Authors: Jaiel Niamat, Faiz Ramjankhan, Niels Van Der Kaaij, Monica Gianoli, Linda W Van Laake, Mostafa M Mokhles
Categories: Tx & Mcs, Heart failure, HeartMate, HeartWare ventricular assist device, Left ventricular assist device, Left ventricular assist device malfunction, Eacts/118, Eacts/116, AcademicSubjects/MED00920
Source: European Journal of Cardio-Thoracic Surgery
Left ventricular assist device (LVAD) therapy has evolved from a short-term bridge-to-transplant strategy into a long-term and often chronic therapy due to long waiting times for heart transplantation and application as destination therapy. Consequently, patients are at risk of developing complications necessitating LVAD exchange. The aim of this study is to assess patient outcomes after LVAD exchange.
Patients who underwent LVAD exchange between January 2010 and December 2022 were included. Logistic and cox regression analyses were used to identify potential risk factors for short and long-term adverse events, respectively. Survival after exchange was assessed using Kaplan–Meier estimates.
Sixty-one patients underwent a total of 80 LVAD exchanges. Most frequently observed short-term complications were pulmonary infections (16.3%) and right heart failure (16.3%). Exit-site infections (34.7%) and device malfunctions (25.3%) were the most often observed long-term complications. HeartWare ventricular assist device as index device was associated with a higher risk of right heart failure [hazard ratio 6.42, 95% confidence interval (CI) 1.80–22.90] and respiratory failure (hazard ratio 7.81, 95% CI 1.95–31.23) compared to HeartMate II and HeartMate 3. Survival was 83% (95% CI 75.5–95.3%) at 1 year and 67% (95% CI 53.9–84.7%) at 6 years after exchange. After 5 years, 25.0% was transplanted, 23.8% had undergone a re-exchange and 32.5% was alive without new intervention.
Although LVAD exchange can be performed with a relatively low mortality, other post-operative adverse events are common. Patients with the HeartWare ventricular assist device as index device may be at higher risk of developing right heart failure and respiratory failure after exchange.
Keywords: Left ventricular assist device, HeartWare ventricular assist device, HeartMate, Heart failure, Left ventricular assist device malfunction
During the past decade, the implantation of the left ventricular assist device (LVAD) has become the most common mode of treatment for patients with advanced heart failure, both as bridge-to-transplant and destination therapy [1, 2]. When elected for this treatment modality, patients are currently provided with centrifugal continuous flow devices that support the left ventricle [3, 4]. In the past, pulsatile flow and axial continuous flow devices were more commonly implanted [5].
Nowadays, patients often require prolonged periods of LVAD support duration due to increased waiting times for heart transplantation and approval of this treatment modality for destination therapy in patients who are no candidate for transplantation [6]. Consequently, patients are at risk of developing complications associated with long-term LVAD support despite an improved prognosis due to technological advancement and increased knowledge [7]. Examples include pump thrombosis, device infections, technical pump failure, gastrointestinal bleeding and late right heart failure [8–10]. Some of these complications may warrant LVAD exchange [8, 11–15]. However, device exchange is associated with a high rate of complications [10, 16, 17]. Yet, limited studies have been conducted assessing survival and risk factors for complications after LVAD exchange.
Therefore, the aim of this study was to (1) review patient characteristics and survival of patients undergoing LVAD exchange, (2) document the rate of adverse events peri- and post-operatively and (3) investigate potential risk factors for major adverse events.
We conducted a retrospective cohort study in which LVAD patients >18 years of age were included that required a pump exchange between January 2010 and December 2022 at the University Medical Center Utrecht, Utrecht, Netherlands. The study was approved by the Medical Ethics Committee (UMC Utrecht, Utrecht, Netherlands) and requirement to obtain individual informed consent to perform this retrospective analysis was waived (METC: 22U-0299).
Patients in which initial LVAD implantation failed due to complications were assessed for either medical therapy, an LVAD exchange or urgent heart transplantation. Eligibility for either of these therapies was assessed by a multidisciplinary team, consisting of (heart failure) cardiologists, cardiac surgeons, anaesthesiologists and intensivists and additional experts if indicated (e.g. geriatrician or pulmonologist).
