Right ventricular afterload sensitivity dramatically increases after left ventricular assist device implantation: a multi-center hemodynamic analysis

Abstract

Background

Right ventricular (RV) failure is a source of morbidity and mortality after left ventricular assist device (LVAD) implantation. We sought to define hemodynamic changes in afterload and RV adaption to afterload both early after implantation and with prolonged LVAD support.

Methods

We reviewed right heart catheterization (RHC) data from participants who underwent continuous-flow LVAD implantation at our institutions (n=244), excluding those on inotropic or vasopressor agents, pulmonary vasodilators, or additional mechanical support at any RHC. Hemodynamic data was assessed at five time intervals: 1) pre-LVAD (within 6 months), 2) early post-LVAD (0–6 months), 3) 7–12 months, 4) 13–18 months and 3) very-late post-LVAD (18–36 months).

Results

Sixty participants met the inclusion criteria. All measures of right ventricular load (effective arterial elastance, pulmonary vascular compliance and pulmonary vascular resistance) improved between the pre- and early post-LVAD time periods. Despite decreasing load and pulmonary capillary artery pressure (PAWP), RAP remained unchanged and the RAP:PAWP ratio worsened early post-LVAD (0.44 [0.38, 0.63] versus 0.77 [0.59, 1.0], p<0.001), suggesting a worsening of RV adaptation to load. With continued LVAD support, both RV load and RAP:PAWP decreased in a steep, linear and dependent manner.

Conclusion

Despite reducing RV load, LVAD implantation leads to worsened RV adaptation. With continued LVAD support, both RV afterload and RV adaptation improve, and their relationship remains constant over time post-LVAD. These findings suggest the RV afterload sensitivity increases after LVAD implantation, which has important clinical implications for patients struggling with RV failure.

Introduction

Right ventricular failure (RVF) has historically provided a major source of morbidity and mortality after left ventricular assist device (LVAD) implantation.^1–4 While definitions of RVF vary, somewhere between 9 and 40% of patients experience RVF after LVAD insertion,^1,5–11 indicating the inexact nature of our understanding of the physiology of the right ventricle during the implant process and with prolonged mechanical circulatory support of the left ventricle.

The explanations for the decline in RV function and the continued high incidence of RVF after LVAD implantation despite reduced RV afterload are myriad. Different studies have implicated changes in RV preload, damage to the RV during surgery, and disadvantageous changes in ventricular interdependence mediated by reduced LV contractility, changes in septal architecture, and alterations in RV shape¹² all in the setting of a suddenly elevated cardiac output.^13–15 Longitudinal hemodynamic assessment of RV function is poorly defined post-LVAD placement, and it is not known whether RV function and afterload change over a prolonged period of LVAD support.

Given the importance of RV function to LVAD patients’ functional capacity,¹⁶ we sought to define hemodynamic changes in RV function and afterload initially and over time with prolonged LVAD support. We compared hemodynamic markers of RV function and afterload beginning before LVAD placement and continuing in four discrete post-operative time periods. To our knowledge, this is the first study to rigorously examine hemodynamic changes in RV function and load after such a prolonged duration of LVAD support.

Methods

Study design and Participants

We retrospectively reviewed right heart catheterization (RHC) data from all participants who underwent continuous-flow LVAD implantation at the Johns Hopkins Hospital and Vanderbilt University between 2005 and April 2014 (n=244). To isolate the LVAD’s effects on RV physiology, participants were excluded from the entire analysis if they were on inotropic or vasopressor agents, pulmonary vasodilators, or additional mechanical support at any RHC assessment. Only participants undergoing LVAD implantation through a median sternotomy and on cardiopulmonary bypass were included. Hemodynamic data was assessed at five discrete time intervals: 1) pre-LVAD (within 6 months), 2) early post-LVAD (0–6 months post-LVAD implantation), 3) 7–12 months post-LVAD, 4) 13–18 months post-LVAD and 3) very-late post-LVAD (18–36 months post-LVAD implantation) (Figure 1). This study was performed with the approval from both institutional review boards.

Figure 1. Right Heart Catheterization Assessment Time Course.

