Recurrent Heart Failure after Left Ventricular Assist Device Placement

Abstract

A host of complications are common after left ventricular assist device (LVAD) surgery. Perhaps none is more challenging to manage than recurrent heart failure (HF). HF in an LVAD patient is associated with substantial morbidity and increased mortality. HF can occur early or late, can present abruptly or insidiously, and can be due to an array of LVAD-specific problems including pump thrombosis and cannula obstruction, or intrinsic cardiac problems such as right ventricular failure or valvular disease. These disparate etiologies require specific testing and distinct therapeutic strategies. This chapter reviews the causes of recurrent HF after LVAD surgery with particular attention to evaluation and management strategies that can identify and treat these distinct etiologies.

1. Introduction

Heart failure (HF) is among the commonest chronic diseases, with a prevalence of 6.2 million adults in the United States (U.S.) and 64.3 million worldwide [1, 2]. Patients with the most advanced form of HF are classified as stage D and have high mortality. Population studies estimate the prevalence of stage D HF to be between 0.2–3.0% [3], with up to 4.5% of patients with chronic (stage C) HF progressing to stage D per year [4]. In the U.S., this is ~150,000–250,000 with stage D disease [3, 5]. Orthotopic heart transplantation (OHT) remains the gold standard therapy for this population. However, there remains a tremendous shortfall of available organs; despite recent increases, only 3817 OHT surgeries were performed in the U.S. in 2021 (~1.5–2.5% of the estimated stage D population) [6].

Left ventricular assist device (LVAD) surgery has revolutionized the treatment of end-stage HF, providing increased longevity and a superior quality of life (QOL) for patients with stage D disease. LVAD technology has progressed dramatically over the last 3 decades, with each successive generation of pump providing robust improvements in outcomes [7]. However, LVAD-associated morbidity remains substantial (Figure 1) [8, 9]. Among the commonest complications is recurrent HF. This is almost uniformly associated with increased mortality and worse QOL.

Figure 1.

Contemporary adverse event rates after LVAD. Events per patient-year (EEPY) occurring >90 days after implant in the STS-INTERMACS registry [8], or from the MOMENTUM 3 continued access protocol (CAP) trial [9]. *No event rate data exist for right HF from STS-INTERMACS.

Recurrent HF after LVAD has many causes and critically, management depends on the underlying etiology. Recurrent HF is either related to problems extrinsic to the LVAD or to a problem with the device itself (Figure 2) [10]. By far the commonest cause is failure of the right ventricle (RV). This chapter will review the causes of recurrent HF and provide strategies to diagnose and manage these distinct problems.

Figure 2.

Causes of recurrent heart failure (HF) after LVAD implant. HF can be caused by intrinsic cardiac disease that is pre-existing or that develops after LVAD implant, or may be secondary to severe anemia in the setting of GI bleeding. Alternately, HF may result from LVAD-specific issues, including outflow or inflow cannula obstruction, pump failure, or simply an inappropriate LVAD set speed.

2. Causes of recurrent heart failure

2.1 Right heart failure

2.1.1 Defining right heart failure

Failure of the RV is the commonest cause of recurrent HF after LVAD surgery. The reported prevalence of RV failure is extremely variable, ranging from 4 to 40% for continuous flow devices [11]. This variance is driven by a lack of standardization in post-operative management, differences in patient characteristics between implanting centers, and the wide range in follow-up time across studies.

The evolution of LVAD technology and usage have impacted the prevalence of RV failure. Though the pulsatile HeartMate XVE was approved for use as destination therapy (DT) in 2003, 2-year survival was low (23% in the REMATCH trial [12], 33% in a post-REMATCH registry [13], and 24% in the HeartMate II (HMII) DT trial [14]) owing at least in part to mechanical failure of this pump. Further, the size of the HeartMate XVE restricted its use to larger patients. These factors limited long-term use, and bridge-to-transplantation (BTT) remained the dominant implant strategy in the first decade of the 2000s [15]. Consequently, analyses of HF in LVAD patients from >10 years ago largely focused on early post-operative RV failure. However, since approval of the HMII for DT in January 2010, DT has become the dominant implant strategy in the U.S. (Figure 3) [16]. This has led to substantially longer time on LVAD support with a concomitant shift in patient characteristics and outcomes. Finally, changes to the United Network for Organ Sharing OHT listing criteria in 2018 and advances in the use of temporary mechanical circulatory support (MCS), have resulted in an even more significant reduction in BTT LVAD usage in the last 3 years [17].

Figure 3.

Evolution of continuous flow LVAD implant strategies in the U.S. (2010–2019). The total number of implants per year is listed below each year. From: STS-INTERMACS database [16].

Analysis of recurrent HF has also been hampered by the variable definitions used for RV failure. Nearly all studies define RV failure when a right ventricular assist device (RVAD) is required, but inotrope use as a criterion has been variable as has the requirement for clinical signs of HF. In 2008, the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) defined RV failure by a central venous pressure (CVP) >18 mmHg, cardiac index <2.0 L/min/m² and either the need for an RVAD, or any use of vasoactive medications >7 days after LVAD implant. The limitations of this definition were quickly recognized, and it was revised in 2014 (Table 1). This refined definition required (1) elevated right sided filling pressures and (2) physical or laboratory evidence for congestion. If these criteria were met, then the severity of RV failure was further qualified.

