Abstract
Late right heart failure (LRHF) following left ventricular assist device (LVAD) implantation remains poorly characterized and challenging to predict. We performed a multicenter retrospective study of LRHF in 237 consecutive adult LVAD patients, in which LRHF was defined according to the 2020 Mechanical Circulatory Support Academic Research Consortium guidelines. Clinical and hemodynamic variables were assessed pre- and post-implant. Competing-risk regression and Kaplan–Meier survival analysis were used to assess outcomes. LRHF prediction was assessed using multivariable logistic and Cox proportional hazards regression. Among 237 LVAD patients, 45 (19%) developed LRHF at a median of 133 days post-LVAD. LRHF patients had more frequent heart failure hospitalizations (p < 0.001) alongside other complications. LRHF patients did not experience reduced bridge-to-transplant rates but did suffer increased mortality (hazard ratio 1.95, 95% confidence interval [CI] 1.11–3.42; p = 0.02). Hemodynamically, LRHF patients demonstrated higher right atrial pressure, mean pulmonary arterial pressure, and pulmonary vascular resistance (PVR), but no difference in pulmonary arterial wedge pressure. History of early right heart failure, blood urea nitrogen (BUN) > 35 mg/dl at 1 month post-LVAD, and diuretic requirements at 1 month post-LVAD were each significant, independent predictors of LRHF in multivariable analysis. An LRHF prediction risk score incorporating these variables predicted LRHF with excellent discrimination (log-rank p < 0.0001). Overall, LRHF post-LVAD is more common than generally appreciated, with significant morbidity and mortality. Elevated PVR and precapillary pulmonary pressures may play a role. A risk score using early right heart failure, elevated BUN, and diuretic requirements 1 month post implant predicted the development of LRHF.
Keywords: LVAD, right ventricle, right heart failure, late onset
The use of durable, continuous-flow left ventricular assist devices (LVAD) has become more common for advanced heart failure (HF) patients, both as bridge-to-transplant (BTT) and increasingly as destination therapy (DT).1 With advancements in LVAD technology, surgical technique, and patient selection, the field has seen improvement in survival on LVAD support over the past decade, with median survival approaching 7 years.2 However, the burden of adverse events post-LVAD remains high, with right heart failure (RHF) being among the more detrimental.3,4 The right ventricle (RV) is vulnerable to failure post-LVAD because of a variety of factors, including alterations in RV preload and afterload, changes to interventricular and pericardial geometry, and underlying disease progression.5 A multitude of studies have sought to characterize and predict early RHF, which plagues the early postoperative period. Late RHF (LRHF), on the other hand, has only recently garnered increased attention.
LRHF historically has been difficult to study because of variable definitions, variable and small cohort studies, and older-generation devices with different competing risks.4,6,7 An updated Interagency Registry for Mechanical Assisted Circulatory Support (INTERMACS) analysis of 6,118 LVAD patients adds clarity to modern-day LRHF, highlighting its association with worsened survival, rehospitalization, and major adverse events.3 Meanwhile, the Mechanical Circulatory Support Academic Research Consortium (MCS-ARC) was convened to establish new consensus definitions for key LVAD adverse events, including LRHF8; these definitions will be used by INTERMACS going forward.9 Despite a better definition and understanding of LRHF, however, much about its clinical and hemodynamic characteristics remain unclear. Such insights might help improve the prediction, identification, and treatment of LVAD patients at risk of LRHF. Therefore, we sought to use the new MCS-ARC definition to investigate post-LVAD LRHF in a contemporary multicenter cohort. In this report, we investigate the characteristics of LRHF, with a focus on its hemodynamic characteristics, and identify clinical predictors that may prove useful in the prediction of LRHF risk.
Methods
Patient Population and Clinical Characterization
A retrospective analysis was performed of all consecutive adult patients (age ≥ 18 years) who underwent LVAD placement at two tertiary care hospitals, the Johns Hopkins Hospital and the Medical University of South Carolina, from November 2010 to March 2020. The study was approved by the institutional review boards at both institutions. All clinical data were collected through a review of the electronic medical record. For all patients, baseline clinical, laboratory, echocardiographic and hemodynamic parameters before LVAD implantation were obtained. Intraoperative variables were also obtained and included cardiopulmonary bypass time, surgical approach to LVAD implantation, and concomitant tricuspid valve replacement. Post-implant variables include development of early RHF (as defined by the MCS-ARC), postoperative course, laboratory data, hemodynamic characteristics, and diuretic doses at 1 month post-LVAD. Additional clinical and hemodynamic characteristics were obtained from patients within 6 to 12 months post-LVAD. Aside from the development of LRHF, other clinical outcomes were also adjudicated, including LVAD-associated complications such as arteriovenous malformation-associated gastrointestinal bleed, cardiac arrhythmias, number of hospital readmissions, BTT, and death. Patients were followed through time of death, cardiac transplant, or latest clinical follow-up at time of conclusion of the study, whichever occurred earliest.
Definitions of RHF
Definitions of RHF adhered to the 2020 MCS-ARC consensus statement. Specifically, LRHF was defined as any hospitalization occurring greater than 30 days post-LVAD implantation requiring right-sided mechanical support (including extracorporeal membrane oxygenation), intravenous diuretics, or inotrope support for at least 72 hours, and that is associated with RHF as evidenced by elevated right atrial pressure (central venous pressure >16 mmHg or jugular venous distention) with concomitant peripheral edema, ascites, renal dysfunction, or liver dysfunction.8 This definition is not limited to rehospitalizations: therefore, patients in our study who met the previously mentioned criteria within their index hospitalization for LVAD were still considered to have LRHF. To help distinguish this cohort from those who presented later with LRHF, in our study, we further distinguished LRHF as either presenting during a rehospitalization after index surgical implantation—“de novo” LRHF—versus LRHF that arose as a continuation of early RHF, or “persistent early” RHF. Patients who met LRHF criteria because of LVAD pump thrombosis or suboptimal speed were excluded from the definition.
