Author + information
- Received May 30, 2018
- Revision received July 19, 2018
- Accepted July 24, 2018
- Published online October 29, 2018.
- Lauren K. Truby, MDa,
- A. Reshad Garan, MDa,
- Raymond C. Givens, MD, PhDa,
- Brian Wayda, MDa,
- Koji Takeda, MD, PhDb,
- Melana Yuzefpolskaya, MDa,
- Paolo C. Colombo, MDa,
- Yoshifumi Naka, MD, PhDb,
- Hiroo Takayama, MD, PhDb and
- Veli K. Topkara, MD, MSca,∗ ()
- aDivision of Cardiology, Department of Medicine, Columbia University College of Physicians and Surgeons, New York, New York
- bDivision of Cardiothoracic Surgery, Department of Surgery, Columbia University College of Physicians and Surgeons, New York, New York
- ↵∗Address for correspondence:
Dr. Veli K. Topkara, Center for Advanced Cardiac Care, Columbia University Medical Center-New York Presbyterian, 622 West 168th Street, PH10-203A, New York, New York 10032.
Objectives This study sought to evaluate the impact of moderate to severe aortic insufficiency (AI) on outcomes in patients with continuous flow left ventricular assist devices (CF-LVADs).
Background Development of worsening AI is a common complication of prolonged CF-LVAD support and portends poor prognosis in single-center studies. Predictors of worsening AI and its impact on clinical outcomes have not been examined in a large cohort.
Methods We conducted a retrospective analysis of patients with CF-LVAD in the INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) study. Development of significant AI was defined as the first instance of at least moderate AI. Primary outcomes of interest were survival after development of significant AI and time to adverse events, including device complications and rehospitalizations.
Results Among 10,603 eligible patients, 1,399 patients on CF-LVAD support developed moderate to severe AI. Prevalence of significant AI progressively increased over time. Predictors of worsening AI included older age, female sex, smaller body mass index, mild pre-implantation AI, and destination therapy strategy. Moderate to severe AI was associated with significantly higher left ventricular end-diastolic diameter, reduced cardiac output, and higher levels of brain natriuretic peptide. Significant AI was associated with higher rates of rehospitalization (32.1% vs. 26.6%, respectively, at 2 years; p = 0.015) and mortality (77.2% vs. 71.4%, respectively, at 2 years; p = 0.005), conditional upon survival to 1 year.
Conclusions Development of moderate to severe AI has a negative impact on hemodynamics, hospitalizations, and survival on CF-LVAD support. Pre- and post-implantation management strategies should be developed to prevent and treat this complication.
The use of continuous flow left ventricular assist devices (CF-LVADs) is rapidly expanding in patients with advanced heart failure, both as a bridge to transplantation and as destination therapy (1). Development of aortic insufficiency during CF-LVAD support has been well documented, affecting 15% to 52% of patients after 1 year of support (2–5). Mechanisms proposed for the cause of worsening AI during CF-LVAD use include leaflet deterioration and/or commissural fusion, aortic sinus dilation, and increased transvalvular gradients, all of which may have implications for device function, thromboembolism, and device explantation for ventricular recovery (6–8). Anticipating that those patients with underlying valvular pathology will have more rapidly progressive aortic valve disease on CF-LVAD support, current consensus guidelines recommend that moderate or greater AI at the time of implantation be treated surgically (Class I, Level of Evidence: C) (9). Concomitant surgical repair can be achieved by over-sewing the atrioventricular (AV) structure, using Park’s stitch or modified Park’s stitch, closure of the ventriculoaortic junction with a surgical patch, or valve replacement using a bioprosthesis (10–12).
In the post-implantation period, transthoracic echocardiography is usually performed every 6 to 12 months to assess frequency of AV opening, severity of AI, and AV structure. In conjunction with surveillance, multiple device management strategies have been proposed to promote AV opening and prevent AI development or progression through optimization of CF-LVAD parameters under transthoracic echocardiography guidance (9). To date, single-center studies have been unable to demonstrate an effect of post-implantation AI on short- or long-term survival on CF-LVAD support, leading some to question whether methods to optimize AV opening and device parameters are clinically indicated. Moreover, the impact of AI on secondary clinical outcomes on LVAD support, such as worsening heart failure, hospital readmission, and device complications, remain largely unknown. In the current study, we used the INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) registry to: 1) characterize prevalence and the natural history of AI during CF-LVAD support; 2) identify predictors of the development of moderate to severe AI on device support; and 3) assess the impact of significant AI on worsening heart failure, adverse events, quality of life, and mortality on CF-LVAD support.