Part 1: Symptoms or findings of right heart failure (need to meet both)
Documented elevation of CVP	Directly measured right atrial pressure > 16 mmHg — OR — Dilated IVC with absence of inspiratory variation by echocardiography — OR — elevated JVP halfway up neck in an upright patient
Manifestations of systemic venous congestion	Peripheral edema — OR — Ascites ± palpable hepatomegaly — OR — Laboratory evidence: serum creatinine >2 mg/dL or total bilirubin >2.0 mg/dL
Part 2: If patient meets the definition for right heart failure (Part 1), then severity is classified as:
Mild	Vasoactive meds (inotropes, vasopressors, vasodilators) are not continued for >7 days post-LVAD implant
Moderate	Vasoactive meds (inotropes, vasopressors, vasodilators) are required for >7 but ≤14 days post-LVAD implant
Severe	Vasoactive meds (inotropes, vasopressors, vasodilators) are required >14 days post-LVAD implant; — AND —; Persistently elevated CVP >16 mmHg
Severe acute	Need for RVAD support at any time following LVAD implant; — OR —; Death during the LVAD implant hospitalization with right heart failure as the primary cause

Table 1.

2014 INTERMACS definition of right heart failure.

CVP – central venous pressure; IVC – inferior vena cava; JVP – jugular venous pressure; RVAD – right ventricular assist device.

Though the 2014 definition was more inclusive, some patients with RV failure were still not captured, and hybrid definitions remained commonplace. In 2020, the Academic Research Consortium convened a multidisciplinary working group to define adverse events related to MCS use (MCS-ARC) [18]. In this simplified definition, RV failure is divided into early or late based on the timing relative to LVAD surgery (Table 2). The need for an RVAD continues to define right HF, while vasoactive medication use without RVAD requires that additional clinical criteria be met including findings of elevated right atrial pressure, or evidence of end organ dysfunction/hypoperfusion. In late 2021, the Society of Thoracic Surgeons (STS)-INTERMACS database adopted this MCS-ARC definition of right HF [19].

Early acute right heart failure	Need for RVAD support (temporary or durable) during the LVAD implant operation
Early post-implant right heart failure	Need for RVAD support (temporary or durable) <30 days following LVAD implant — OR — Failure to wean vasoactive medication (inotropes, vasopressors or inhaled pulmonary vasodilators) ≤14 days following LVAD implant OR initiation of vasoactive medication support ≤30 days of LVAD implant for a duration of ≥14 days PLUS 2 clinical findings* or 1 manifestation† of RV failure
Late right heart failure	Need for RVAD support (temporary or durable) >30 days following LVAD implant — OR — Hospitalization >30 days following LVAD implant and which requires IV diuretics and/or inotropic support for ≥72 hours PLUS 2 clinical findings* or 1 manifestation† of RV failure

Table 2.

2020 MCS-ARC & 2021 STS-INTERMACS definition of right heart failure.

^*

Clinical findings of RV failure: (1) ascites; (2) functional/limiting peripheral edema; (3) elevated JVP halfway up the neck in an upright patient; (4) measured CVP >16 mmHg.

^†

Manifestations of RV failure: (1) serum creatinine >2 times baseline; (2) ALT/AST ≥2 times the upper limit of normal or total serum bilirubin >2.0 mg/dL; (3) reduction in LVAD pump flow >30% below baseline in the absence of cardiac tamponade; (4) central or mixed venous blood oxygen <50%; (5) cardiac index <2.2 L/min/m2; (6) serum lactate >3.0 mmol/L.

2.1.2 Right heart function in the LVAD patient

The RV is anatomically and physiologically distinct from the LV [20]. Under normal conditions, RV output is roughly equal to that of the LV. However, the mechanics of RV contraction are distinct. The RV is thin walled and shaped like a tetrahedron (Figure 4). It is highly compliant and pumps blood into the low impedance pulmonary vasculature. Consequently, the RV requires only one sixth the energy of the LV per contraction [20]. As the LV contracts, it twists around its longitudinal axis; this twisting motion (akin to wringing a wet towel) contributes significantly to septal contraction. Meanwhile the RV contracts along longitudinal and transverse axes, with longitudinal shortening being the major driver of RV stroke volume (Figure 5) [21]. Importantly, a significant portion of longitudinal RV contractility is derived from the septum.

Figure 4.

Anatomy and geometry of the right ventricle. Transverse (A), coronal (B) and sagittal (C) views of the heart showing the unique tetrahedral or half-ellipsoid shape of the RV. The sagittal view (C) is from the perspective of the interventricular (IV) septum in the foreground and looking “into” the RV towards the RV free wall. Lines correspond to the following: WX = junction of the interatrial and IV septa; WY = anterior atrioventricular sulcus; XZ = posterior atrioventricular sulcus; WZ = anterior IV sulcus; XZ = posterior IV sulcus; YZ = margo acutus. RV regions correspond to the following: WXY = approximate valvualr plane; WYZ = anterior RV free wall; XYZ = posterior RV free wall; WXZ = IV septum.

Figure 5.

Vectors of ventricular contraction. (A) In the normal heart, the LV contracts in a wringing, or spiral, motion around its long axis while the normal RV contracts in perpendicular planes along its longitudinal and transverse axes. This LV twisting motion augments septal contractility and RV stroke volume. (B) An LVAD limits LV twisting and triggers a leftward shift of the interventricular septum by decreasing LV preload, in turn reducing septal contraction and overall RV longitudinal contractility. If RV free wall contractility cannot increase to compensate, then total RV cardiac output may decline.

RV function is primarily governed by three physiologic parameters: (1) preload; (2) contractility; and (3) afterload. In chronic HF, primary RV dysfunction (i.e., independent of LV failure) is common, being identified in about half of patients with HF and reduced ejection fraction [22]. RV afterload is determined by pulmonary vascular resistance (PVR) and compliance. The RV displays a steep decline in cardiac output with increasing PVR [20]. In chronic HF, PVR rises and compliance declines secondary to (1) elevated left heart pressures and (2) pulmonary arterial remodeling, thus substantially increasing RV afterload.