The MCS-ARC definition of early RHF was also used in this study. This defines early RHF as persistent need for inotrope support beyond 14 days post-LVAD, need for right-ventricular assist device (RVAD) within 1 month of LVAD, or death caused by RHF within 14 days of LVAD.8 Unlike INTERMACS criteria, the MCS-ARC definition does not consider the need for inotropes within the first 14 days post-LVAD to sufficiently warrant classification as early RHF, which essentially eliminates INTERMACS cases of mild or moderate early RHF.8
Statistical Analysis
Data are presented as mean ± standard deviation for continuous variables and as proportions for categorical variables, unless otherwise indicated. Comparison of continuous variables between LRHF and non-LRHF patient groups was done using Student’s t-test. For categorical variables, χ2 tests were used. To determine predictors of LRHF, univariate logistic regression identified any preoperative, intraoperative, and early postoperative (at 1 month post implant) variables that were predictive of LRHF. Patients who died or required RVAD within 1 month of LVAD did not contribute to 1 month post-implant predictors. To generate the final multivariable model, forward stepwise regression was performed using candidate univariate predictors. Competing-risk regression and Kaplan–Meier survival analysis was used to determine the rate of death and transplant between LRHF and non-LRHF patient groups within the DT and BTT populations. Cox proportional hazards model was used to predict time-to-LRHF. A risk score predicting the development of LRHF was generated using the same variables included in our multivariable model; significance was determined by log-rank testing and area under the receiver operating characteristic curve (AUC). Statistical analyses were performed using Stata 16.0 (StataCorp, College Station, TX, USA); graphics were made using Stata 16.0 and GraphPad Prism 9.1.1 (GraphPad Software, San Diego, CA, USA).
Results
Incidence of LRHF
Of the 237 patients who underwent LVAD placement at both institutions within the study period, 45 patients (19% of the overall cohort) developed LRHF. Of the 45 patients in the LRHF group, 32 patients (71% of the LRHF cohort, 13% of the overall cohort) were considered to have de novo LRHF with onset beyond day 30 of LVAD implantation, whereas 13 LRHF patients (29% of the LRHF cohort, 5% of the overall cohort) had persistence of early RHF. Figure 1 highlights the timing of LRHF patients who developed persistence of early RHF versus de novo LRHF. The onset of LRHF occurred at a median of 133 days after LVAD implantation (Figure 1). Among the 45 LRHF patients, 18 (40%) required reinitiation of inotropes and 5 (11%) required RVAD.
Figure 1.
Timing of LRHF onset. Histogram showing number of LRHF patients by time of onset from left ventricular assist device implantation, in 3 month increments. LRHF patients further classified as those with persistence of early RHF (teal) versus de novo LRHF (navy). LRHF, late right heart failure.
Preoperative and Perioperative Characteristics
Preoperative characteristics of all patients at the time of LVAD implantation, stratified by LRHF, are shown in Table 1. Patients in the LRHF group were more likely to be male, have a higher body surface area (BSA), and have a higher blood urea nitrogen (BUN) level than those in the non-LRHF group. The indication for LVAD was more likely to be DT in the LRHF group. Notably, INTERMACS profiles were similar between groups, as were both preoperative echocardiographic and hemodynamic indices of right heart dysfunction, including RV systolic function, right atrial pressure (RAP), the RAP-to-pulmonary arterial wedge pressure (RAP/PAWP) ratio, and pulmonary artery pulsatility index.
Table 1.
Pre- and Early Postoperative Characteristics of 203 Patients Based on Development of LRHF
| Variable | Total (N = 203) | No LRHF (N = 158) | LRHF (N = 45) | p value |
|---|---|---|---|---|
|
| ||||
| Age (years) | 52.9 ± 13.5 | 53.3 ± 13.0 | 51.5 ± 14.9 | 0.385 |
| Female sex, n (%) | 55 (27) | 49 (31.0) | 6 (13.3) | 0.019 |
| BSA (m2) | 2.1 ± 0.3 | 2.0 ± 0.3 | 2.2 ± 0.3 | 0.020 |
| Race, n (%) | ||||
| White | 91 (44.8) | 74 (46.8) | 17(37.8) | 0.281 |
| Black | 89 (43.8) | 64 (40.4) | 25 (55.6) | 0.073 |
| Other | 6 (3) | 6 (3.9) | 0 (0) | 0.352 |
| iCM, n (%) | 106 (52.2) | 78 (49.4) | 28 (62.2) | 0.128 |
| CF-LVAD type, n (%) | ||||
| HM2 | 70 (34.5) | 57 (36.1) | 13 (28.9) | 0.371 |
| HW | 73 (36.0) | 58 (36.7) | 15 (33.3) | 0.677 |
| HM3 | 60 (29.6) | 43 (27.2) | 17 (37.7) | 0.171 |
| DT indication, n (%) | 111 (54.7) | 80 (50.6) | 31 (68.9) | 0.030 |
| INTERMACS, n (%) | ||||
| 1 | 31 (15.3) | 25 (15.8) | 6 (13.3) | 0.682 |
| 2 | 52 (25.6) | 43 (27.2) | 9 (20.0) | 0.328 |