Study design, variables, and definitions
The INTERMACS registry was queried to identify patients who received a durable CF-LVAD between 2006 and 2016. Excluded patients were those without a pre-implantation assessment of AI, those without follow-up echocardiograms, and those who received a pre-implantation or concomitant AV procedure (Figure 1). Echocardiographic assessment of AI was available at serial time points from 1 week post-implantation to 8 years post-implantation; AI at these time points was graded as none, mild, moderate, or severe. Significant AI was defined as the first instance of moderate or severe AI during the follow-up period in patients with no or mild pre-implantation AI.
Descriptive analyses were conducted for all baseline variables and are presented as mean ± SD for continuous variables and as numbers (percentages) for categorical variables. Non-normally distributed variables are presented as median and interquartile range. Differences between those who would go on to develop significant AI and those who did not were assessed by using Student’s t-test and Kruskal-Wallis tests. Freedom from significant AI was assessed using Kaplan-Meier survival estimates with log-rank tests for comparison among subgroups. Univariate and multivariate Cox proportional hazard regression analysis was used to identify predictors of significant AI among patients with no or mild pre-implantation AI. Proportional hazards assumptions were tested by visual assessments of Kaplan-Meier estimates. All variables in the final model were tested for interactions. The impact of AI on New York Heart Association (NYHA) functional class, left ventricular ejection fraction (LVEF), and LV end diastolic diameter, and quality of life were assessed at serial time points during the study period and were compared between patients with and without moderate to severe AI as assessed at that time point. All p values were reported as two-sided tests with p value <0.05 considered statistically significant. The effect of AI on survival and adverse events was assessed using Kaplan-Meier survival methods conditional upon survival to 1 year, with comparisons between those patients with and without a diagnosis of significant AI in the first year of support. STATA version 13.1 software (Stata Corp., College Station, Texas) was used to perform statistical analysis.
Baseline clinical characteristics based on pre-implantation aortic insufficiency
A total of 10,925 patients were identified as eligible for analysis in the current study. Among them, 10,603 patients had no AI (n = 7,362, 69.4%) or mild AI (n = 3,241, 30.6%) at the time of device implantation. A total of 322 patients had moderate or severe pre-implantation AI and did not undergo a concomitant AV procedure during the index operation (Figure 1). A total of 2,296 patients died during the follow-up period, and 2,671 patients underwent transplantation with a mean follow-up of 13.4 months.
Baseline demographics and medical histories of patients with no or mild pre-implantation AI were compared between those who did and did not develop moderate to severe AI (Table 1). Patients who developed moderate to severe AI on CF-LVAD support were older and more likely to be female. Development of moderate to severe AI was more prevalent in those with a lower body surface area, those with ischemic cardiomyopathy, and those with peripheral vascular disease. Patients who developed worsening AI were twice as likely to have mild AI as opposed to no AI at the time of CF-LVAD implantation. Pulmonary hypertension, peripheral vascular disease, and chronic kidney disease were more common in patients with moderate to severe AI. Significant AI was more common in patients who received a CF-LVAD as destination therapy, with moderate to severe AI patients spending more time on device support. Overall, 80% of patients had NYHA functional class IV heart failure symptoms at the time of CF-LVAD implantation, and 16% of patients were INTERMACS profile 1. Eighty-three percent of the devices used were axial, and 96% were used in an isolated LVAD configuration. In terms of device strategy, 57.1% of patients received implants (bridge-to-transplantation [BTT] patients), whereas 42.2% were candidates only for destination therapy.
Pre-implantation laboratory values and hemodynamics are summarized in Table 2. Patients who developed significant AI had higher levels of blood urea nitrogen and creatinine and lower levels of albumin before implantation. In addition to having a higher concentration of B-type natriuretic peptide (BNP), patients with moderate to severe AI had larger left ventricular end-diastolic diameters (LVEDDs), lower blood pressures, and lower cardiac output values at baseline.
Natural history of worsening aortic insufficiency on CF-LVAD support
Among patients with no or mild pre-implantation AI, a total of 31,571 echocardiograms were performed during the follow-up period. The distribution of AI at each time point during the first 2 post-operative years is displayed in Figure 2A. Among patients with no or mild pre-implantation AI and follow-up echocardiographic data, the proportion of patients with mild AI increased dramatically during the study period, such that by 6 months’ follow-up 55% of patients with echocardiograms had at least mild AI. Similarly, the proportion of patients with moderate AI increased from 1% at 1 week follow-up to 10% at 1 year and 14% at 2 years. Kaplan-Meier estimates of freedom from moderate or severe AI are shown in Figure 2B.