Right HF after LVAD implant is multifactorial. The abrupt increase in venous return to the right heart can cause HF in a myopathic RV that fails to adequately compensate for the increased preload. RV contractility can be compromised by LVAD support in multiple ways (Figure 5): (1) reduced LV preload leads to a leftward shift of the septum, limiting the contribution of septal contraction to RV force generation; (2) apical LVAD insertion coupled with the loss of pericardial constraint reduces LV contractility (especially twisting), which also limits septal contraction; and (3) the leftward shift of the septum can lead to stretching of the tricuspid valve (TV) annulus with a concomitant increase in tricuspid regurgitation (TR). Finally, the abrupt reduction in left heart filling pressures after LVAD typically improves PVR, reducing RV afterload and improving contractility [23]. However, PVR (and therefore RV afterload) may remain elevated due to vascular remodeling, thus further contributing to post-LVAD RV failure.

2.1.3 Early right heart failure

Early right HF increases the risk of death and end-organ dysfunction, prolongs hospital length-of-stay (LOS), delays recovery, and reduces functional capacity [24, 25, 26]. The most consistent defining feature of early right HF is the need for RVAD support. In clinical trials, early RVAD support has been steady: HMII BTT trial, 6%; [27] ADVANCE trial of the HeartWare LVAD (HVAD), 2.1%; [28] MOMENTUM 3 trial of the HM3, 4.1% [9]. Registry data have shown a similar prevalence of RVAD use in patients with continuous flow LVADs: INTERMACS, 4.1%; [8] EUROMACS 2017–2020 cohort (European Registry for Patients with Mechanical Circulatory Support), 5.4%; [29] IMACS (ISHLT Mechanically Assisted Circulatory Support registry), 6.1%; [30] and MOMENTUM 3 continued access protocol (CAP) registry, 7.6% [9].

However, the reported prevalence of all early right HF events has been plagued by variable definitions. In the HMII BTT trial, early right HF was defined as the need for RVAD or inotrope support for at least 14 days following LVAD implant [27]. Using this definition, 13% had early right HF. In the HMII DT trial, early right HF was identified in 17.5%, with 43% of those with early right HF dying within 30 days of LVAD surgery [31]. In the ADVANCE trial, 14.3% were diagnosed with early right HF [28]. The prevalence of early right HF was 25% in a meta-analysis of 36 studies (4428 LVADs), however differences were noted in study design, right HF definition, and proportion of continuous flow devices [32]. In the INTERMACS database, the prevalence of right HF 1 month after LVAD implant was 24% [33]. Notably, right HF resolved in 96.5% of these individuals by 12 months. Similarly, in EUROMACS, the prevalence of early right HF was 21.7% [34].

Peri-operative factors contribute significantly to early right HF. Most LVADs are implanted with cardiopulmonary bypass (CPB) support. While CPB maintains adequate organ perfusion and gas exchange, blood contact with the circuit provokes an inflammatory response that leads to increased capillary permeability, vasoplegia, and acute organ dysfunction [35, 36]. The large volume of priming solution administered upon CPB initiation may cause volume overload and RV dysfunction [37]. Blood loss, platelet dysfunction and coagulopathy often mandate transfusion, which can also contribute to right HF [38]. Finally, myocardial stunning [39], pericardiotomy-associated changes in RV contraction [40, 41], pulmonary hypertension [42], and inadvertent air embolism to the right coronary artery [43] may all contribute to acute RV dysfunction.

2.1.4 Late right heart failure

Even more than early right HF, analysis of late right HF has been plagued by variable definitions, different clinical parameters (e.g., time from surgery, type of support, presence of HF symptoms) and study types (e.g., single-center, clinical trials) that may not be representative of the general LVAD population [44]. Study duration is also critical: 59% of right HF was diagnosed >1 year after LVAD implant in the HMII DT trial [31], and de novo late right HF in the STS-INTERMACS database developed at a relatively constant rate of 5–10% [33, 45].

In the HMII BTT trial, late right HF was defined as initiation of inotropes >14 days after implant. Using this simplistic definition, 7% had late RV failure [27]. However, the median duration on LVAD support was only 126 days [46]. In the HMII DT trial, RV failure occurred in 21% over a median follow-up of 1.7 years at a rate of 0.13 events per patient-year (EPPY) [47]. When divided into early and late right HF (causing hospitalization >30 days post-LVAD), late right HF was identified in 8% of DT patients at a median of 480 days after LVAD implant [31].

In the ADVANCE trial, inotropes were used beyond 30 days in 6% (0.12 EPPY) [28]. Longer follow-up from the HVAD registry found RV failure in 9% (0.10 EPPY), though events were not adjudicated as early or late [48]. Similarly, the MOMENTUM 3 trial did not split early from late right HF events, simply defining RV failure as “symptoms and signs” of RV dysfunction with either RVAD implant, or therapy with inhaled nitric oxide or inotropes for >1 week at any point after LVAD surgery [49]. In MOMENTUM 3-CAP, right HF (early and late) was identified in ~37% (0.27 EPPY) [9].