All patients receiving an LVAD at our centre are included in a prospective dedicated database and followed over time. This database includes baseline characteristics, peri-operative and postoperative data and follow-up information. Therefore, this is a retrospective study using a prospective database. All complications were classified according to the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) definitions. Pump infection was defined as a pocket infection for HeartMate II devices, this was not the case for intrapericardially placed HeartWare ventricular assist device (HVAD) and HeartMate 3 devices.
Short-term mortality was defined as 30-day mortality. Mortality after 30 days was considered long-term mortality. There was 1 patient who died after 45 days within the same hospital admission after surgery and was, therefore, considered as long-term mortality.
Normality of continuous variables was assessed by performing the Shapiro–Wilk test for normality and by inspecting Q–Q plots. Continuous variables are presented as mean ± standard deviation or as median with interquartile range (IQR) if assumption of normality could not be met. All variables with a P-value of lower than 0.05 were considered statistically significant.
Survival was assessed using a Kaplan–Meier analysis, patients were censored in case of a heart transplantation or re-exchange. Logistic and cox regression analyses were used to identify potential risk factors for short-term and long-term adverse events, respectively. Potential prognostic factors and target events are listed in Supplementary Material, Table S1. Odds ratios (ORs) and hazard ratios (HRs) are reported with 95% confidence intervals (95% CI). Correlation between variables was assessed using Pearson or Spearman correlation coefficient whenever appropriate. In case of statistically significant correlation between variables, the clinically most relevant variable was selected for further analysis. Multivariable analyses were not performed due to a relatively low number of events. Patients were censored at the time of heart transplantation, re-exchange or death.
Missing data was handled by using missing value analysis and multiple imputation was applied to impute missing data, according to Papageorgiou et al. [18]. To assess the mechanism of the missing values, binary logistic regression was utilized to exclude a missing not at random mechanism. All relevant variables used for analysis with a missing value percentage below 25% were accepted for imputation (Supplementary Material, Table S2). For our analyses, 5 multiple imputation datasets were generated using SPSS software. The imputation model included all relevant baseline variables. All analyses were performed using SPSS version 26 (IBM Corporation, Armonk, NY, USA).
Sixty-one unique patients were identified, who underwent a total of 80 LVAD exchanges (Fig. 1). The annual frequency of LVAD exchange was not linear, it fluctuated with an increase over the years. On average, approximately 6 exchanges were performed per year. During the study period, a total of 403 primary LVADs were implanted. Seventeen patients underwent a re-exchange of which 1 underwent 2 more re-exchanges due to a device infection, multiple episodes of pump thrombosis and a failed LVAD explanation (Supplementary Material, Table S3). Median time of follow-up was 2.0 years (IQR, 0.7–3.9). Mean age at the exchange was 49.7 ± 12.5 years, and 37.7% were female. At the initial LVAD implantation, the reported cause of heart failure dilated cardiomyopathy in 70.5%, ischemic CMP in 26.2% and (dilated) hypertrophic CMP in 3.3% of the patients. Most LVADs were implanted as bridge-to-transplantation or bridge-to-decision/candidacy (85.2%). Index devices were HeartMate II (62.3%), HVAD (29.5%) and the HeartMate 3 (8.2%). Median duration from index implantation to LVAD exchange was 891 days (IQR, 297–2188), 787 days (IQR, 299–1194) and 134 days (IQR, 59–1260) for the HVAD, HeartMate II and HeartMate 3 as index device, respectively. Patient characteristics stratified by index LVAD are displayed in Supplementary Material, Table S4. Indications for LVAD exchange were device malfunction (88.5%), device infections (8.2%), prophylactic exchange for the risk of HVAD dysfunction (1.6%) and recurring gastrointestinal bleeding (1.6%). Implanted devices were HeartMate II (60.7%), Heartmate 3 (29.5%) and HVAD (9.8%) (Table 1).