| 3–7 | 119 (58.6) | 89 (56.3) | 30 (66.7) | 0.214 |
| Creatinine (mg/dl) | 1.5 ± 0.6 | 1.5 ± 0.6 | 1.6 ± 0.5 | 0.243 |
| BUN (mg/dl) | 40.0 ± 16.1 | 29.2 ± 15.5 | 37.3 ± 16.6 | 0.003 |
| Total bilirubin (mg/dl) | 1.5 ± 1.4 | 1.4 ± 1.3 | 1.6 ± 1.6 | 0.473 |
| Hemoglobin (g/dl) | 11.4 ± 2.3 | 11.3 ± 2.2 | 11.7 ± 2.7 | 0.380 |
| Atrial fibrillation, n (%) | 99 (48.8) | 75 (47.4) | 25 (53.3) | 0.487 |
| LVEF (%) | 13 ± 7 | 13 ± 7 | 14 ± 8 | 0.422 |
| Echocardiographic RV function, n (%) | ||||
| Normal | 69 (34) | 54 (34.2) | 16 (33.3) | 0.905 |
| Reduced | 79 (39) | 62 (39.2) | 15 (17.7) | 0.625 |
| Preoperative hemodynamics | ||||
| RAP (mmHg) | 13 ± 7 | 12 ± 7 | 14 ± 6 | 0.300 |
| PASP (mmHg) | 56 ± 14 | 56 ± 14 | 59 ± 15 | 0.203 |
| PADP (mmHg) | 29 ± 8 | 29 ± 8 | 30 ± 8 | 0.393 |
| mPAP (mmHg) | 39 ± 10 | 38 ± 9 | 40 ± 10 | 0.262 |
| PAWP (mmHg) | 27 ± 8 | 26 ± 7 | 28 ± 9 | 0.126 |
| RAP/PAWP | 0.5 ± 0.2 | 0.5 ± 0.2 | 0.5 ± 0.2 | 0.448 |
| CI (L/min/m2) | 2.3 ± 1.3 | 2.3 ± 1.3 | 2.5 ± 1.5 | 0.455 |
| PVR (Wood units) | 4.0 ± 2.3 | 4.0 ± 2.3 | 3.9 ± 2.5 | 0.821 |
| PAPi | 3.1 ± 2.7 | 3.3 ± 2.9 | 2.7 ± 1.8 | 0.257 |
| Preoperative IABP | 31 (15.3%) | 26 (16.5%) | 5 (11.1%) | 0.379 |
| CPB Time (min) | 89.2 ± 37.8 | 88.2 ± 39.0 | 93.4 ± 33.2 | 0.444 |
| Lateral thoracotomy, n (%) | 17 (8.4) | 15 (9.5) | 2 (4.4) | 0.272 |
| TV replacement, n (%) | 14 (6.9) | 11 (7.0) | 3 (6.7) | 0.945 |
| Need for reexploration, n (%) | 34 (16.7) | 20 (12.7) | 15 (31.1) | 0.004 |
| Early RHF, n (%) | 41 (20.2) | 24 (15.2) | 17 (37.8) | 0.001 |
| Labs at 1 month postoperative | ||||
| Creatinine (mg/dl) | 1.3 ± 0.8 | 1.2 ± 0.8 | 1.5 ± 0.8 | 0.044 |
| BUN (mg/dl) | 21.8 ± 16.2 | 19.4 ± 13.6 | 30.2 ± 21.2 | 0.0001 |
| Total bilirubin (mg/dl) | 1.0 ± 1.9 | 0.8 ± 0.6 | 1.5 ± 3.8 | 0.038 |
| Atrial fibrillation at 1 month postoperative, n (%) | 81 (39.9) | 58 (37.2) | 23 (48.9) | 0.170 |
| Diuretic dose at 1 month postoperative (furosemide equivalent/day, mg) | 92 + 88 | 82 ± 79 | 129 ± 109 | 0.002 |
BSA, body surface area; BUN, blood urea nitrogen; CF-LVAD, continuous-flow left ventricular assist device; CI, cardiac index; CPB, cardiopulmonary bypass; DT, destination therapy; HM2, HeartMate 2; HW, HeartWare; HM3, HeartMate 3; iCM, ischemic cardiomyopathy; IABP, intra-aortic balloon pump; INTERMACS, Interagency Registry for Mechanically Assisted Circulatory Support; LVEF, left ventricular ejection fraction; mPAP, mean pulmonary artery pressure; PADP, pulmonary artery diastolic pressure; PAPi, pulmonary artery pulsatility index; PASP, pulmonary artery systolic pressure; PAWP, pulmonary artery wedge pressure; PVR, pulmonary vascular resistance; RAP, right atrial pressure; RHF, right heart failure; RV, right ventricle; TV, tricuspid valve.
Table 1 also summarizes intra- and early postoperative characteristics. There was no difference between groups with regard to cardiopulmonary bypass time, rate of tricuspid valve replacement, or surgical approach used (sternotomy versus lateral thoracotomy). Patients in the LRHF group were more likely to have required reexploration in the early postoperative period (p = 0.004). Patients with LRHF were more likely to have more severe renal (p = 0.001) and liver dysfunction (p = 0.038) at 1 month post-implant. Notably, patients who developed early RHF were also more likely to develop LRHF (p = 0.001).
Hemodynamic and Clinical Characteristics of LRHF
Clinical outcomes and hemodynamic characteristics of LRHF versus control patients were compared in Table 2. Among LRHF subjects, hemodynamic parameters at 6–12 months post implant were notable for significantly higher RAP (11 ± 5 vs. 8 ± 5 mmHg, p = 0.020), higher diastolic pulmonary arterial (19 ± 5 vs. 15 ± 6 mmHg, p = 0.017), and mean pulmonary arterial (26 ± 5 vs. 22 ± 7 mmHg, p = 0.041) pressures, and higher pulmonary vascular resistance (PVR) (2.7 ± 1.4 vs. ± 0.8 Wood units, p = 0.019) (Table 2). Meanwhile, there was no significant difference in PAWP between groups. These hemodynamic findings were in contrast to the hemodynamics of both groups pre-LVAD, which were statistically similar (Figure 2). Laboratory markers of right-sided congestion, including total bilirubin and BUN, were higher among LRHF patients (Table 2). Clinically, LRHF patients required higher loop diuretic dosages than those without LRHF (145 ± 111 vs. 60 ± 91 mg/day of furosemide equivalents, p = 0.0001). Although overall rates of hospitalizations were similar between groups, LRHF patients were hospitalized more frequently for HF, as would be expected by the definition. There was also a trend toward more atrial fibrillation, ventricular tachycardia, and arteriovenous malformation-related gastrointestinal bleeding in the LRHF group (Table 2).
Table 2.