In the National Readmission Database, 4.2% of all patients discharged after the implant hospitalization were readmitted with recurrent HF within 30 days of discharge (13.4% of all readmissions) [50]. When using the 2014 INTERMACS definition (Table 1) in patients who survived 3 months after LVAD surgery, the incidence of new, mild RV failure was 5–6% at 12-months, with moderate HF in an additional 4–5%, and severe HF being very rare as a late presentation (≤0.2%). In a single center study of DT patients who survived 1-year post-LVAD surgery without right HF, 45% developed right HF at a steady rate over a mean of 3.5 years [51]. Importantly, this incidence of de novo right HF, while highest in the early post-op period, appears to stabilize at 5–10% by 3–6 months post-LVAD implant and remains steady for at least 4–5 years [33, 45, 51, 52].

The prevalence of right HF after LVAD implant is ~10% by 3 months and remains constant for ≥3 years [33]. After diagnosis of late right HF, 9–20% will die and an additional ~25–33% will have persistent HF within 3–6 months [33, 45]. Two factors seem to predict persistence of RV failure. First is the time from LVAD implant to diagnosis of right HF, with HF that develops later associated with a higher rate of persistence [33]. Second is the severity of HF at diagnosis: of patients with no right HF 3 months post-LVAD, only 3.4% developed HF at 6- or 12-months after surgery. By contrast, of those with moderate right HF 3 months after implant, HF persisted in 32.5% and 11.5% at 6- and 12-months, respectively [45].

2.1.5 Outcomes associated with right heart failure

Right HF is a morbid complication in LVAD patients; this includes increased rehospitalizations, excess complications, poorer functional metrics and, critically, worse survival. Right HF has been adjudicated as the cause of death in 11–13% in the STS-INTERMACS [8] and IMACS registries [30]. Right HF as the cause of death in clinical trials has been more variable: 5% for continuous flow devices in the HMII DT trial [14], 12% in the ROADMAP trial (also HMII) [53], 17% in the ADVANCE trial (HVAD) [28], and 28% in the MOMENTUM 3 trial (HM3) [7].

In the HMII BTT trial, early RV failure was associated with a lower combined end point of (1) survival to OHT, (2) recovery, or (3) continuing support at 180 days (71% vs. 89% without right HF; p < 0.001); those requiring an RVAD had the poorest outcomes [27]. In a single high-volume center, 6-month mortality with early RVAD use was 41% [54]. This study also showed that (1) successful RVAD weaning was associated with ~3-fold better survival; and (2) planned biventricular support during the index surgery yielded better outcomes than later RVAD implant. A meta-analysis of retrospective studies found that RVAD use after LVAD was associated with significantly worse survival, and increased rates of bleeding and stroke [55]. Finally, data from INTERMACS [24] and EUROMACS [56] show that RVAD use is associated with lower 1-month, 6-month and 1-year survival.

Late right HF is also associated with reduced survival [31, 45, 57]. In the HMII DT trial, patients with late right HF had lower 1-year (78% vs. 84%), 2-year (58% vs. 81%) and 3-year survival (36% vs. 56%, p < 0.001) [31]. Notably, when analyzed from the time of diagnosis of de novo right HF, 1-year survival in this DT cohort was only 38%. BTT patients have reduced survival to OHT if right HF develops. In STS-INTERMACS, the presence and severity of late right HF predicted worse outcomes, including mortality (Figure 6) [45]. Perhaps not surprisingly, those with persistent right HF have the worst outcomes [33]. Finally, late right HF was associated with modest but significantly more strokes, arrhythmias and infections [45]. However, a causal link to right HF has not been established.

Figure 6.

Clinical outcome at 12-months in patients who survived to 3-months after LVAD implant. Patients are grouped by right heart failure (RHF) status 3-months post-LVAD. From: STS-INTERMACS database [45].

Post-LVAD right HF may also put patients at elevated risk after OHT. Patients requiring an RVAD with a BTT LVAD had a 22% increase risk of death post-OHT [58]. A retrospective, large, single-center study showed reduced post-OHT survival up to 5-years in BTT LVAD patients who developed right HF [57]. A study of 2 large European transplant centers found a post-OHT 1-year survival of 75% in BTT LVAD patients with right HF; [59] while not significant in their cohort, this is substantially lower than the ~93% 1-year post-OHT survival in the International Thoracic Organ Transplant Registry [60]. The mechanism for this risk is not clear. BTT LVAD use increases the risk for primary graft dysfunction (PGD) [61], and 2 single center analyses found an increased risk for PGD in BTT patients with pre-OHT right HF [57, 62].

Finally, right HF after LVAD has a negative impact on functional capacity and possibly QOL. Early RVAD use is associated with poorer QOL in some patients [63]. In the HMII DT trial, those with late right HF had lower QOL as assessed by the Kansas City Cardiomyopathy Questionnaire [31]. This was not true of patients in STS-INTERMACS using a visual acuity scale; [45] these investigators noted ample missing data that could have biased the analysis, and a lack of agreement as to the optimal tool for assessing QOL in LVAD patients. Functionally, those with right HF have a reduced 6-minute walk distance, supporting a detrimental effect of recurrent HF in LVAD patients [31, 45, 64, 65].

2.1.6 Predicting right heart failure after LVAD surgery

Numerous attempts have been made to predict post-LVAD RV failure in order to better guide patient selection and improve post-operative outcomes. Close to 100 variables have been found to be associated with post-LVAD right HF across dozens of studies [11, 66]. A handful of these risk factors have been consistently identified in multiple studies (Table 3). Although some variables are not actionable, others can be mitigated. While these predictors generally have high specificity, their sensitivity and negative predictive value are low, limiting their utility in clinical practice. A meta-analysis found that no single parameter was sufficiently sensitive to predict post-LVAD right HF [32].