Figure Flowchart of subjects and exchanged devices. No. 1 corresponds with the index device placed in patients before undergoing LVAD exchange surgery (n = 61), subsequent levels in the flowchart show further (re-)exchanges and device succession with a total of 80 exchange procedures. LVAD: left ventricular assist device.
Median cardiopulmonary bypass time at the exchange was 73.0 min (IQR, 31.5–171.8). Concomitant cardiac surgery was performed in 10 cases, of which 2 (3.3%) underwent an aortic valve replacement, 1 (1.6%) a supracoronary ascending aorta replacement with aortic valve replacement and 4 (6.6%) received a temporary right ventricular assist device implantation (Supplementary Material, Table S5).
A total of 15 (24.6%) patients died after LVAD exchange of which 5 within 30 days and 10 thereafter. Ischemic cerebrovascular accidents (13.3%), multi-organ failure (13.3%) and pump failure (6.7%) were the causes of death in the short-term. The most common causes for long-term mortality were intra-cranial haemorrhage (26.7%), multi-organ failure (20.0%) and pump failure (6.7%) (Table 2).
Survival was 83% (95% CI 75.5–95.3%) and 67% (95% CI 53.9–84.7%) at 1 and 6 years, respectively (Fig. 2A). After 1 year of follow-up the rates of transplantation, re-exchange and being alive with the replaced LVAD in situ were 6.3%, 3.8% and 75.0%, respectively (Fig. 2B). After 5 years of follow-up, these rates were 25.0%, 23.8% and 32.5%, respectively. Freedom from re-exchange was 94.6% (95% CI 84.3–98.2%) and 45.5% (95% CI 25.8–63.3%) at 1 and 4 years, respectively (Fig. 2C).
Figure (A) Long-term Kaplan–Meier survival curve for LVAD exchange recipients (n = 61). Survival after 1 year is 83% and 67% after 6 years. Censoring rate is 75%, reasons are heart transplantations and alive without event. (B) Graphical view for competing outcomes after LVAD exchange procedures (n = 80). Analysis was conducted procedure-based. At any point in time, the total sum of proportions of each outcome is equal to 1. (C) Long-term Kaplan–Meier curve for freedom from LVAD re-exchange (n = 61). Survival after 1 year is 95% and 47% after 4 years. Censoring rate is 72%, reasons are transplantation, death or alive without event. LVAD: left ventricular assist device.
The most frequently observed short-term complications after LVAD exchange procedures were right heart failure (16.3%), respiratory failure (16.3%), pulmonary infection (16.3%) and ventricular tachycardia (13.8%) (Table 3). Re-sternotomy due to bleeding or pericardial effusion was performed after 10 procedures (12.5%). After 9 (11.3%) procedures, a septic episode was endured.
In the long-term, exit-site infections (34.7%), device malfunctions (25.3%), hepatic dysfunction (16.0%) and major haemolysis (16.0%) were the most frequently observed complications after an exchange procedure (Table 4).
HVAD as index LVAD (OR 12.00, 95% CI 1.24–116.53) was univariably associated with an increased risk of post-operative cerebrovascular accidents (Table 5). HVAD as index LVAD (OR 13.05, 95% CI 2.37–71.88) was also associated with a higher risk of developing respiratory failure. Patients with HVAD as index LVAD (OR 4.85, 95% CI 1.39–16.87) had a significantly higher risk of developing post-operative right heart failure. Lastly, an elevated LDH (OR 7.09, 95% CI 1.35–37.29) and the HVAD as index device (OR 4.88, 95% CI 1.12–20.19) were associated with a higher risk of resternotomy for haemothorax after exchange. No significant risk factors were identified for sepsis, ventricular tachycardia, pulmonary infection and renal dysfunction (Supplementary Material, Table S6).