Characteristics and Outcomes of LRHF Patients
| Variable | Total (N = 203) | No LRHF (N = 158) | LRHF (N=45) | p value |
|---|---|---|---|---|
|
| ||||
| Hemodynamics within 6–12 months post-LVAD | ||||
| RAP (mmHg) | 9 ± 5 | 8 ± 5 | 11 ± 5 | 0.020 |
| PASP (mmHg) | 36 ± 10 | 35 ± 10 | 37 ± 10 | 0.351 |
| PADP (mmHg) | 16 ± 6 | 15 ± 6 | 19 ± 5 | 0.017 |
| mPAP (mmHg) | 23 ± 7 | 22 ± 7 | 26 ± 5 | 0.041 |
| PAWP (mmHg) | 14 ± 6 | 13 ± 6 | 15 ± 5 | 0.210 |
| RAP/PAWP | 0.7 ± 0.3 | 0.7 ± 0.3 | 0.8 ± 0.4 | 0.181 |
| CO (L/min) | 4.5 ± 1.3 | 4.5 ± 0.9 | 4.4 ± 1.9 | 0.96 |
| CI (L/min/m2) | 2.2 ± 0.6 | 2.2 ± 0.5 | 2.0 ± 0.7 | 0.239 |
| PVR (Wood units) | 2.3 ± 1.0 | 2.1 ± 0.8 | 2.7 ± 1.4 | 0.019 |
| PAPi | 3.0 ± 2.2 | 3.1 ± 2.0 | 2.7 ± 2.6 | 0.445 |
| Laboratory characteristics at 1-year post-LVAD | ||||
| Creatinine (mg/dl) | 1.4 ± 0.6 | 1.4 ± 0.7 | 1.4 ± 0.5 | 0.793 |
| BUN (mg/dl) | 22.6 ± 10.7 | 21.6 ± 9.3 | 26.5 ± 14.3 | 0.059 |
| Total bilirubin (mg/dl) | 0.8 ± 0.8 | 0.7 ± 0.6 | 1.1 ± 1.4 | 0.035 |
| Clinical features and outcomes | ||||
| Diuretic dose at 6 months postoperative (furosemide equivalent/day, mg) | 79 ± 102 | 60 ± 91 | 145 ± 111 | 0.0001 |
| AVM/GIB, n (%) | 41 (20.2) | 28 (17.7) | 13 (28.9) | 0.106 |
| Atrial fibrillation at 6 months postoperative, n (%) | 51 (25.1) | 35 (22.2) | 16 (35.6) | 0.051 |
| VT (sustained), n (%) | 45 (22.2) | 31 (19.6) | 14 (31.1) | 0.114 |
| Total hospitalizations up to 5 year post-VAD or until death/transplant, n | 5.0 ± 5.6 | 4.8 ± 5.4 | 5.7 ± 6.3 | 0.362 |
| Hospitalizations for HF, n | 0.8 ± 1.9 | 0.5 ± 1.5 | 2.0 ± 2.5 | <0.0001 |
| Need for inotropy at any point post discharge, n (%) | 31 (15.3) | 13 (8.2) | 18 (40.0) | <0.0001 |
| Transplant, n (%) | 65 (32.0) | 56 (35.4) | 9 (20.0) | 0.051 |
| Death, n (%) | 55 (27.1) | 38 (24.0) | 17 (37.8) | 0.049 |
AVM/GIB, arteriovenous malformation/gastrointestinal bleed; BUN, blood urea nitrogen; CI, cardiac index; CO, cardiac output; HF, heart failure; LVAD, left ventricular assist device; mPAP, mean pulmonary artery pressure; PADP, pulmonary artery diastolic pressure; PAPi, pulmonary artery pulsatility index; PASP, pulmonary artery systolic pressure; PAWP, pulmonary artery wedge pressure; PVR, pulmonary vascular resistance; RAP, right atrial pressure; VT, ventricular tachycardia.
Figure 2.
Hemodynamic characteristics of LRHF. Hemodynamic parameters were obtained before and 6–12 months after LVAD implantation. Mean ± standard error of the mean is noted for each hemodynamic parameter at both time points. *p < 0.05 for significant change after LVAD implantation. Among LRHF subjects, hemodynamic parameters at 6–12 months post implant were notable for significantly higher RAP, higher PADP and mPAP, and higher PVR. Meanwhile, there was no significant difference in PASP or PAWP between groups. Hemodynamic parameters pre-LVAD were statistically similar in both groups. LVAD, left ventricular assist device; LRHF, late right heart failure; mPAP, mean pulmonary artery pressure; RAP, right atrial pressure; PADP, pulmonary artery diastolic pressure; PASP, pulmonary artery systolic pressure; PAWP, pulmonary arterial wedge pressure; PVR, pulmonary vascular resistance.
Outcomes of LRHF Patients
In the overall cohort, LRHF conferred a nearly two-fold increased risk of death within 5 years (subdistribution hazard ratio 1.95, 95% confidence interval [CI] 1.11–3.42; p = 0.02), assessed using competing-risks regression, which takes into account the competing outcome of heart transplantation (Figure 3). Among those implanted with BTT intent, there was also an increased risk of death (subdistribution hazard ratio 2.98, 95% CI 1.14–7.74; p = 0.025). However, among BTT patients, there was no significant difference in the rate of successful bridge to heart transplantation (subdistribution hazard ratio 0.75, 95% CI 0.26–2.13, p = 0.59).
Figure 3.
Impact of LRHF on outcomes. Competing-risk regression of LRHF on death in the overall cohort, with heart transplantation as competing outcome. LRHF conferred a nearly two-fold increased risk of death (subdistribution hazard ratio 1.95, 95% CI1.11–3.42, p = 0.02). Notably, LRHF did not significantly impact the rate of successful bridge to heart transplantation in the BTT cohort (subdistribution hazard ratio 0.75, 95% CI 0.26–2.13, p = 0.59). BTT, bridge-to-transplant; CI, confidence interval; LRHF, late right heart failure; SHR, subdistribution hazard ratio.