Visualization of temporal relationship of RHC assessments evaluated for each patient and summary of hemodynamic variables measured and calculated for each right heart catheterization (RHC). Abbreviations: RAP: Right atrial pressure; sPAP: Systolic pulmonary artery pressure; mPAP: mean pulmonary artery pressure; dPAP: Diastolic pulmonary artery pressure; PAWP: Pulmonary artery wedge pressure; CO: Cardiac output; SV: Stroke volume; HR: Heart rate; PVR: Pulmonary vascular resistance; E_A: Effective arterial elastance; P_CA: Pulmonary vascular compliance.

Right Heart Catheterization

All RHC procedures were performed in the cardiac catheterization labs at the Johns Hopkins Hospital or Vanderbilt University Hospital by a heart failure cardiologist. Hemodynamic variables were analyzed in the pre-implant period and compared to values in each post-implant period. Cardiac output (CO) was obtained using the thermodilution method measured in triplicate. Thermodilution is the preferred method of obtaining CO,¹⁷ is a validated method for assessing CO in LVAD patients,^18,19 and has been shown to be independent of tricuspid regurgitation.^20,21 To assess RV load, three separate pulmonary vascular parameters easily obtained from a routine right heart catheterization were evaluated: effective arterial elastance (E_A), pulmonary vascular compliance (P_CA), and pulmonary vascular resistance (PVR). P_CA was calculated by dividing the stroke volume (SV) by the pulmonary pulse pressure.²² E_A was calculated by dividing the systolic pulmonary artery pressure (sPAP) by the SV.^23–25 To evaluate RV function, we calculated the ratio of the right atrial pressure (RAP) and pulmonary arterial wedge pressure (PAWP), (RAP:PAWP).

Statistical Analysis

Values of continuous variables were expressed as median [interquartile range]. The primary outcome measure was RAP:PAWP as a measure of RV adaptation over time. Secondary outcomes were three measures of RV load (E_A, P_CA, PVR) and load expressed as E_A indexed to both RAP and RAP:PCWP (E_A/RAP and E_A/RAP:PAWP). Outcome measures pre- and early post-LVAD were compared via Wilcoxon signed rank sum test to account for repeated measures. Dependent variables post-LVAD were assessed over time using a mixed linear model with repeated subject measurements accounted for random effects. This appropriately accounted for the repeated measurement of some subjects while allowing for the missing assessments noted among others.

To further investigate the possibility of a healthy survivor bias, given an expected high attrition rate during prolonged post-LVAD follow-up, we evaluated longitudinal hemodynamics in the subset of participants who had qualifying RHC assessments pre-LVAD, and early (0–6 months) and very late (19–36 months) post-LVAD as a pre-planned sensitivity analysis. Comparison of hemodynamic outcome measurements was performed from each of these three time points via Wilcoxon signed rank sum in this analysis.

Finally, a series of trends in RV load and adaptation were presented graphically using simple plots. A best fit line to describe relationships was established using linear regression, and the coefficient of determination (R²) was calculated to determine the ratio of the calculated to total variation between measures of RV load and adaptation post-LVAD.

A p-value < 0.05 was considered statistically significant. Statistical analysis was performed using SigmaPlot version 12.5 (Systat Software Inc., San Jose, California) and Stata-version 11 (StataCorp LP, College Station, Texas).

Results

Study sample

Of the 244 participants reviewed, 60 met the inclusion criteria and had qualifying hemodynamic measurements via right heart catheterization (RHC) pre-LVAD (within 6 months of implant) and at least once more post-LVAD. Seventeen participants received a Heartware HVAD device (Heartware; Framingham, MA), and the remaining 43 participants received a HeartMateII device (Thoratec; Pleasanton, CA). Most participants were male (75%) and had non-ischemic cardiomyopathy (68%). Baseline subject characteristics pre-LVAD implantation are further delineated in Table 1.

Table 1.

Baseline subject characteristics.

