response of systemic cardiovascular parameters and pump performance. Artif Organs 2013.37: 754–762.
14. Moazami N, Fukamachi K, Kobayashi M, et al. Axial and centrifugal continuous-flow rotary pumps: A translation from pump mechanics to clinical practice. J Heart Lung Transplant 2013.32: 1–11.
15. Lim HS, Howell N, Ranasinghe A. The physiology of continuous flow left ventricular assist devices. J Card Fail 2016. 23: 169–180, Available at:
16. Brassard P, Jensen AS, Nordsborg N, et al. Central and peripheral blood flow during exercise with a continuous-flow left ventricular assist device: Constant versus increasing pump speed: A pilot study. Circ Heart Fail 2011.4: 554–560.
17. Topilsky Y, Hasin T, Oh JK, et al. Echocardiographic variables after left ventricular assist device implantation associated with adverse outcome. Circ Cardiovasc Imaging 2011.4: 648–661.
18. Imamura T, Kinugawa K, Nitta D, et al. Opening of native aortic valve accomplished after left ventricular assist device implantation in patients with insufficient preoperative beta-blocker treatment. Int Heart J 2015.56: 303–308.
19. Patel SR, Saeed O, Murthy S, et al. Combining neurohormonal blockade with continuous-flow left ventricular assist device support for myocardial recovery: A single-arm prospective study. J Hear Lung Transplant 2013;32:305–312. Available at:

left ventricular assist device; cardiopulmonary exercise testing; optimization; LVAD support

Copyright © 2018 by the American Society for Artificial Internal Organs