Incidence and outcome of Levitronix CentriMag support as rescue therapy for early cardiac allograft failure: a United Kingdom national study

Abstract

Objective: Primary graft failure is the most common cause of mortality early after heart transplantation. The availability of relatively low-cost short-term mechanical support devices has altered the management of primary graft failure but there are few data on clinical outcome. Here, we describe the UK experience with Levitronix CentriMag support following heart transplantation across multiple centres. Methods: Data for all adult heart transplants and all CentriMag devices used within 30 days of heart transplantation in the UK between November 2003 and July 2008 were collected. Transplant characteristics were compared for those who did and did not receive CentriMag support, and device outcomes and survival rates were summarised. Results: A total of 572 heart transplants were performed in this period. As many as 38 patients (6.6%) were implanted with CentriMag devices for primary graft failure. Four patients received extracorporeal membrane oxygenation concurrently and were excluded from further analysis. There were no significant differences in transplant characteristics between the patients who received CentriMag support and those who did not. Twelve patients were explanted; nine survived but three died shortly afterwards. Five underwent acute retransplantation; two survived and three died. Seventeen patients died on support. The 30-day and 1-year survival rates were 50% (95% confidence interval (CI) 32–65%) and 32% (95% CI 18–48%), respectively. Patients who previously had a bridge-to-transplant ventricular assist device (VAD) had significantly better survival than those who did not (1-year survival 71% vs 22%, p = 0.029). Conclusions: Primary graft failure remains an important early complication of heart transplantation. Levitronix CentriMag support led to the salvage of 32% of patients with severe allograft failure.

Heart transplantation, Heart-assist devices, Primary graft dysfunction

1 Introduction

In heart transplantation (HTx), the most common cause of early mortality is failure of the allograft, accounting for 41% of deaths within 30 days [1]. This figure includes deaths due to Primary alloGraft Failure (PGF) which, by definition, excludes immunological and technical causes of graft loss, among others, and is believed to largely represent the effects of ischaemia-reperfusion injury. Various risk factors have been identified, including donor age, recipient pulmonary artery pressure and vascular resistance and ischaemia time [2,3], but these factors do not explain its full incidence, and PGF remains a risk for all HTx operations. PGF is the most extreme form of the more common primary graft dysfunction (PGD). There is no universally agreed definition of PGF or PGD in the literature, and the diagnosis is normally made following the exclusion of technical and immunological causes of allograft failure, including hyperacute rejection. Medical treatment comprising inotropic support, pulmonary vasodilators, intra-aortic balloon counterpulsation (IABP) and, more recently, levosimendan have established roles in the management of PGD, but the more profound haemodynamic insufficiency of PGF demands additional measures including extracorporeal membrane oxygenation (ECMO) or ventricular assist device (VAD) implantation. These devices may be used as a bridge to allograft recovery, or a bridge to retransplantation, but there are limited data on clinical outcome.

Prior to the advent of short-term VAD support, PGF was likely to be uniformly fatal except for isolated cases where emergency salvage retransplantation was possible. In recent years, the availability of short-term, relatively low-cost VADs has altered the management of PGF in the United Kingdom (UK), as these devices can be deployed rapidly allowing management to cardiac recovery or consideration of retransplantation. The Levitronix CentriMag^® (Levitronix, LLC, Waltham, MA, USA) is an extracorporeal, centrifugal VAD that uses magnetically levitated bearingless motor technology [4], which reduces friction and wear, and can therefore reduce the risk of haemolysis and thrombus formation. The pump can be used for uni- or biventricular support and the manufacturers report that flow rates of up to 10 l min⁻¹ can be achieved with a pump speed of 5500 revolutions min⁻¹. The CentriMag Ventricular Assist System is CE Marked and approved for use for up to 30 days (http://www.levitronix.com, accessed 13 August 2010). It is commercially available in Europe, but, in the USA, it is limited to investigational use only, with clinical trials underway. UK cardiac transplant centres have had access to this device for the management of PGF since 2003.

Patient outcome data following the use of mechanical support for PGF is limited and most reports are from single-centre series [5–10]. We therefore examined a national cohort of patients, who received a Levitronix CentriMag device for the treatment of PGF in UK transplant centres. Our aims were to describe the UK experience with CentriMag support following HTx, to identify any donor, recipient or operative factors that influenced the requirement for post-transplant CentriMag support and to report patient outcomes following device implantation.