Clinical predictors of worsening aortic insufficiency on CF-LVAD support
Clinically relevant patient and device characteristics were entered as univariate and multivariate Cox proportional hazard regression analyses to determine predictors of progression to moderate or severe AI among patients with no or mild pre-implantation AI (Table 3). Multivariate analysis identified age of >60 years (hazard ratio [HR]: 1.75; confidence interval [CI]: 1.50 to 2.04; p < 0.001), female sex (HR: 1.29; 95% CI: 1.09 to 1.52; p = 0.002), body surface area <2.0 m2 (HR: 1.30; 95% CI: 1.13 to 1.49; p < 0.001), and mild pre-implantation AI (HR: 1.87; 95% CI: 1.64 to 2.13; p < 0.001) as significant predictors of worsening AI on CF-LVAD support.
Impact of aortic insufficiency on ventricular remodeling during CF-LVAD support
To try to quantify the impact of significant AI on ventricular structure and function, we analyzed serial echocardiograms and invasive hemodynamic testing of patients during the study period. As displayed in Figures 3A and 3B, LVEDD and pro-BNP was higher at multiple time points during the study compared between those with no or mild AI and moderate to severe AI at a given time point. In addition, the percentage of patients with at least moderate mitral regurgitation was significantly higher in patients with moderate to severe AI at all time points assessed in the first 2 years of support (Figure 3C). This translated into lower systolic blood pressures (Figure 3D) and lower cardiac output (Figure 3E). There were no significant differences in mean right atrial pressure, pulmonary vascular resistance, or pulmonary capillary wedge pressures between the 2 groups at serial time points (Online Figure S1). Six-min walk distances and Kansas City Cardiomyopathy-12 Questionnaire scores were lower in those with moderate to severe AI, although not to a significant degree (Online Figure S2).
Impact of aortic insufficiency on survival on CF-LVAD support
In the overall cohort, on-device survival at 5 years was 46.9%. The impact of moderate to severe AI on survival on CF-LVAD support was first evaluated using episode splitting. In this approach, patients with significant AI remained in the control arm until the time their condition was first diagnosed with moderate to severe AI and were then switched to the significant AI arm. Using this approach, survival on CF-LVAD support was significantly lower in patients with significant AI than in those with no or mild AI (49.1% vs. 36.5%, respectively, at 5 years; p < 0.001), even after adjusting for age at implantation, INTERMACS profile, and chronic kidney disease (Figure 4A). As an alternative approach, we analyzed freedom from death conditional on survival to 1 year on CF-LVAD support based on the presence or absence of moderate to severe AI within the first year of support. Even when adjusted for covariates, conditional survival analysis also suggested a significant difference in survival (77.2% vs. 71.4%, respectively, at 2 years; p = 0.005) (Figure 4B). These trends persisted in a sensitivity analysis of destination therapy patients (Online Figure S3A).
Impact of worsening AI on device complications and rehospitalizations
Freedom from rehospitalization, device malfunction, arrhythmia, and stroke by presence or absence of moderate to severe AI within the first year of CF-LVAD support, conditional on survival to 1 year, is shown in Figure 5. As shown, patients who developed significant AI within the first year of CF-LVAD support had significantly lower freedom of rehospitalization (32.1% vs. 26.6%, respectively, at 2 years; p = 0.015) (Figure 5A). No significant differences were observed among rates of bleeding, arrhythmia, and stroke (Figures 5B to 5D).
The current study investigated the incidence and impact of moderate to severe AI during CF-LVAD support. Important findings include: 1) AI as a progressive disease that develops during CF-LVAD support with well over 50% developing mild disease at 2 years of support and 15% developing moderate to severe disease; 2) old age, female sex, small body size, and presence of mild AI at the time of CF-LVAD implantation predict development of moderate to severe AI on CF-LVAD support; 3) development of moderate to severe AI is associated with adverse left ventricular remodeling on CF-LVAD support; and 4) our data suggest that CF-LVAD support patients with moderate to severe AI are at higher risk for rehospitalizations and mortality after 1 year of pump support.