Clinical risk factors	Lower heart rate Female sex Lower body surface area Non-ischemic etiology for heart failure HeartWare LVAD use LVAD implant as destination therapy Presence of pulmonary vascular disease Prior coronary artery bypass or valve surgery
Laboratory risk factors	Elevated white blood cell count Anemia (hemoglobin ≤10 g/dL), thrombocytopenia Abnormal renal function (elevated BUN, Cr) Abnormal liver function (ALT, AST, total bilirubin) Elevated INR
Echocardiographic risk factors	Qualitatively severe RV dysfunction Reduced RV free wall longitudinal strain Increased RV end diastolic diameter Lower LV end diastolic diameter Increased ratio of RV to LV diastolic area Higher LV ejection fraction Moderate/severe tricuspid regurgitation Increased left atrial volume
Hemodynamic risk factors	Low systolic blood pressure Elevated central venous pressure (CVP) High CVP/pulmonary capillary wedge pressure ratio Low pulmonary artery pulsatility index Low RV stroke work index Elevated pulmonary vascular resistance Cardiac index ≤2.2 L/min/m^2 Pre-operative need for inotropes/MCS/IABP
Perioperative risk factors	Need for mechanical ventilatory support INTERMACS profile Hemodialysis or ultrafiltration with 48 hours of LVAD Circulatory support (e.g., ECMO, percutaneous VAD) Inotrope use Vasopressor use Intraoperative bleeding or need for re-operation Prolonged cardiopulmonary bypass time Other concomitant procedure performed at the time of LVAD surgery

Table 3.

Risk factors for post-LVAD right heart failure.

BUN – blood urea nitrogen; Cr – creatinine; ALT – alanine aminotransferase; AST – aspartate aminotransferase; INR – international normalized ratio; LV – left ventricle; RV – right ventricle; MCS – mechanical circulatory support; IABP – intraaortic balloon pump; ECMO – extracorporeal membrane oxygenation.

Scoring systems have been developed that use a combination of these risk factors to predict post-LVAD right HF. More than 20 such models exist, most from single-center cohorts [11]. Fewer than 40% have been validated in ≥2 external cohorts. The validation studies are fraught with bias and have consistently shown poor discriminatory power with C-statistics of only 0.53–0.65 [67]. Modeling has been hindered by the many variable definitions of right HF. Further, nearly all models were derived using data from pulsatile or early-generation continuous flow LVADs, limiting their applicability. Consequently, results have been disappointing, and have limited wider use of these predictive models in clinical practice.

2.2 Valvular heart disease

2.2.1 Aortic regurgitation

Significant aortic valve insufficiency (AI) can create a short circulation loop in LVAD patients whereby a substantial fraction of the blood pumped by the LVAD to the aorta returns directly to the LV (Figure 7). This reduces effective perfusion, causes LV distention and elevates left heart filling pressures, ultimately causing HF. Consequently, moderate to severe AI is a contraindication to LVAD unless valve intervention is planned at the time of surgery [68].

Figure 7.

Aortic valve insufficiency (AI) with an LVAD. Significant AI creates a short circulation loop whereby a substantial portion of LVAD flow regurgitates back into the LV. This reduces functional cardiac output and increases left heart filling pressures, ultimately causing heart failure.

Among patients in the STS-INTERMACS database implanted between 2016 and 2020, only 0.7% had severe AI at the time of LVAD surgery [8]. Moderate or severe AI was present at implant in 4.5% of over 16,000 patients in the IMACS database [30]. By contrast, mild AI prior to LVAD is relatively common, being found in 29.7% of patients in INTERMACS [69] and 31.2% in IMACS [70].

Importantly, AI can progress or develop de novo after LVAD implant; this is clearly a result of LVAD support rather than disease progression (Figure 8A) [71]. Two mechanisms drive the development of AI on LVAD support. First, full LVAD support substantially reduces or completely eliminates aortic valve (AV) opening during systole. This promotes fusion of the commissures between valve leaflets; the resulting fibrosis causes retraction of the leaflets and AI with a central regurgitant jet. Second, decompression of the LV and the high volume delivered to the proximal ascending aorta generates a substantial and continuous pressure gradient across the valve that favors flow into the LV.

Figure 8.

Natural history of aortic insufficiency (AI) after LVAD implant. (A) Proportion of patients with progression of AI from baseline up to 4-years in those receiving an LVAD as compared to patients with end-stage HF who did not undergo LVAD surgery (NS-ESHF). Reprinted with permission [71]. (B) Progression and severity of AI in LVAD patients from the INTERMACS database, with number of patients assessed listed above each bar. Reprinted with permission [69].

The prevalence of AI increases with the duration of LVAD support (Figure 8B); in INTERMACS, 13.2% developed moderate/severe AI over a mean follow-up of 13.4 months [69]. Of those with no AI prior to LVAD, 10.7% developed moderate or severe AI, while 18.9% with mild AI prior to LVAD developed moderate or severe AI. By 6-months post-LVAD, 55% had developed at least mild AI.

Key factors that increase the risk of significant AI across multiple studies are (1) older age at LVAD implant; (2) female sex; (3) low body surface area; (4) longer duration of LVAD support; (5) baseline mild (vs. no) AI; (6) no AV opening; and (7) continuous flow (vs. pulsatile) pump [69, 71, 72, 73]. The presence of moderate or severe AI is associated with a modest but significant increase in hospitalizations [69]. Moderate/severe AI is also associated with greater severity of MR, but surprisingly, is not associated with significantly worse QOL or reduced 6-minute walk distance. Finally, the presence of moderate or severe AI (vs. no or mild AI) is associated with lower survival (49.1% vs. 35.6% at 5-years, p < 0.001) [69].