2 Materials and methods

Data were analysed from the comprehensive UK Cardiothoracic Transplant Audit, which has been described previously [11]. We analysed all adult UK heart transplants and all CentriMag devices used within 30 days of transplant between November 2003 and July 2008. Medical records for the CentriMag patients were reviewed to collate the incidence of adverse events during device support: bleeding, renal failure, infection, neurological dysfunction and return to theatre (excepting for device explant or retransplantation). Data were collected for the first adverse event of each type. Bleeding was defined as suspected internal or external bleeding requiring surgical intervention or transfusion of packed red blood cells; renal failure as the need for dialysis or haemofiltration; neurological dysfunction as any new, temporary or permanent, focal or global neurological deficit; and infection as any infection requiring antibiotic treatment.

Donor, recipient and operative characteristics were compared for those who did and did not receive CentriMag support using Fisher’s exact test, the two-sample t-test or the Mann–Whitney test, as appropriate. Risk factors included those associated with PGF following HTx in the UK [3]. Missing data ranged between 0% and 13% across factors, and were imputed using a random number generator for categorical data, and the mean or median value for continuous data.

Duration of intensive care unit (ICU) stay, hospital stay, patient survival, graft survival (time from implant to the earlier of retransplantation or death), survival while on VAD support (time from implant to death, censoring for explantation or retransplantation) and survival free from adverse events were summarised using the Kaplan–Meier estimate of the survivor function. Risk factors considered were: recipient age, sex, ethnicity, primary disease and graft number; donor age, sex and cause of death; and ischaemia time, urgency, day of implant, need for biventricular support and previous use of a VAD as a bridge to HTx. Univariate analysis was performed for each factor using the log rank test; due to the small number of patients in the study, no multivariable analysis was performed. The cumulative incidence of explantation, retransplantation and death were estimated using competing risks analysis.

Transplant and VAD data were initially collected for the audit on the basis of presumed consent, but, more recently, informed consent has been sought when listing the patient or implanting the device, respectively. In the UK, audit projects do not require separate research ethics committee approval (Defining Research, ref: 0987 December 2009, available at http://www.nres.npsa.nhs.uk/news-and-publications/publications/general-publications, accessed 13 August 2010). All statistical analyses were undertaken using SAS (Windows) version 9 (SAS Institute Inc., Cary, NC, USA), and a p-value less than 0.05 was considered statistically significant.

3 Results

3.1 Incidence and nature of VAD support

Between November 2003 and July 2008, 572 adult heart transplants were performed in the UK. The overall 30-day mortality was 11% and, of the 63 early deaths, 39 were reported to be secondary to PGF. We identified 38 patients (6.6%) in whom a CentriMag device was implanted for PGF (univentricular or biventricular failure) within 30 days of HTx. None of these patients had a positive donor-specific crossmatch or a technical cause of graft failure. Four patients received ECMO concurrently and were excluded from further analysis; two of these were explanted and two died on support. Recipient, donor and operative characteristics for the 34 patients who received isolated CentriMag support and the 534 patients who did not receive ventricular support are compared in Table 1 . A larger proportion of the CentriMag patients were of non-white ethnicity and received a VAD for bridging to heart transplant, but none of the differences attained statistical significance.

Table 1

Recipient, donor and transplant characteristics.

The majority (22) of patients received biventricular support, but 10 received a right VAD only (RVAD) and two patients a left VAD only (LVAD). Fig. 1 presents the day of VAD implant and duration by 1-year survival status. Twenty-six patients (76%) were implanted on the day of transplant, three patients on day 1, four patients on day 3 and one patient was implanted on day 12. The duration of VAD support ranged between 0 and 45 days but 56% were on support for ≪10 days. The median duration of support was 8 days.

Fig. 1

Survivors (hashed) and non-survivors (solid) by day of implant in relation to heart transplant (a) and duration of VAD support (b).