Univariable analyses showed that preoperative increased C-reactive protein (CRP) was associated with an increased risk of pump thrombosis (HR 1.00, 95% CI 1.00–1.02) during long-term follow-up (Table 6). Preoperative increased CRP (HR 1.01, 95% CI 1.00–1.02), elevated LDH (HR 4.80, 95% CI 1.25–18.46) and having HVAD as index LVAD (HR 6.42, 95% CI 1.80–22.90) were all univariably associated with an increased risk of long-term right heart failure (Table 6). HVAD LVAD as index device (HR 7.81, 95% CI 1.95–31.23) was significantly related with the development of respiratory failure (Table 6). Elevated LDH (HR 3.15, 95% CI 1.08–9.22) was associated with an increased risk of LVAD re-exchange (Table 6). No significant risk factors were found for cerebrovascular accident and gastrointestinal bleeding during long term follow-up in this study (Supplementary Material, Table S7).
Increased CRP was significantly associated with short-term (OR 1.01, 95% CI 1.00–1.03) and long-term mortality (HR 1.01, 95% CI 1.00–1.02) (Tables 5 and 6).
The present study investigated patient characteristics, survival and complications after LVAD exchange. This study indicates that having an HVAD as index LVAD is associated with a higher risk of developing right heart failure and respiratory failure after device exchange. Furthermore, our results show that the 6-year mortality after LVAD exchange is relatively low.
The 1-year survival in our study is comparable to what has been reported before [11, 14, 16]. To our knowledge, there are 2 studies that report survival up to 4 years [11, 19]. In the 1st study, the survival after 4 years was 50% which is lower in comparison to our study (67%). A possible explanation is that Anand et al. only included axial flow pumps and not centrifugal pumps. Centrifugal pumps are more durable and provide a longer survival, hence the reason for the difference in survival rates [3, 4, 20]. In addition, a recent study published by Contreras et al. found similar survival rates with a comparable group of index LVADs, survival analyses were however performed with insufficient numbers [19]. Furthermore, in that study, pre-operative extracorporeal membrane oxygenation, mechanical ventilation and multiple sternotomies were found to be associated with mortality or the need for a subsequent exchange [19]. This is different from our study, which might be explained by population differences. In this study, we have a larger group of women, more patients with dilated cardiomyopathy and our study was performed in Europe versus the United States. Additionally, re-exchange and mortality were not considered as a composite outcome in our study but were analysed separately.
In our study, pre-operative elevated CRP is associated with higher postoperative and long-term mortality after device exchange. This finding is in line with a study published by Tie et al. [21], which suggests that CRP is a comprehensive predictor for short and long-term mortality in LVAD recipients. The exact relationship between elevated CRP levels and mortality is still unclear. The current hypothesis states that a systemic inflammatory response is induced which could pose an increased risk of mortality [22].
To our knowledge, this is the 1st study that reports on RHF after LVAD exchange. However, there is literature on this subject in patients undergoing primary LVAD implantation. The reported incidence ranges between 3.9% and 53.0%, depending on patient population and type of LVAD implanted [23]. The hypotheses behind the pathophysiology for RHF during LVAD therapy vary and may include anatomical reasons where the blood flow from the LVAD causes the interventricular septum to shift to the left ventricle due to decreased left ventricular peak pressure, leaving more diastolic compliance on one hand but reducing the efficiency of right ventricular contractility on the other hand [24]. Another explanation could be the shift in volume distribution; LVADs increase forward flow to the systemic circulation and increase venous return beyond the capacity of the right ventricle [25].
In our study, HVAD as index LVAD was associated with a higher risk of developing RHF in both short- and long-term follow-up. This finding is in accordance with previous studies that reported a higher incidence of RHF in HVAD patients at primary implantation [4]. However, patient groups are not equal across different studies and the HVAD patient group in general is at disadvantage in regard to many patient characteristics. For example, our previous study indicates that after propensity matching, there remains a difference in survival, it is however a lot smaller than originally thought [26]. Ultimately, the mechanism behind this association is still unknown and further research is needed to investigate this finding.
Respiratory failure has also been described as a complication after LVAD implantation with rates reported in literature between 14% and 38% [27]. In our study, 16.3% has developed respiratory failure during the postoperative phase and 2.7% during the long-term follow-up.