2.2.2 Mitral regurgitation

Hemodynamically significant mitral regurgitation (MR) is the commonest valve lesion at the time of LVAD implant. It is almost always functional MR secondary to LV dilation. Among >26,000 patients in STS-INTERMACS, 22.8% had severe MR at baseline [8]. Similarly, at implant, 57% had moderate/severe MR in IMACS [30], and 46% had moderate/severe MR, or underwent concomitant MV surgery in the MOMENTUM 3 trial [74].

With offloading of the LV, MR improves in the majority of LVAD patients [74]. When moderate to severe MR persists, the impact on outcomes remains uncertain. In the MOMENTUM 3 trial, persistent MR was uncommon (~6.5% at 1-year, n = 619), and was not associated with survival, adverse events including right HF, or functional capacity [74]. However, in INTERMACS (n = 8364), persistent MR was ~3-fold more common (18.8% at a median of 15 months), and was associated with increased rates of right HF and renal failure, and a modest 16% increase in mortality of borderline significance (p = 0.07) [75]. Single center studies also show mixed results, with frequent right HF but little to no impact on survival.

2.2.3 Tricuspid regurgitation

TR is also common in advanced HF. In the STS-INTERMACS database, 11.5% had severe TR at baseline [8], and 41% had moderate or severe TR at implant in the IMACS database [30]. Improvements in PVR and afterload with LVAD support often leads to a reduction in TR [23, 76]. Outcomes data on post-LVAD TR are very limited, but large single center studies as well as data from the EUROMACS registry suggest that moderate or severe TR after LVAD is associated with a small but significant increase in mortality [76, 77].

2.3 Device malfunction

Right HF can be caused by device malfunction, cannula obstruction or inappropriate LVAD speed. LVAD pump thrombosis is discussed elsewhere in this textbook. Data on device malfunction and outcomes are scant. Further, malfunction caused by manufacturing problems often leads to a recall, which abruptly changes incidence [78]. LVAD failure as a cause of death occurs in ~2% [8, 30], with events declining in recent years [8]. In a large single-center study, device malfunction occurred at a rate of 3.06 events per 1000 patient-days [79]. Notably, device malfunction is pump-type specific, with more events in the HMII. The only data for the HM3 comes from the European ELEVATE registry, which showed device malfunction in 3.9% [80]. However, almost 90% of these events were due to outflow graft twisting, a problem since corrected by the manufacturer.

Device malfunction can be grouped by the component that failed: (1) controller; (2) pump/driveline; and (3) peripheral components (e.g., cables, batteries, monitor). Importantly, not all malfunction results in right HF, with pump or driveline failure the most likely to result in a clinically significant event (i.e., HF or death). Controller malfunction was commonest (~30%), while pump or driveline malfunction constituted 13% of malfunction events [79].

Inflow cannula obstruction is a rare event, typically associated with thrombus and/or cannula malposition [81]. Abnormal inflow cannula position is associated with increased left heart filling pressures and a > 2-fold increased risk of recurrent HF [82]. By contrast, outflow cannula obstruction is more common and likely underappreciated; dozens of case reports and case series exist, with an event rate of 0.03 EPPY in the largest study [83]. The commonest pathology seems to be external compression from buildup of an acellular fibrinous material between the outflow graft and the protective GoreTex wrap that is frequently placed around the graft at implant. Notably, ~80% of patients with clinically significant outflow cannula obstruction will have recurrent HF [83].

3. Management of Recurrent Heart Failure

Recurrent HF is a morbid event that limits functional capacity and survival, and therefore warrants aggressive treatment. However, management depends on the underlying etiology (Figure 9). Importantly, assessment of post-LVAD HF should begin prior to implant, with attention given to optimization of RV function and planning for the management of significant valvular disease at the time of surgery.

Figure 9.

Algorithm for the assessment and management of recurrent heart failure (HF) in the LVAD patient. Depending on the severity of the underlying lesion, patients may be hemodynamically stable or unstable at initial evaluation. In the unstable patient, a more targeted diagnostic evaluation should be performed to rapidly identify the cause of HF. LFT = liver function testing; CBC = complete blood count; LDH = lactate dehydrogenase; CT = computed tomography; ICD = implantable cardioverter defibrillator; SBRT = stereotactic body radiation therapy. LVAD-specific causes in red; intrinsic cardiac causes in blue; non-cardiac causes in yellow.

The simplest tenet of LVAD management is ensuring an appropriate LVAD speed, the only pump parameter that can be adjusted by providers. If the speed is too low, cardiac output will remain insufficient, resulting in persistent HF. By contrast, if the speed is too high, HF can result from worsening RV dysfunction (Figure 5) and/or worsening valve disease. Though widely used and given a Class I recommendation in the MCS guidelines [68], data supporting the utility of ramp echocardiographic studies are limited. Further, recent evidence suggests that hemodynamic-guided management by right heart catheterization may be superior to echo-guided ramp testing [84]. Ultimately, more data are needed.

3.1 Management of right heart failure

Preemptive strategies to mitigate right HF post-LVAD (whether early or late) are an essential component in management. Preoperatively, this consists of optimizing RV preload (lowering CVP), and afterload (lowering PVR and left heart filling pressures). Intraoperative strategies include judicious fluid and blood product use, and limiting time on CPB. Immediately post-LVAD, RV support with inotropes and pulmonary vasodilators is routine, as is managing vasoplegia to limit myocardial ischemia, and optimizing LVAD speed [85].