Multiple prior studies have addressed the natural history of AI during CF-LVAD support. In a single-center study from 2005 to 2013, Holley et al. (5) identified 210 patients with CF-LVAD among whom 32 patients (15.2%) developed moderate to severe AI in a median time of 482 days. At 5 years after implantation, 35% of patients had developed de novo AI (5). Similarly, Cowger et al. (9,14) analyzed 166 patients over 291-person years and discovered 36 patients who developed moderate to severe AI (0.17 persons per year). In a meta-analysis of 7 observational studies including 657 patients, the rate of AI development was 4% per month of support (13). In the current study, we report a total of 1,399 patients (13.2%) who developed significant AI in the current study. When all echocardiograms available at each time point were combined, we demonstrated a progressive increase in the number of patients with mild, moderate, or severe AI while on CF-LVAD support. Thus, our study supports the observations of previously published, small, single-center studies which suggest that AI on CF-LVAD support is a progressive disease.
In those patients with no or mild AI at the time of implantation, our study confirms the findings of multiple previous studies that have suggested that older age, female sex, and smaller body size are risk factors for progression to moderate or severe disease (13,14). Although the current study focused only on CF-LVADs, previous studies that have included both continuous and pulsatile devices have identified CF-LVADs as conferring higher relative risk for AI (15). Similarly, although not available in the INTERMACS data set, much attention has been paid to the opening status of the AV and its impact on AI (2,14). In addition to the risk factors which were confirmed in the present study, we identified mild AI and elevated BNP at the time of implantation as independent predictors of progression to moderate or severe AI. It is likely that structural and anatomic factors (e.g., body mass index) as well as factors that contribute to prolonged CF-LVAD support (e.g., destination therapy indication) contribute to increased risk of progression of AV disease.
To assess the impact of AI on heart failure, both structurally and functionally, we analyzed the impact of AI development on LVEDD, LVEF, pulmonary capillary wedge pressure, central venous pressure, and cardiac output. We found that a comparison of LVEDD among patients who would go on to develop moderate to severe AI showed LVEDD was higher in those patients who had AI already develop. This trend persisted when LVEF was assessed. This translated into lower systolic blood pressures and lower cardiac output. Recently, Sayer et al. (16) compared invasive hemodynamics and echocardiographic assessment of AI on CF-LVAD support and reported similar elevations in central venous pressure and pulmonary capillary wedge pressure among patients with AI (16). Overall, it does appear that AI has both hemodynamic consequences that must be managed aggressively, whether medically or surgically, before and during CF-LVAD support in order to prevent worsening clinical status. Additional studies assessing the impact of AI on LV recovery are warranted.
The effect of AI on device adverse events was also assessed in the current study by using AI status and conditional survival to 1 year to assess freedom from bleeding, arrhythmia, stroke, and rehospitalization. We demonstrate that after 1 year of support, patients with moderate to severe AI experienced decreased freedom from rehospitalizations. Taken together with the data regarding LV size, moderate mitral regurgitation, blood pressure, and cardiac output, it is conceivable that the changes in LV structure and function may predispose patients to rehospitalization for worsening heart failure symptoms.
Finally, the effect of moderate or severe AI on survival was assessed. All prior single-center studies have failed to demonstrate a difference in survival based upon the development of AI (5,14,17). Rather than stratifying patients by development of AI and assessing their survival from the time of implantation, we chose to compare survival before and after development of significant AI as well as conditional upon 1 year’s survival. In this way, we demonstrate decreased survival in patients after the development of AI when censored for transplantation or device exchange as well as decreased survival after 1 year of support after the development of AI. Although the reason for death is not directly analyzed in this study, we hypothesize that, in part, worsening heart failure secondary to worsening AI may play a role.
Given the significant impact of moderate to severe AI on clinical endpoints in patients with CF-LVAD, prevention and management strategies need to be developed for patients with this condition. In particular, the dilemma of mild pre-implantation AI must be addressed. Our analysis suggests that patients with mild AI at the time of implantation have both an increased risk of AV disease progression and a shorter time to development of significant AI compared to those without pre-existing AI (Online Figure S3B). Given its associated risks, consideration may be given to concomitant AV repair high-risk patients, particularly those with expected prolonged time on device support, that is, destination therapy patients. At our institution, our current practice is to concomitantly repair mild AI by using Park’s stitch in patients who receive implants for indications of deceleration time. In BTT patients, decision to repair mild AI is made on a case-by-case basis based on likelihood of prolonged support such as high-level human leukocyte antigen sensitization and becoming a destination therapy patient after implantation due to worsening renal function, aging, and so forth. We avoid repairing AI in patients who have possibility of myocardial recovery and device explantation, particularly patients with young ages, those with nonischemic causes (including myocarditis), and shorter duration of heart failure. In addition, early interventions such as speed adjustment echocardiograms allowing for aortic valve opening whenever feasible may potentially reduce risk of worsening AI in high-risk individuals. Transplantation candidates who are severely symptomatic from AI on CF-LVAD support should be considered for status upgrade. Those who are ineligible for transplantation may potentially benefit from rapidly evolving percutaneous therapies such as transcatheter aortic valve replacement and occluder devices. However, safety, durability, and efficacy of these approaches require further investigation.