Predictors of LRHF
We sought to identify LRHF predictors using preoperative and postoperative patient characteristics. Univariable logistic regression revealed several predictors (Table 3), which were used to generate a multivariable model. The final multivariable model included the following postoperative variables: development of early RHF, BUN > 35 mg/dl at 1 month post implant, and diuretic dose at 1 month post implant (Table 4). Each variable predicted LRHF independent of the others (Table 4). Each of these variables also independently predicted the time to onset of LRHF in a multivariable Cox proportional hazards regression model (Table 4).
Table 3.
Univariable Predictors of LRHF
| Variable | OR | SE | 95% CI | p value |
|---|---|---|---|---|
|
| ||||
| Baseline characteristics | ||||
| Female sex | 0.34 | 0.16 | 0.14–0.86 | 0.023 |
| BSA | 4.26 | 0.63 | 1.20–15.18 | 0.025 |
| Diabetes | 1.84 | 0.63 | 0.94–3.58 | 0.075 |
| DT indication | 2.16 | 0.78 | 1.07–4.36 | 0.032 |
| BUN | 1.03 | 0.01 | 1.01–1.05 | 0.004 |
| HM3 | 1.62 | 0.58 | 0.81–3.26 | 0.173 |
| RAP > 10 mmHg | 1.60 | 0.54 | 0.83–3.14 | 0.160 |
| PAWP | 1.04 | 0.03 | 0.99–1.09 | 0.128 |
| Perioperative factors | ||||
| Need for reexploration | 3.09 | 1.24 | 1.41–6.79 | 0.005 |
| Early RHF (INTERMACS) | 3.09 | 1.56 | 1.14–8.33 | 0.025 |
| Early RHF (severe/MCS-ARC) | 3.39 | 1.28 | 1.61–7.13 | 0.001 |
| Creatinine at 1 month postoperative | 1.47 | 0.30 | 0.99–2.19 | 0.056 |
| BUN at 1 month postoperative | 1.04 | 0.01 | 1.02–1.06 | <0.0001 |
| BUN at 1 month postoperative > 35 mg/dl | 4.65 | 1.98 | 2.01–10.72 | 0.0004 |
| Hemoglobin at 1 month postoperative | 0.78 | 0.14 | 0.55–1.10 | 0.155 |
| Atrial fibrillation at 1 month postoperative | 1.6 | 0.56 | 0.80–3.20 | 0.180 |
| Diuretic dose at 1 month postoperative (per 100 mg furosemide equivalent) | 1.64 | 0.30 | 1.15–2.34 | 0.006 |
OR, SE, 95% CI, and p values are reported for each.
BSA, body surface area; BUN, blood urea nitrogen; CI, confidence interval; DT, destination therapy; HM3, HeartMate 3; INTERMACS, Interagency Registry for Mechanically Assisted Circulatory Support; MCS-ARC, Mechanical Circulatory Support Academic Research Consortium; OR, odds ratios; PAWP, pulmonary arterial wedge pressure; RAP, right atrial pressure; RHF, right heart failure; SE, standard error.
Table 4.
Multivariable Regression Analyses
| Multivariable logistic regression for LRHF | ||||
|---|---|---|---|---|
|
| ||||
| Variable | OR | SE | 95% CI | p value |
|
| ||||
| BUN at 1-month post-op > 35 mg/dl | 3.23 | 1.50 | 1.31–7.98 | 0.011 |
| Early RHF | 2.52 | 1.06 | 1.11–5.74 | 0.027 |
| Diuretic dose at 1-month post-op (per 100mg furosemide equivalent) | 1.62 | 0.31 | 1.11–2.35 | 0.012 |
| Cox proportional hazards regression for time-to-LRHF | ||||
| Variable | HR | SE | 95% CI | p value |
| BUN at 1-month post-op > 35 mg/dl | 2.98 | 1.01 | 1.53–5.81 | 0.001 |
| Early RHF | 2.22 | 0.72 | 1.18–4.19 | 0.013 |
| Diuretic dose at 1-month post-op (per 100mg furosemide equivalent) | 1.47 | 0.21 | 1.12–1.94 | 0.006 |
OR or HR, SE, 95% CI, and p values are reported for each.
95% CI, 95% confidence intervals; BUN, blood urea nitrogen; HR, hazard ratios; OR, odds ratios; RHF, right heart failure; SE, standard error.
Next, each variable was incorporated into an LRHF Risk Score to predict the development of LRHF post-LVAD. Points were derived using relative odds ratios as follows: 3 points for BUN > 35 mg/dl, 2.5 points for early RHF, and 1.5 points per 100 mg/day of furosemide equivalents. This LRHF Risk Score strongly predicted the development of LRHF in our cohort (Figure 4, log-rank p < 0.0001). Receiver operating characteristic analysis determined that our risk score was able to predict LRHF with an AUC of 0.73. Patients with a risk score of 0 had a <15% chance of developing LRHF within 3 years, whereas those with a risk score of >3 had >50% chance of developing LRHF within 3 years (Figure 4).
Figure 4.
LRHF Risk Score. A risk score predicting development of LRHF was generated using elevated BUN, early RHF, and diuretic requirements 1 month post implant, with 3 points for BUN > 35 mg/dl, 2.5 points for early RHF, and 1.5 points per 100 mg/day of furosemide equivalents. This risk score strongly predicted the development of LRHF in our combined cohort (log-rank p < 0.0001, receiver operating characteristic curve 0.73). Patients with a risk score of 0 had a <15% chance of developing LRHF within 3 years, whereas those with a risk score of >3 had >50% chance of developing LRHF in that timeframe. BUN, blood urea nitrogen; LRHF, late right heart failure.
While male sex was also independently predictive of LRHF in a multivariable logistic regression model—alongside early RHF, BUN, and diuretic requirement—it was not independently predictive in the multivariable Cox proportional hazard regression model (Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A848 ). For this reason and because of cohort size, male sex was excluded from the final model. That said, because of possible underpowering, we included a modified LRHF Risk Score with the inclusion of sex in our Supplemental data (Supplemental Figure 1, Supplemental Digital Content 1, p < 0.0001 http://links.lww.com/ASAIO/A848 ).