Chronic management of the LVAD patient includes maintenance of proper RV preload with diuretics, treatment of hypertension to permit optimal pump function and resumption of neurohumoral blockade, which can improve functional capacity and survival [86, 87]. Collectively these approaches are likely to limit the incidence of recurrent HF, though little data exists to support this indication. Once right HF develops, management differs depending on timing: more aggressive strategies are favored early and can frequently yield good outcomes, while late right HF usually merits a more conservative approach and has a more uniformly poor prognosis.

3.1.1 Early right heart failure

Timely identification of RV failure in the immediate postoperative period may be challenging but is of critical importance. As noted, early post-operative diuretic and inotrope use are routine. However, rising lactic acid levels, evidence of end organ dysfunction, and/or persistent or increasing vasoactive medication doses signal that RV mechanical support may be required. Importantly, the timing of RVAD initiation affects prognosis. In a multicenter retrospective analysis of 91 patients requiring RV support, an RVAD was implanted at the time of the LVAD in 44% with 56% receiving an RVAD after the initial operation. Concomitant RVAD implant was associated with >2-fold lower mortality vs. RVAD implant after completion of LVAD surgery [88]. In the IMACS registry, RVAD use was associated with substantially lower survival. Further, within the RVAD group, there was progressively worse survival with longer times between LVAD and RVAD surgery (1-year survival: LVAD only 82.9%; RVAD at time of LVAD implant, 58.5%; RVAD ≤14 days after LVAD, 51.6%; RVAD 15–30 days after LVAD, 32.4%) [30]. A major limitation of these studies is a lack of propensity matching and the inability to fully control for baseline differences, identifying an urgent area for future research.

3.1.2 Late right heart failure

There is almost no substantive data guiding management of late right HF. First, it is important to identify the cause of right HF (Figure 2). LVAD speed should be optimized and arrhythmias managed as indicated [89]. Importantly, the only manifestation of ventricular tachyarrhythmias in an LVAD patient may be HF. Pulmonary hypertension usually improves with offloading of the LV, but could worsen in the setting of MR. However, most cases of late right HF are likely due to intrinsic RV dysfunction in the myopathic heart.

As with early right HF, the cornerstone of therapy is diuresis. However, this is often insufficient and many patients require initiation of inotropic therapy. In patients developing late right HF in the STS-INTERMACS database, 33–50% required inotropic support [45]. Unfortunately, inotrope use portends a very poor prognosis with substantially elevated mortality even beyond those with late right HF that do not require inotropes [33, 45]. The need for extended inotrope use also increases the risk of infection due to chronic indwelling intravenous catheter placement, and may be associated with poorer functional status and QOL, though data are lacking. As no pumps are approved for hospital discharge, RVAD support is virtually never employed in late right HF, being used in <0.2% of those in the STS-INTERMACS database [45].

3.2 Management of Valvular Disease

Valve surgery at the time of LVAD implant is common and uniformly increases morbidity relative to LVAD surgery alone. Mortality data with valve surgery at the time of LVAD are mixed. In the HMII trials, 21.9% underwent valve surgery with modestly increased mortality (1-year survival: 69% vs. 75%, p = 0.004) [90]. In the ADVANCE trial, 19.6% had concomitant valve surgery, and though the absolute difference in survival was the same as with HMII, the result was not significant (79% vs. 85%, p = 0.33) [91]. In the MOMENTUM 3 CAP registry, 21.8% had valve surgery with equivalent survival at 2-years (81.7% vs. 80.8%, p = 0.6) [92]. Registry data show a similar prevalence of valve surgery at LVAD implant (IMACS, 12.1%; [30] EUROMACS, 19.3% [93]). When those having valve surgery in EUROMACS were propensity matched to LVAD patients not undergoing valve surgery, 1-year survival was the same (67.9% vs. 66.4%, p = 0.25) [93].

Notably, early right HF was actually higher with concurrent valve surgery in the HMII [90], ADVANCE [91], and HM 3 trials [92], while the propensity matched cohort in EUROMACS had equivalent rates of RVAD use [93]. Late right HF was not different in the HMII or ADVANCE trials but was increased with concurrent valve surgery in the MOMENTUM 3 CAP registry.

3.2.1 Aortic valve disease

Hemodynamically significant AI can be addressed via 3 methods (reviewed in detail elsewhere) [94]: (1) complete closure (oversewing) of the AV; (2) AV repair (e.g., central closure via Park’s stitch [95]); or (3) AV replacement (AVR). Closure successfully eliminates AI, but acute LVAD malfunction may be rapidly fatal as this method leaves no native cardiac output. AV repair closes the central orifice of the AV while still allowing blood flow through the lateral commissures. This prevents blood stasis and thrombosis in the aortic root, and durably limits AI [94]. Of note, if AVR is pursued, mechanical valves are not recommended due to decreased valve opening and blood flow that could increase the chance of thrombosis.

Outcomes data distinguishing the best approach are extremely limited. Among those with concurrent AV and continuous flow LVAD surgery between 2006 and 2012 in INTERMACS, survival was lower with AV closure than either AV repair or AVR, suggesting that preservation of AV opening is beneficial [96]. More recent data from IMACS showed reduced survival with concurrent procedures, with AV repair having numerically better survival than AVR [70]. However, data were compared to no AV surgery, leaving unclear if the difference between repair and AVR was significant. In both studies, CPB time and LOS were longer with AV procedures.

Though mortality may be increased, residual confounding is possible due to the lack of prospective, randomized data. Whether AV procedures are of benefit in a subpopulation of LVAD patients remains unclear. Among concomitant AV procedures in IMACS, ~50% were performed in those with mild AI [70]. Interestingly, when this analysis was limited to those with moderate or severe AI, survival was the same between AV repair, AVR and those not receiving AV surgery. This suggests that the benefit may be restricted to those with more severe disease.