First, because the data were collected from a large national registry, they are subject to error in entry as well as missing data. Second, all patients did not have echocardiographic data at all time points during the study. In addition, echocardiographic data were limited, and we were thus not able to assess aortic valve and aortic root dilation, device settings, and other important details for each patient at each time point. We also recognize that there is significant interobserver variability both within and among centers in terms of interpreting echocardiograms. Because of the large numbers of patients included in the current registry-based study, many comparisons between groups have reached statistical significance. Caution is advised, however, to acknowledge these differences only when clinically relevant. Importantly, we recognize that AI is a time-dependent phenomenon and that many analyses are subject to influence by time on device support.
This study demonstrates that AI after CF-LVAD implantation is a progressive disease that contributes to worsening heart failure, increased rehospitalizations, and decreased survival. Those patients who are older, with a smaller body size, and with a CF-LVAD placed for destination therapy are at higher risk of AV disease progression, particularly if they have mild AI before CF-LVAD implantation. In this patient population, AV intervention at the time of CF-LVAD implantation may be warranted.
COMPETENCY IN MEDICAL KNOWLEDGE: The current study highlights worsening AI as a common, progressive disease of CF-LVAD support with significant subclinical and clinical implications. AI is a time-dependent phenomenon and thus disproportionately affects patients receiving CF-LVADs support as destination therapy. Patients with mild AI prior to implantation, those in whom concomitant repair is not currently recommended, are at higher risk. Future studies should identify high-risk patients and evaluate the efficacy of concomitant repair of mild AI in this particular subpopulation. Development of AI may limit survival, and indications of aortic valve repair on CF-LVAD support may expand as percutaneous technologies continue to evolve.
TRANSLATIONAL OUTLOOK: Translation of this research to the care of the individual patient may help cardiologists and surgeons identify those patients at high risk for the development of worsening AI during device support and help to facilitate a discussion of the risks and benefits of concomitant aortic valve repair. In addition, on a more global scale, the results of the study suggest that a prospective, randomized controlled trial of concomitant aortic valve repair in those with mild pre-implantation disease is warranted, particularly in the destination therapy population.
The authors thank the INTERMACS investigators, coordinators, and participating institutions for the data they provided for this registry.
Supported by National Center for Advancing Translational Sciences, National Institutes of Health grants UL1TR001873 (to Dr. Topkara) and KL2TR001874 (to Dr. Garan), and Lisa and Mark Schwartz, and Program to Reverse Heart Failure, New York Presbyterian Hospital-Columbia University. Dr. Colombo has consulted for and received research support from Abbott. Dr. Naka has consulted for Thoratec and Heartware. All other authors have reported that they have no industrial relationships relevant to the contents of this paper to disclose. The content is solely the responsibility of the authors and does not necessarily represent the official views of the INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) or National Institutes of Health.
- Abbreviations and Acronyms
- aortic insufficiency
- aortic valve
- bridge to transplantation
- continuous flow left ventricular assist device
- left ventricular end diastolic diameter
- left ventricular ejection fraction
- pulmonary vascular resistance
- Received May 30, 2018.
- Revision received July 19, 2018.
- Accepted July 24, 2018.
- 2018 American College of Cardiology Foundation
- Jorde U.P.,
- Uriel N.,
- Nahumi N.,
- et al.
- Holley C.T.,
- Fitzpatrick M.,
- Roy S.S.,
- Alraies M.C.,
- Cogswell R.,
- Souslian L.,
- Eckman P.,
- John R.
- Cowger J.,
- Rao V.,
- Massey T.,
- Sun B.,
- May-Newman K.,
- Jorde U.,
- Estep J.D.
- Cowger J.,
- Pagani F.D.,
- Haft J.W.,
- Romano M.A.,
- Aaronson K.D.,
- Kolias T.J.
- Sayer G.,
- Sarswat N.,
- Kim G.H.,
- et al.