Finally, we investigated the ability of these models to predict only de novo LRHF. In a sensitivity analysis of patients who developed de novo LRHF, excluding cases of persistent early RHF, the LRHF Risk Score remained strongly predictive of LRHF (Supplemental Figure 2, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A848, p = 0.01). Moreover, logistic and Cox proportional hazard regression models using only BUN > 35 mg/dl and diuretic requirements remained overall predictive of de novo LRHF (Supplemental Table 2, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A848).
Discussion
LRHF significantly contributes to morbidity and mortality post-LVAD, and yet remains poorly understood.3,4 In this multicenter study, with LRHF defined according to an updated consensus,8 we find that (1) LRHF post-LVAD is actually more prevalent than previously appreciated; (2) LRHF is associated with considerable morbidity and increased mortality, but fortunately similar rates of BTT; (3) patients with LRHF have higher residual PVR, associated with greater pulmonary and right atrial pressures; and (4) a risk score using early RHF, elevated BUN, and diuretic requirements 1 month post implant was able to predict the development of LRHF. Our findings improve our understanding of the clinical characteristics and outcomes of LRHF, suggest possible hemodynamic etiologies for LRHF development, and offer a means to predict patients at greater risk of LRHF post-LVAD, with the hope that earlier identification of risk can lead to improved observation and intervention.
The prevalence of LRHF in this multicenter cohort is 19%. Although this figure is higher than previously reported estimates of 7%–11%,4,6,7 it is notably identical to a recent, large INTERMACS report of 6,118 LVAD patients that found that 19% of LVAD patients met criteria for moderate RHF at 1 month post implant.3 Even if persistent early RHF cases were excluded, the prevalence of LRHF (i.e., only de novo LRHF cases) was still 13%. Identifying a higher prevalence of LRHF post-LVAD has important implications for modern LVAD management. Not only does it change the paradigm with which we consider LRHF post-LVAD, but it also affects our assessment, differential diagnosis, and management of HF in post-LVAD patients. One reason for this evolving prevalence may be that most prior reports of LRHF employed variable definitions that restricted LRHF patients to those who were readmitted for RHF sometime after index discharge or required reinitiation of inotropes.4,6,7 These studies were rather specific in their definition of LRHF. This is in contrast to the updated 2020 MCS-ARC definition, which includes not only those hospitalized 1 month post implant for RHF requiring inotropic support but also those with RHF hospitalizations requiring intravenous diuretics, those requiring delayed RVAD during their index hospitalization, and those with persistent early RHF.8
Consistent with prior reports, LRHF is associated with increased mortality and morbidity, primarily in the form of increased HF hospitalizations. In our cohort, LRHF was unsurprisingly associated with reduced survival, confirming multiple prior reports.3,8 In the current study, we found no difference among LRHF patients in the rate of successful BTT, similar to prior reports.3,4 It should be noted, however, that LVAD patients with LRHF have been shown to suffer worse outcomes post-transplantation because of the impact of LRHF on hepatorenal end-organ dysfunction.10 The increased morbidity and mortality of LRHF were consistent across all LVAD types, including axial, intrapericardial, and intrapericardial-fully magnetically levitated pumps. The effect of LRHF on LVAD HF hospitalizations and mortality is important to note in this current era of LVAD implantation. As DT patients live longer on LVAD support and BTT patients wait longer for heart transplantation in the current heart allocation era, the prevalence of LRHF will only increase, along with its associated morbidity and mortality.1,11
Our multicenter registry allowed for an in-depth assessment of the hemodynamic profile of LRHF. LRHF LVAD patients had higher PVR and greater pulmonary artery diastolic, mean pulmonary artery, and RA pressures as compared with their non-LRHF counterparts. Elevated pulmonary pressures were present in spite of comparable PAWP between groups and, intriguingly, indistinguishable precapillary hemodynamic profiles before LVAD. These findings are in line with other studies showing an association between RHF and an elevated diastolic pulmonary gradient (DPG) in the LVAD population.12,13 There was also a trend toward increased atrial fibrillation in the LRHF group. Collectively, these data suggest persistent pulmonary vascular pathology and atrial arrhythmias as possible culprits of LRHF post-LVAD.14 Alternatively, elevated right-sided filling pressures may be indicative of intrinsic RHF simply unmasked after LVAD surgery; in this scenario, elevated pulmonary pressures may be a by-product of enhanced pericardial restraint resulting from right heart dysfunction. Indeed, the aforementioned association between DPG and RHF could, in some cases, be mitigated by LVAD speed adjustments, suggesting not all these cases are due to left heart disease inducing precapillary pulmonary hypertension.12,13 Other factors including residual mitral regurgitation may also contribute to persistent pulmonary hypertension post-LVAD.15 These hemodynamic characteristics offer possible areas for intervention in the LRHF cohort. Overall, the pathophysiologic connections between RHF and the left heart, pulmonary vasculature, and pericardial space remain a ripe area for investigation in the LVAD population.
Although many risk factors have been identified and prediction models developed for early RHF after LVAD, very little is known about predictors of LRHF.16–21 Interestingly, our study showed no significant impact of preimplant hemodynamic indices of right heart dysfunction, consistent with previous reports,4,6 though our study may have been underpowered to detect preimplant differences.3 For example, male sex was predictive of LRHF as a univariate predictor as well as in the multivariable model, but not in the Cox model. Male sex as a risk factor for LRHF has plausible pathophysiology, as it has been shown in pulmonary arterial hypertension that men have reduced RV–pulmonary arterial coupling because of poorer intrinsic RV contractility.22,23 Additionally, preimplant elevation in BUN and BSA were associated with LRHF, findings also seen in prior studies.4 Previous studies examining the impact of body mass index on LVAD outcomes have shown higher complication rates in obese patients with LVAD.24 A possible link between obesity and intrinsic RV contractile dysfunction, seen in HF patients with preserved ejection fraction, may be at play in this population as well.25,26 DT indication was also significantly associated with LRHF, but this was likely because of the increase in age and comorbidities, and the reduced rate of heart transplantation inherent to this cohort. As for operative factors, lateral thoracotomy was not found to be protective of LRHF, statistically speaking. That said, of the 17 patients in our cohort who underwent lateral thoracotomy, 15/17 were spared of LRHF whereas only 2/17 developed it (p = 0.27). Ultimately, these results were limited by the low rate of lateral thoracotomy usage (8.4%) in this cohort. Given growing data that lateral thoracotomy protects against early RHF,27,28 it stands to reason that it may protect against LRHF. Future studies will hopefully uncover this. Postoperative predictors included the need for surgical reexploration, which may have mediated its risk by increasing the risk of early RHF, perhaps by repeated insults to pericardial integrity27,28; that said, reexploration for reasons like bleeding and tamponade may have just correlated with early severe RHF itself, which would explain why it did not remain an independent predictor in multivariable modeling.