Age at implant (years)	51.6 + 12
Male	45/60 (75%)
Nonischemic etiology	41/60 (68%)
Body Mass Index (kg/m2)	30.4 + 7.2
LVAD type Heartmate II	43/60 (72%)
INTERMACS classification	3.1 + 1.1
Prior cardiac surgery	15/60 (25%)
Concomitant cardiac surgery	13/60 (22%)
Echocardiographic Parameters
•LV Ejection Fraction	16.3% + 7
• LVEDD (mm)	71 + 11
• Mitral regurgitation > moderate	27/60 (45%)
• Tricuspid regurgitation > moderate	23/60 (38%)
• RV dilation > moderate	19/60 (32%)
• RV size at base (mm)	38 + 8

Abbreviations: LV – Left ventricular; RV = Right Ventricular; LVEDD = Left ventricular end diastolic diameter.

Pre-LVAD versus Early Post-LVAD (0–6 months)

Forty-five participants had a qualifying RHC assessment early post-LVAD (within 6 months after LVAD implant) and were compared to their pre-LVAD hemodynamic measurements as shown in Table 2. Median time of early post-LVAD RHC assessment was 83 [44, 121] days. As expected, PAWP declined compared to the pre-LVAD assessment, and all three measures indicated a reduction of RV load at early post-LVAD RHC compared to pre- LVAD assessment: E_A (0.59mmHg/mL [0.42, 0.9] versus 1.31mmHg/mL [0.7, 1.62], p<0.001); P_CA (3.02mL/mmHg [2.07, 4.03] versus 1.77 mL/mmHg [1.26, 2.51], p<0.001), and PVR (2.04 WU [1.68, 2.64] versus 2.66 WU [2.09, 3.72], p<0.001). Despite the reduced LV filling pressure and RV load however, the RAP did not change, and RAP:PAWP increased early post-LVAD (0.77 [0.59, 1.0] versus 0.44 [0.38, 0.63], p<0.001), suggesting worsened RV function.

Table 2.

Hemodynamics pre-LVAD versus 0–6 months post-LVAD.

	Pre-LVAD (n=45)	0–6 Months Post (n=45)
Time of post-LVAD RHC (days)		83 [44, 121]
RAP (mmHg)	11 [5, 16]	10 [5, 15]
PAWP (mmHg)	23 [17, 30]	12 [7, 17]*
RAP:PAWP	0.44 [0.38, 0.63]	0.77 [0.59, 1.0]*
PVR (WU)	2.66 [2.09, 3.72]	2.04 [1.68, 2.64]*
PCA (mL/mmHg)	1.77 [1.26, 2.51]	3.02 [2.07,4.03]*
EA (mmHg/mL)	1.31 [0.7, 1.62]	0.59 [0.42, 0.9]*

^*p<0.001 compared to pre-LVAD

Post-LVAD hemodynamic measures (0–6, 7–12, 13–18, and 19–36 months)

Table 3 delineates hemodynamic measures at all time points post-LVAD: 0–6 months (median 83 [44, 120.5] days, n=45), 7–12 months (261 [213, 290] days, n=24 patients), 13–18 months (406 [398, 455] days, n=15 patients), and 19–36 months (721 [627, 860] days , n=11). There was no difference in median pump speed between the early post-LVAD assessment and any interim time period among HeartMateII recipients. There was no significant difference in pump speed between the 12 HVAD participants included at 0–6 months and 3 HVAD participants included at 7–12 months (2800 [2600, 2930] versus 3000 rpm [2920, 3120], p=0.09). Only 1 HVAD participant was included in the 13–18 and 19–36 month time periods obviating further comparison. RAP and PAWP were unchanged across all time points post-LVAD. RAP:PAWP decreased significantly post-LVAD (p=0.011). Likewise, all measures of RV load declined post-LVAD: E_A (p=0.036), P_CA (p=0.033), and PVR (p=0.003). Figure 2 illustrates the temporal trends in markers of RV load and RAP:PAWP.

Table 3.

Hemodynamics 0–6 months post-LVAD, 7–12 months post-LVAD, 13–18 months post-LVAD, and 19–36 months post-LVAD.