Our analyses also show that the HVAD as index device is possibly associated with an increased risk of developing respiratory failure. While current literature reports no causal mechanism, the same pathophysiology may be at play as in the mechanisms for RHF after HVAD exchange since respiratory failure is described as a deleterious complication of worsening heart failure [28].
Our study is subject to several limitations. First, the analysis was conducted retrospectively and the data are from a single centre. However, the majority of the data were collected prospectively in a dedicated database with a consistent follow-up. Furthermore, complication management, surgeon-specific experience and the conditions in which patients are treated may not be homogenous between different hospitals and might limit the generalizability of our results. In addition, our low numbers of events may have caused unreliable risk estimates and prevented the development of multivariable models. Therefore, the results should be interpreted with caution. However, in our study, we documented the complete range of relevant complications, and explored potential risk factors for major adverse events associated with a device exchange. Additionally, this is the 1st study to report the 5-year results with a sufficient amount of patients in this specific group of patients which will further grow in size with an increasing number of LVADs being implanted as a destination therapy.
Lastly, currently a significant group of patients is still supported by the HeartMate II or HVAD. The current study can potentially contribute to pre-operative risk assessment of these patients if LVAD exchange is needed. To date, this is one of the few studies providing insight in the long-term results following LVAD exchange, a procedure that is increasingly more often performed in the last few years.
In June 2021, the HVAD LVAD was withdrawn from the market due to an increased incidence of neurological adverse events and mortality associated with the internal pump [29]. However, current literature suggests that it is not recommended to replace the HVAD pump unless serious complications arise warranting exchange [30]. In our study, we found that device exchange is a viable option with an acceptable survival rate. On the other hand, our research suggests that the patients who have an HVAD as index device may be at increased risk of developing respiratory failure and right-sided heart failure after device exchange. Considering the current evidence, it may be concluded that device exchange remains a viable option, but the risk of complications should be carefully weighed.
To validate our findings, a multicentre study with a larger group of patients should be conducted to provide more precise risk estimates for complications following LVAD exchange.
Although LVAD exchange can be performed with a relatively low mortality, other post-operative adverse events are common. Patients with the HVAD as an index device may be at higher risk of developing right heart failure and respiratory failure after exchange.
Jaiel Niamat, Department of Cardiothoracic Surgery, University Medical Center Utrecht, Utrecht, Netherlands.
Faiz Ramjankhan, Department of Cardiothoracic Surgery, University Medical Center Utrecht, Utrecht, Netherlands.
Niels Van Der Kaaij, Department of Cardiothoracic Surgery, University Medical Center Utrecht, Utrecht, Netherlands.
Monica Gianoli, Department of Cardiothoracic Surgery, University Medical Center Utrecht, Utrecht, Netherlands.
Linda W Van Laake, Department of Cardiology, University Medical Center Utrecht, Utrecht, Netherlands.
Mostafa M Mokhles, Department of Cardiothoracic Surgery, University Medical Center Utrecht, Utrecht, Netherlands.
Supplementary material is available at EJCTS online.
No funding was provided for this project.
**Conflict of ** Jaiel Niamat, Faiz Ramjankhan, Niels Van Der Kaaij, Monica Gianoli and Mostafa Mokhles have no conflict of interest. The UMCU, as employer of Linda Van Laake, received consultancy fees or educational grants from Medtronic, Abbott, Vifor and Novartis outside the submitted work.
The data underlying this article will be shared on reasonable request to the corresponding author.
Jaiel Niamat: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Visualization; Writing—original draft; Writing—review & editing. Faiz Ramjankhan: Data curation; Supervision; Writing—review & editing. Niels van der Kaaij: Writing—review & editing. Monica Gianoli: Writing—review & editing. Linda van Laake: Writing—review & editing. Mostafa Mokhles: Formal analysis; Methodology; Writing—original draft; Writing—review & editing.
European Journal of Cardio-Thoracic Surgery thanks J.F. Matthias Bechtel, Stefano Mastrobuoni and Imre János Barabás for their contribution to the peer review process of this article.