Early prediction of LRHF risk would greatly help identify at-risk patients in need of increased monitoring and right heart optimization. In our study, an LRHF prediction model incorporating three easily attainable early postoperative predictors—early RHF, elevated BUN, and high diuretic requirements—not only predicted incidence but also time to LRHF. To our knowledge, this is the first risk score developed to predict post-LVAD LRHF. Given its associated increased morbidity and mortality, the ability to predict the development of LRHF as early as 1 month post implant would help to guide earlier, more aggressive treatment strategies aimed at preventing downstream consequences. Although LRHF can manifest years post-LVAD, our data show that many develop LRHF within a few months, further highlighting the value of early risk prediction. Additionally, RHF is an important indication for BTT status upgrade, and LRHF among BTT LVAD candidates may worsen outcomes even post-transplant. Thus, among BTT patients, earlier identification of LRHF risk might help improve BTT listing strategies while also helping to avoid the consequences of LRHF-mediated end-organ dysfunction, which might mitigate the effect of LRHF on posttransplant survival.10 For the DT population, earlier identification of patients at risk for LRHF would also help guide more aggressive HF mitigation strategies. Angiotensin receptor-neprilysin inhibitors (ARNIs), for example, are known to lower pulmonary pressures in end-stage HF patients and play a useful role in mitigating residual diuretic requirements and HF symptoms in the LVAD population.29,30 ARNIs may demonstrate similar utility for post-LVAD LRHF patients as well. Exactly what interventions might prove best for LRHF admittedly remains unclear and should be the focus of further study, but earlier identification nonetheless remains the first step.
Limitations
Limitations of this study include its retrospective design, which led to the variable acquisition of pre- and post-implant data. That said, the characteristics studied all represent commonly obtained variables that are also obtained for registry purposes. The inclusion of more centers would have improved generalizability and multivariable modeling. In particular, male sex may have proven to be another independent risk factor in multivariable modeling if our cohort had been larger. Lastly, although our risk score was developed using multicenter data, the score was applied to the patient population from which it was developed and has not been externally validated. Future directions should be targeted toward prospective data collection with MCS-ARC criteria in mind, as well as validation of our LRHF-prediction risk score in an external cohort.
Conclusions
LRHF may be more common after LVAD than previously appreciated, occurring in 19% of our combined cohort. The onset of LRHF is associated with increased morbidity, including more frequent HF hospitalizations, as well as increased mortality. Hemodynamic characteristics of LRHF patients reveal elevated PVR and precapillary pulmonary and right atrial pressures, suggesting potential contributions to the development of LRHF. A simple LRHF Risk Score including the development of early RHF and elevated BUN and diuretic requirements post-LVAD predicted LRHF with excellent discrimination. Future studies should further investigate LRHF pathophysiology and treatment strategies.
Supplementary Material
Acknowledgments
S. Hsu received the following sources of funding: P.I., NIH/NHLBI K23-HL146889–01; Co-investigator., NIH/NHLBI R01-HL114910–06.
Footnotes
Disclosure: The only disclosures to note are those of Dr. Ryan Tedford, who reports no direct conflicts of interest related to this manuscript but does report general disclosures to include consulting relationships with Medtronic, Abbott, Aria CV Inc., Acceleron, Itamar, Edwards LifeSciences, Eidos Therapeutics, Lexicon Pharmaceuticals, Gradient and United Therapeutics. Dr. Tedford is on a steering committee for Medtronic, Acceleron, and Abbott as well as on a research advisory board for Abiomed. He also does hemodynamic core lab work for Actelion and Merck.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text, and links to the digital files are provided in the HTML and PDF versions of this article on the journal’s Web site (www.asaiojournal.com).
The other authors have no conflicts of interest to report..