	0–6 Months Post (n=45)	7–12 Months Post (n=24)	13–18 Months Post (n=15)	19–36 Months Post (n=11)	p-value *
Time of RHC (days)	83 [44, 120.5]	261 [213, 290]	406 [398, 455]	721 [627, 860]	-
RAP (mmHg)	10 [5, 14.5]	8.5 [4.25, 16.75]	8 [4, 17]	6 [4, 12]	p=0.25
PAWP (mmHg)	12 [7, 17]	13 [5.25. 19.5]	14 [8, 20]	12 [8,17]	p=0.424
RAP:PAWP	0.77 [0.59, 1.0]	0.65 [0.56, 0.88]	0.63 [0.43, 0.89]	0.56 [0.5, 0.67]	p=0.011
PVR (WU)	2.04 2.64] [1.68,	2.05 [1.27, 2.88]	1.96 [1.19, 2.09]	1.17 [0.89, 1.35]	P=0.003
PCA (mL/mmHg)	3.02 [2.07,4.03]	3.8 [2.79, 4.39]	3.81 [3.2, 5.74]	4.42 [3.3, 6.08]	p=0.033
EA (mmHg/mL)	0.59 [0.42, 0.9]	0.53 [0.35, 0.85]	0.47 [0.38, 0.77]	0.38 [0.31, 0.62]	p=0.036
Pump speed for HMII (RPM)	9200 [9190, 9600]	9600 [9200, 9600]	9600 [9150, 10,050]	9300 [9200, 9850]	p=0.205

^*Statistical comparison performed via linear mixed effects regression models across all time points.

Figure 2. Hemodynamics over time.

All measures of load improve from pre-LVAD to the 0–6 months post-LVAD measurement (E_A, P_CA, and PVR). However, RAP:PAWP worsens. Post- LVAD, all markers of RV load continue to improve as does RAP:PAWP. Abbreviations: RAP: Right atrial pressure; PAWP: Pulmonary artery wedge pressure; Ea: Effective arterial elastance; PCA: Pulmonary vascular compliance; PVR: Pulmonary vascular resistance

RV load versus adaption analysis

We analyzed all participants by plotting RAP versus E_A (Figure 3). Post-LVAD, we appreciated a steep, linear relationship between total RV load (E_A) and RAP (R²=0.98). The slope of this line indicates that for every 0.1 mmHg/mL decrease in E_A, RAP declined by 1.3 mmHg. However, this steep, linear relationship was not noted between the pre-LVAD and early post-LVAD assessment, where only minimal and non-significant decline in RAP was noted despite a marked decline in E_A (for every 0.1 decrease in E_A, RAP only decreased 0.36 mmHg). Calculating the ratio of E_A/RAP permits a mathematical expression of the RV load to adaptation relationships at each discrete time point pre-LVAD and post-LVAD (Table 4). Pre-LVAD, median E_A/RAP was 0.11 [0.08, 0.16], which was significantly higher than early post-LVAD (0.07 [0.05,0.11], p=0.003). For all subsequent time points post-LVAD, the E_A to RAP ratio remained unchanged (p=0.35)

Figure 3. Right atrial pressure (RAP) versus total RV load (E_A).

Median and SD values represented. There is a linear relationship between RAP and E_A post-LVAD (black line, R²=0.98). For each 0.1mmHg/mL decline in E_A, RAP declines by 1.3mmHg. However, the pre- LVAD relationship between RAP and RV load (blue line) demonstrates that despite a marked decline in RV load (EA), RAP only minimally declines (for every 0.1 decrease in E_A, RAP only decreased 0.36mmHg).

Table 4.

RV adaptation (RAP and RAP:PAWP) versus RV load (E_A and P_CA) over time. Both E_A/RAP and E_A/RAP:PAWP decline significantly after LVAD insertion, indicating worsening RV adaptation to load conditions. PCA X RAP and PCA X RAP:PAWP both increase significantly after-LVAD insertion. Post-LVAD, there is no significant different between any mathematical expression of RV load versus adaptation at each successive time point, indicating a constant RV load/function relationship.