Finally, as noted, AI will develop and/or progress in the majority of LVAD patients (Figure 8B). Despite this, data on the optimal approach to these patients is markedly limited. Post-LVAD AI has been managed with (1) open AVR; (2) percutaneous closure of the aortic valve with an occlusion device (e.g., those used for septal defects); or (3) percutaneous AVR (TAVR). Percutaneous methods have increased over the last decade but no large-scale studies have been performed to study different approaches. In a meta-analysis of 15 case series (only 29 patients), percutaneous treatments were durable and showed no difference in mortality between occlusion or TAVR [97]. A study using the Nationwide Readmission Database found no difference in mortality between surgical AVR and TAVR but showed substantially lower morbidity with TAVR [98]. Given the prevalence of post-LVAD AI, this is an urgent area for future research.

3.2.2 Mitral valve disease

Intraoperative management of moderate or severe MR remains controversial. Among those with moderate or severe MR in INTERMACS, only 5.3% underwent MV surgery (95.8% repair, 4.2% replacement) at the time of LVAD [99]. MR severity 3-months after LVAD was equivalent in both groups (moderate/severe MR in 20% with MV procedure, 25% with LVAD alone, p = 0.2). Importantly, there was no survival benefit between: (1) those with moderate/severe MR undergoing LVAD and MV surgery vs. LVAD alone; (2) those with baseline moderate vs. severe MR; nor (3) those with baseline no/mild MR vs. moderate/severe. A trend (p = 0.09) towards benefit of concurrent MR surgery 2-years after LVAD was noted in DT patients. MV surgery was associated with longer LOS and CPB time, but fewer rehospitalizations. The incidence of right HF was the same and 6-minute walk distance was not different with or without concurrent MV surgery [99].

These data suggest there is little benefit to correcting MR at the time of LVAD and risk predictors have not been established to identify subgroups who might derive benefit. However, in INTERMACS, the presence of moderate or severe MR at least 3-months after LVAD implant was associated with a nearly 2-fold increased risk of right HF and a trend towards lower survival [75]. Data are lacking to guide management in LVAD patients with significant residual MR.

3.2.3 Tricuspid valve disease

Surgical treatment of moderate or severe TR at the time of LVAD is similarly controversial. Among those with moderate/severe TR in INTERMACS, 16.5% underwent TV surgery (>95% repair) at the time of LVAD [100]. TV surgery was associated with slightly lower survival (hazard ratio 1.13, p = 0.04) and significantly higher rates of stroke, bleeding and arrhythmia. Concurrent TV surgery did not impact patient-reported QOL [100]. In a propensity matched cohort from EUROMACS, TV surgery had no impact on survival, readmission, or right HF after LVAD [101]. Further, at 1-year, the prevalence of moderate or severe TR was similar between those with TV surgery or LVAD alone [101]. Other large single center studies have confirmed a high failure rate (~30–40%) of TV surgery in LVAD patients [102, 103]. Notably, the rate of concurrent TV surgery has declined in the last decade, possibly in response to the mounting evidence for a lack of benefit. Whether subgroups that do benefit can be identified remains to be determined.

3.3 Management of LVAD device malfunction

While failure of any part of the LVAD can be life threatening, the majority of issues with external components are readily remedied without harm [79]. By contrast, LVAD thrombosis almost always requires therapy (Figure 9). Pump or driveline failure triggering LVAD exchange occurred in 0.6% of HMII recipients in an early INTERMACS analysis [104]. Driveline malfunction (more common with the HMII than HVAD) rarely requires urgent intervention. Major driveline issues occurred in ~2.2% of >13,000 HMII patients, and of those only 20% required urgent surgical intervention (driveline repair, pump exchange, OHT) [105]. There are as yet no data on the incidence and outcomes of driveline issues with the HM3.

Lastly, LVAD cannula problems should also be considered in the differential diagnosis of recurrent HF. Although rare, malalignment of the inflow cannula may require surgical correction. Outflow cannula obstruction is predominantly caused by external compression of the outflow graft or stenosis of the aortic anastomosis [83]. While surgical correction is an option, the safety and long-term durability of outflow graft stenting has recently been confirmed and can be performed with very low morbidity [83].

4. Conclusions

Recurrent HF is a very common complication after LVAD implant and portends a poor prognosis with increased morbidity and mortality. The causes are varied and identifying the correct etiology is critically important to proper management (Figure 9). Given the highly specialized nature of many of these etiologies, it is recommended that HF in the LVAD patient be managed in an active LVAD center. A wealth of evidence exists defining the incidence and prevalence of LVAD-associated HF. However, only limited data are available to guide therapies.

In our opinion, future research to reduce morbidity associated with recurrent HF should focus on 3 major areas. (1) Preemptive strategies to prevent right HF: for instance, pre-LVAD temporary MCS usage has increased with time in the STS-INTERMACS database [8], but whether it will impact post-VAD outcomes remains to be determined. (2) Perioperative improvements and standardization of surgical methods: for instance, a meta-analysis of LVAD implant via lateral thoracotomy suggests a significantly lower incidence of post-LVAD right HF, an approach being tested in the ongoing SWIFT trial [106]. And, finally, (3) prospective assessment to identify treatment strategies that provide significant benefit for patients with recurrent HF. These could be pharmacologic strategies, or possibly development of durable RV support. Regardless, the breathtaking pace of LVAD technological development will no doubt continue to benefit patients with advanced HF.

Conflict of interest

None