References
- 1.Molina EJ, Shah P, Kiernan MS, et al. : The Society of Thoracic Surgeons Intermacs 2020 Annual Report. Ann Thorac Surg 111: 778–792, 2021. [DOI] [PubMed] [Google Scholar]
- 2.Zimpfer D, Fiane AE, Larbalestier R, et al. : Long-term survival of patients with advanced heart failure receiving an left ventricular assist device intended as a bridge to transplantation. Circ Hear Fail 13: e006252, 2020. [DOI] [PubMed] [Google Scholar]
- 3.Rame JE, Pagani FD, Kiernan MS, et al. : Evolution of late right heart failure with left ventricular assist devices and association with outcomes. J Am Coll Cardiol 78: 2294–2308, 2021. [DOI] [PubMed] [Google Scholar]
- 4.Takeda K, Takayama H, Colombo PC, et al. : Incidence and clinical significance of late right heart failure during continuous-flow left ventricular assist device support. J Heart Lung Transplant 34: 1024–1032, 2015. [DOI] [PubMed] [Google Scholar]
- 5.Lampert BC, Teuteberg JJ: Right ventricular failure after left ventricular assist devices. J Heart Lung Transplant 34: 1123–1130, 2015. [DOI] [PubMed] [Google Scholar]
- 6.Kormos RL, Teuteberg JJ, Pagani FD, et al. ; HeartMate II Clinical Investigators: Right ventricular failure in patients with the HeartMate II continuous-flow left ventricular assist device: Incidence, risk factors, and effect on outcomes. J Thorac Cardiovasc Surg 139: 1316–1324, 2010. [DOI] [PubMed] [Google Scholar]
- 7.Rich JD, Gosev I, Patel CB, et al. ; Evolving Mechanical Support Research Group (EMERG) Investigators: The incidence, risk factors, and outcomes associated with late right-sided heart failure in patients supported with an axial-flow left ventricular assist device. J Heart Lung Transplant 36: 50–58, 2017. [DOI] [PubMed] [Google Scholar]
- 8.Kormos RL, Antonides CFJ, Goldstein DJ, et al. : Updated definitions of adverse events for trials and registries of mechanical circulatory support: A consensus statement of the mechanical circulatory support academic research consortium. J Heart Lung Transplant 39: 735–750, 2020. [DOI] [PubMed] [Google Scholar]
- 9.Teuteberg JJ, Cleveland JC Jr, Cowger J, et al. : The Society of Thoracic Surgeons Intermacs 2019 Annual Report: the changing landscape of devices and indications. Ann Thorac Surg 109: 649–660, 2020. [DOI] [PubMed] [Google Scholar]
- 10.Takeda K, Takayama H, Colombo PC, et al. : Late right heart failure during support with continuous-flow left ventricular assist devices adversely affects post-transplant outcome. J Heart Lung Transplant 34: 667–674, 2015. [DOI] [PubMed] [Google Scholar]
- 11.Patel JN, Chung JS, Seliem A, et al. : Impact of heart transplant allocation change on competing waitlist outcomes among listing strategies. Clin Transplant 35: e14345, 2021. [DOI] [PubMed] [Google Scholar]
- 12.Imamura T, Chung B, Nguyen A, et al. : Decoupling between diastolic pulmonary artery pressure and pulmonary capillary wedge pressure as a prognostic factor after continuous flow ventricular assist device implantation. Circ Heart Fail 10: e003882, 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Imamura T, Narang N, Kim G, et al. : Decoupling between diastolic pulmonary artery and pulmonary capillary wedge pressures is associated with right ventricular dysfunction and hemocompatibility-related adverse events in patients with left ventricular assist devices. J Am Heart Assoc 9: e014801, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Frankfurter C, Molinero M, Vishram-Nielsen JKK, et al. : Predicting the risk of right ventricular failure in patients undergoing left ventricular assist device implantation: A systematic review. Circ Heart Fail 13: e006994, 2020. [DOI] [PubMed] [Google Scholar]
- 15.Gulati G, Ruthazer R, Denofrio D, Vest AR, Kent D, Kiernan MS: Understanding longitudinal changes in pulmonary vascular resistance after left ventricular assist device implantation. J Card Fail 27: 552–559, 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.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. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Fitzpatrick JR 3rd, Frederick JR, Hsu VM, et al. : Risk score derived from pre-operative data analysis predicts the need for biventricular mechanical circulatory support. J Heart Lung Transplant 27: 1286–1292, 2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Soliman OII, Akin S, Muslem R, et al. ; EUROMACS Investigators: Derivation and validation of a novel right-sided heart failure model after implantation of continuous flow left ventricular assist devices: The EUROMACS (European Registry for Patients with Mechanical Circulatory Support) right-sided heart failure risk score. Circulation 137: 891–906, 2018. [DOI] [PubMed] [Google Scholar]
- 19.Atluri P, Goldstone AB, Fairman AS, et al. : Predicting right ventricular failure in the modern, continuous flow left ventricular assist device era. Ann Thorac Surg 96: 857–863; discussion 863, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Drakos SG, Janicki L, Horne BD, et al. : Risk factors predictive of right ventricular failure after left ventricular assist device implantation. Am J Cardiol 105: 1030–1035, 2010. [DOI] [PubMed] [Google Scholar]
- 21.Wang Y, Simon MA, Bonde P, et al. : Decision tree for adjuvant right ventricular support in patients receiving a left ventricular assist device. J Heart Lung Transplant 31: 140–149, 2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Jacobs W, van de Veerdonk MC, Trip P, et al. : The right ventricle explains sex differences in survival in idiopathic pulmonary arterial hypertension. Chest 145: 1230–1236, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tello K, Richter MJ, Yogeswaran A, et al. : Sex differences in right ventricular-pulmonary arterial coupling in pulmonary arterial hypertension. Am J Respir Crit Care Med 202: 1042–1046, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Jaiswal A, Truby LK, Chichra A, et al. : Impact of obesity on ventricular assist device outcomes. J Card Fail 26: 287–297, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Agrawal V, Fortune N, Yu S, et al. : Natriuretic peptide receptor C contributes to disproportionate right ventricular hypertrophy in a rodent model of obesity-induced heart failure with preserved ejection fraction with pulmonary hypertension. Pulm Circ 9: 2045894019878599, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Aslam MI, Hahn VS, Jani V, Hsu S, Sharma K, Kass DA: Reduced right ventricular sarcomere contractility in heart failure with preserved ejection fraction and severe obesity. Circulation 143: 965–967, 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Carmona A, Hoang Minh T, Perrier S, et al. : Minimally invasive surgery for left ventricular assist device implantation is safe and associated with a decreased risk of right ventricular failure. J Thorac Dis 12: 1496–1506, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Saeed D, Muslem R, Rasheed M, et al. : Less invasive surgical implant strategy and right heart failure after LVAD implantation. J Heart Lung Transplant 40: 289–297, 2021. [DOI] [PubMed] [Google Scholar]
- 29.Zern EK, Cheng S, Wolfson AM, et al. : Angiotensin receptor-neprilysin inhibitor therapy reverses pulmonary hypertension in end-stage heart failure patients awaiting transplantation. Circ Heart Fail 13: e006696, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Goldberg RL, Freed KE, Klemans N, et al. : Angiotensin receptor-neprilysin inhibition improves blood pressure and heart failure control in left ventricular assist device patients. ASAIO J 67: e207–e210, 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.