	Pre LVAD N=45	0–6 months N=45	7–12 months N=24	13–18 months N=15	19–36 months N=11	p-value: pre- LVAD vs. 0–6 months*	p-value: post- LVAD time points†
EA/RAP	0.11 [0.08, 0.16]	0.07 [0.046, 0.108]*	0.059 [0.05, 0.09]*	0.061 [0.04, 0.116]*	0.064 [0.045, 0.093]*	0.003	0.35
EA/RAP:PAWP	2.6 [1.7, 4.27]	0.97 [0.51, 1.44]*	0.84 [0.54, 1.16]*	0.81 [0.56, 1.05]*	0.83 [0.52, 1.1]*	<0.001	0.46
PCA X RAP	18.5 [12.4, 25.1]	26.9 [16.9, 39.4]*	31.4 [19.8, 43.6]*	30.5 [16.0, 54.0]*	23.4 [18.3, 46.3]*	<0.001	0.264
PCA X RAP:PAWP	0.8 [0.5, 1.2]	2.0 [1.4, 3.5]*	2.6 [1.8, 3.5]*	3.0 [1.6, 3.6]*	2.1 [ 1.9, 3.8]*	<0.001	0.904

^*Statistical comparison via Wilcoxon signed rank sum

^†Statistical comparison via linear mixed effects regression models across all time points.

We next plotted E_A versus RAP:PAWP to further assess RV response to the decreasing load conditions over time. Comparing pre-LVAD to early post-LVAD, we noted a decline in E_A/RAP:PAWP (2.6 [1.7, 4.3] versus 1.0 [0.5, 1.4], p<0.001) indicating a worsening RV adaption to load relationship. Post-LVAD, we again noted a linear relationship between E_A and RAP:PCWP (R²=0.9) (Figure 4) and an unchanged expression of E_A/RAP:PAWP (p=0.46) indicating a constant RV adaption to load relationship.

Figure 4. Right atrial pressure to pulmonary arterial wedge pressure ratio (RAP:PAWP) versus total RV load (E_A).

Median and SD values represented. With VAD implant, RAP:PAWP increases despite a decrease in RV load (red line). However, post-LVAD, RAP:PAWP and RV load decline in a linear fashion (blue line, R²=0.9). Abbreviations: RAP: Right atrial pressure; PAWP: Pulmonary artery wedge pressure.

To evaluate a second marker of RV load, we also plotted P_CA versus RAP and RAP:PAWP, finding a similar relationship between RV load and adaptation over time (Supplemental Figures 1 and 2) – namely that RV function improves post-LVAD in a linear relationship to an improving P_CA (post-LVAD R² for P_CA relationship to RAP:PCWP = 0.96, for RAP = 0.99). Given the expected inverse relationship between P_CA and markers of RV adaptation (RAP and RAP:PCWP) we calculated P_CA X RAP and P_CA X RAP:PAWP, again finding that these mathematical expressions of RV load to adaptation relationship changed significantly with LVAD implantation (P_CA X RAP: 18.5 [12.4, 25.1] versus 26.9 [16.9, 39.4], p<0.001; P_CA X RAP:PAWP: 0.80 [0.54, 1.16] versus 1.96 [1.42, 3.45], p<0.001). However, all post-LVAD values (whether P_CA was multiplied by RAP or RAP:PAWP) were unchanged, again indicating a constant RV adaption to load relationship (Table 4).

Sub-group Analyses

To further assess for survivor bias, the subgroup of “survivors” in our cohort (patients who had RHC assessments pre-LVAD, and at 0–6 and 19–36 months (n=11) was evaluated. Hemodynamic measures were compared from each of these three time points, and revealed similar findings to the total cohort’s (Table 5). PAWP declined between pre-LVAD and early post-LVAD (25mmHg [17, 32] versus 10mmHg [5, 15], p<0.001) and remained unchanged at 19–36 months. All assessments of RV load (E_A, P_CA, and PVR) improved comparing pre-LVAD to early post-LVAD. However, similar to the total cohort, RAP:PAWP worsened in the “survivors” from pre-LVAD to early post-LVAD (0.47 [0.41, 0.64] versus 0.85 [0.68, 1.04], p<0.001). Similar to the larger cohort, RAP:PAWP improved in the survivor cohort post-VAD comparing the 0–6 and 19–36 month time periods (0.85 [0.68, 1.04] versus 0.56 [0.5, 0.67], p=0.024).

Table 5.

Hemodynamics pre-LVAD 0–6 months post LVAD and 19–36 months post-LVAD in “survivors,” only patients that had qualifying assessments at all three time points.

	Pre-LVAD (n=11)	0–6 Months Post (n=11)	19–36 Months Post (n=11)
Time of RHC (days)		70 [18.5, 78]	721 [627, 860]
RAP (mmHg)	14 [8, 16]	8 [4, 13.5]	6 [4, 12]
PAWP (mmHg)	25 [17, 32]	10 [5.5, 15]*	12 [8,17]*
RAP:PAWP	0.47 [0.41, 0.64]	0.85 [0.68, 1.04]*	0.56 [0.5, 0.67] †
PVR (WU)	2.22 [1.65, 2.78]	1.50 [0.92, 2.08]*	1.17 [0.89, 1.35]* †
PCA (mL/mmHg)	2.00 [1.8, 3.35]	4.03 [1.1, 4.6]*	4.42 [3.3, 6.08]*†
EA (mmHg/mL)	1.04 [0.61, 1.43]	0.43 [0.32, 0.56]*	0.38 [0.31, 0.62]*
Pump speed for HMII (RPM)	9200 [9200, 9400]	9300 [9200, 9850]	9300 [9200, 9850]

^*p<0.05 compared to pre-LVAD

^†p<0.05 compared to 0–6 months post-LVAD

Similar analyses for three more subgroups were completed: participants with ischemic etiology, participants with non-ischemic etiology, and participants with HeartMateII devices (only one subject with a HVAD device was included in the final and penultimate time periods, obviating subgroup analysis). Regardless of subgroup, the hemodynamic changes were analogous to those of the general cohort.

Discussion

Right heart failure (RVF) continues to be a major source of morbidity and mortality in patients after LVAD implantation. However, relatively few studies have rigorously evaluated the isolated hemodynamic effect of mechanical left ventricular support on RV load and function among those who do receive LVAD support. This study represents the largest such study comparing hemodynamic RV load and function before and after continuous flow LVAD in patients free of inotropic, additional mechanical, or pulmonary vasodilator support at the time of assessment. Importantly, this study is not designed to describe or predict RV failure, and thus we do not describe clinical outcomes in this cohort. This study is instead designed to be purely descriptive to elucidate isolated hemodynamic effects of LVAD implantation and prolonged support on RV load and adaptation. To isolate LVAD hemodynamic effects, we purposefully excluded patients on medications that might alter RV load or adaptation (pulmonary vasodilators, inotropes, vasopressors) or alternate mechanical support (intra-aortic balloon pump, Impella device) during any RHC. The major findings of our study are: 1) RV afterload declines early after LVAD and continues to decline with prolonged support and 2) the RV adaptation to load worsens after implantation and this relationship remains constant over time. These all point to increasing afterload sensitivity of the RV after LVAD implantation.

Hemodynamic changes early post-LVAD

In this study, PVR and E_A declined and P_CA increased early post-LVAD implantation, confirming that LVAD implantation alone reduces both resistive and pulsatile components of RV load. Despite this, however, right atrial pressure did not decline early post-LVAD. RAP:PAWP increased and the relationship of RV load to RV adaption (whether expressed by E_A/RAP or E_A/RAP:PAWP) worsened. An elevated RAP:PAWP has been shown to be an independent predictor of RV failure post-LVAD,²⁶ is associated with adverse clinical outcomes in patients with heart failure,²⁷ and is associated with lower cardiac output and mean arterial pressure after cardiac surgery.²⁸ Furthermore, RAP:PAWP has been shown to remain constant over time in heart failure patients. Data from the ESCAPE trial^27,29 demonstrates that RAP:PAWP remains stable in heart failure patients despite aggressive heart failure therapy including inotropes (Supplemental Table). Similarly, a retrospective analysis of over 4000 patients over a 14 year period showed that the RAP:PAWP relationship remained constant over time, and that an elevated RAP:PAWP ratio is associated with increased mortality after heart transplant.³⁰ Thus our observed alterations in RAP:PAWP with LVAD insertion and over time with LVAD support represents a significant departure from the normal physiologic state.

It is reasonable to presume that the elevation in RAP:PAWP in the face of reduced RV load is caused by either RV systolic or diastolic dysfunction. Moon et al demonstrated in a closed-chest dog model that LVAD support did not alter RV diastolic function;¹³ thus it is likely that the observed increase in RAP:PCWP early post-LVAD is a marker of reduced RV contractile function. Our findings corroborate those of a small, early study wherein Morita and colleagues employed echocardiographic and beat-to-beat pressure recordings in 8 patients undergoing Novacor LVAD implantation, demonstrating that RV systolic pump function worsened despite a reduction in RV afterload with LVAD implantation.³¹ In our study, it is notable that a decrement in RV function was observed even in a cohort of patients who, relative to the larger population receiving LVAD support, were relatively well as indicated by their lack of requirement for pharmacologic or inotropic support at the time of the pre-LVAD RHC.

The earliest study comparing long-term hemodynamics in LVAD patients was performed in 58 consecutive patients receiving pulsatile LVAD support between 1996 and 2003.³² This study demonstrated a reduction in trans-pulmonary gradient and PVR with LVAD support, though median timing of late hemodynamic assessment was performed at only 75 days and included patients on inotropic, intra-aortic balloon, and additional mechanical support both pre- and post-LVAD. Three previous studies have reported early hemodynamic findings in patients supported by continuous-flow LVAD. In 2010, Lee et al compared RHC hemodynamics in 40 bridge-to-transplant patients supported with HeartMate II devices pre-implant and post-implant (mean time of 139.3 days),³³ and in 2013, Morgan et al evaluated 105 patients with the a HeartMateII or HVAD device pre-implant and at 1 and 6 months post-implant.³⁴ Both studies found that left heart filling pressure (PAWP) and pulmonary vascular resistance declined post-LVAD, though diverged from this study in finding that RAP also declined early post-LVAD. In contrast, Herod et al found no reduction in RAP in 17 HeartMateII patients at a mean time of 136 days after LVAD implantation.³⁵ Though not specifically reported, all of these studies also saw a rise in the ratio of mean RAP to mean PAWP. Each of these studies included patients on inotropic, intra-aortic balloon, and additional mechanical support both pre- and post-LVAD making it difficult to separate these factors’ hemodynamic effects from the effects of LVAD implantation. Further, these and other studies have evaluated post-LVAD RV function and load by echocardiogram, largely relying on tricuspid annular systolic plane excursion (TAPSE). However, as Raina et al have recently reported, TAPSE becomes a poor marker of RV function after pericardiectomy and cardiopulmonary bypass.³⁶ Similarly, several previous studies have reported right ventricular stroke work index (RVSWI) as a post-LVAD hemodynamic marker of RV function. However, RVSWI is load dependent, and thus a declining RVSWI post-LVAD may be a marker of reduced RV load instead of a marker of intrinsic RV dysfunction.

Hemodynamic changes with prolonged LVAD support

The current study is also the first to evaluate the course of RV load and functional hemodynamic parameters for such an extended duration after continuous-flow LVAD implantation (up to 3 years). Morgan et al reported a stable reduction in transpulmonary gradient at 1 and 6 months post-LVAD, but did not offer any later assessments.³⁴ Mikus et al observed a significant reduction in PVR in 27 patients at up to one year post-LVAD, though did not report any later hemodynamic assessments, concluding that there is likely no further decrease in PVR after 6 months.³⁷ However, in our study, we find that all markers of RV load improve significantly at assessments after 18 months compared to early post-LVAD assessments, and that there are appropriate temporal directional trends in all three parameters noted at interval assessments. Further, concomitant with this reduction in load we noted a linear temporal improvement in RAP and RAP:PAWP. This suggests that the observed improvement in RV adaption was dependent on the decreasing RV load, supported by the constant mathematical expressions of RV load adaptation, E_A/RAP and E_A/RAP:PAWP at all time points post-LVAD. In 2013, Dandel et al found that a similar load adaptation index derived from echocardiographic parameters provided an accurate assessment of RV adaptation to load and carried a high predictive value for RV failure.³⁸ That study employed tricuspid regurgitation velocity time integral as a surrogate for RV load and RV end-diastolic area to long axis dimension as a surrogate for RV function. We believe E_A/RAP and E_A/RAP:PAWP represent hemodynamic correlates of this RV load adaptation index.

Longitudinal shortening via the interventricular septum has been shown to be responsible for nearly 80% of RV function during in normal physiologic states.³⁹ The oblique fiber orientation of the septum and its twisting motion allows for ejection of blood into a high afterload circuit.⁴⁰ However, after cardiothoracic operations involving pericardiotomy and cardiopulmonary bypass, the right ventricle drastically changes its contractile pattern to rely heavily on transverse shortening.³⁶ The RV may compensate via this enhanced transverse function as long as afterload is low, but may be unable to do so when load is higher. The reliance on transverse motion after cardiac surgery may provide an explanation for the strikingly tight and steep correlation observed in this study between RV adaption and load post-LVAD.

Limitations

Survivor analysis suggests that our findings in the total cohort were not due to healthy survivor bias, and statistical analysis by linear mixed effects regression models across all time points post-LVAD accounts for both subject attrition over time and repeated measures analysis. Our findings were consistent for both ischemic and non-ischemic patients and when the analysis was limited to HeartMateII patients only. However, this study has several limitations. First, despite collaboration between two relatively high-volume LVAD centers, our strict exclusion criteria in an attempt to isolate LVAD hemodynamic effects on the RV limited the number of patients. Second, despite its clinical implications and demonstrated stability over time in several heart failure non-LVAD cohorts, RAP:PAWP remains a surrogate marker for RV function. Given its retrospective nature and the aforementioned difficulties in rigorously assessing the RV echocardiographically post-LVAD, our study does not include echocardiographic measurements of RV load or adaptation. Our findings do corroborate those of Herod et al, who found that at a relatively late post-LVAD assessment (mean 234 days, correlating roughly with 7–12 month time period in our study), the majority of their 17 patient cohort demonstrated improvements in RV systolic strain and strain rate.³⁵ Finally, this study is unable to comment on the etiologic mechanisms of the observed hemodynamic changes. While it is logical to presume that the early post-LVAD reduction in RV load is due largely to reduced LV filling pressures and associated improvement in vascular compliance, this study cannot definitively elucidate the cause of the early decrement in RV function. Nor can it comment on the biologic mechanism of the subsequent gradual improvement in RV load, though the possibility of pulmonary vascular reverse remodeling with continued mechanical unloading is intriguing and calls for further study.

Clinical Perspectives

Our findings have important clinical implications. For patients struggling with borderline or poor RV function post-LVAD, given the increased afterload sensitivity of the RV after implant, efforts should be aimed at keeping afterload as low as possible. This would include ensuring adequate mechanical unloading of the LV to minimize any contribution from elevated left heart filling pressures and its effect on pulmonary vascular compliance.^41–43 It may also support the use of adjunctive pharmacologic therapy to treat a pre-capillary component of pulmonary hypertension persisting after optimal LV unloading and correction of peri-operative factors which can induce pulmonary vasoconstriction (i.e. hypoxia, acidosis, etc). Additionally, because RV load continues to decrease with prolonged support, our study suggests that “tincture of time” will lead to concomitant improvements in RV performance, although this may take months to occur. Our findings also suggest that the LVAD implant procedure itself likely leads to a decrement in RV function. Strategies aimed at minimizing RV damage, perhaps through pericardial-sparing techniques or avoiding cardiopulmonary bypass, should be further studied.^44,45

In conclusion, this study confirms that RV load sharply declines early post-LVAD. Despite this decline, the relationship of RV adaption to load worsens. The RV appears more afterload sensitive and this relationship remains constant over time. Because RV load continues to slowly decline with prolonged LVAD support, RV performance also improves.