3.2 Outcome of support

Fig. 2 gives the 1-year outcome of CentriMag support. Twelve patients (35%) were explanted; two died soon afterwards on day 5 and day 20 post-explant, one further patient died on day 71, but the other nine survived 1-year post-explant. The three patients who died following explantation were those with the longest device durations (>30 days). Five patients (15%) underwent retransplantation, 1, 2, 7, 7 and 8 days following initial transplantation. Two patients died within 30 days (days 0 and 13), a further patient died on day 153 and two patients survived 1-year following retransplantation. The remaining 17 patients (50%) died on support between day 0 and day 45. There were no device failures. Fig. 3 shows the cumulative incidence of each outcome over time. At 30 days post-implant, 44% of patients had died on support, 26% had been explanted and 15% re-transplanted, leaving 15% still on support.

Fig. 2

One-year outcome of VAD support.

Fig. 3

Incidence of patients in each outcome group over time, estimated using competing risks analysis.

3.3 Patient and graft survival

Fig. 4 presents overall patient and graft survival following device implantation. Patient survival is calculated regardless of other events, such as explantation or retransplantation. The majority of deaths occurred early in the post-implant period; 7-day survival was 79% (95% confidence interval (CI) 62–90%), 30-day survival was 50% (95% CI 32–65%) and 1-year survival was 32% (95% CI 18–48%). The equivalent allograft survival rates were 68% (95% CI 49–81%), 41% (95% CI 25–57%) and 26% (95% CI 13–42%). Fig. 5 presents patient survival while on VAD support, censoring at explantation or retransplantation; 7-day survival was 78% (95% CI 60–89%) and 30-day survival was 37% (95% CI 17–57%).

Fig. 4

Kaplan–Meier estimate of patient survival (solid line) and graft survival (dashed line) after VAD implantation.

Fig. 5

Kaplan–Meier estimate of patient survival while on VAD support, censoring at explantation or retransplantation.

For the 17 patients who died on support, one patient died of haemorrhage, one due to mesenteric infarction and in one, the cause of death was unknown. The other 14 died of irreversible donor organ failure and its complications. The three patients, who died after explantation, died due to a cardiac cause.

In univariate analysis, only one of the risk factors considered was found to be significantly associated with patient survival following CentriMag implantation (regardless of other events): the previous use of a VAD as a bridge to HTx (p = 0.029). The seven patients who had a bridging VAD had significantly better 1-year patient survival (71%, 95% CI 26–92%) compared with the 27 patients who did not (22%, 95% CI 9–39%). None of the risk factors considered were found to be significantly associated with survival while on CentriMag support.

3.4 Hospital stay and adverse events

Both ICU and hospital stay were significantly longer in the CentriMag group compared with the non-CentriMag group (p ≪ 0.0001). Median ICU stay was 39 days (95% CI 14–64 days) compared with 4 days (95% CI 4–4 days), while median hospital stay was 59 days (95% CI 15–103 days) compared with 24 days (95% CI 23–25 days).

Full medical records were available for review for 30 of the 34 patients who received CentriMag support. Adverse events were frequent (Table 2 ); 24 of 30 patients (80%) experienced at least one adverse event, with a maximum of five adverse events for two patients. The average number of adverse events per patient was 2.2. When including death, 87% of patients experienced at least one adverse event, with an average of 2.8 adverse events per patient. Bleeding was the most common adverse event affecting a total of 21 patients (70%) and 20 patients underwent further surgery. Approximately half (47%) of patients experienced renal failure and 20% had infections. Survival free from adverse events (including death) was estimated as 13% (95% CI 4–28%) at 6 days (the maximum period of device support without any adverse events).

Table 2

Adverse events while on CentriMag support (30 patients whose notes could be accessed).

4 Discussion

To our knowledge, this is the first multicentre study to examine the use of short-term VADs in the treatment of PGF after heart transplantation. In the era prior to mechanical support techniques, PGF was almost uniformly lethal unless emergency retransplantation was possible, and, even in these circumstances, outcomes were poor [12–14]. Our data reflect the perilous clinical condition of patients with PGF, but provide some encouraging information about the natural history and treatment of this condition. In our series, 6.6% of adult HTx patients received a CentriMag device within 30 days of HTx; a similar proportion reported receiving mechanical support for PGF (excluding IABP) in other studies (3.9% and 7.0%) [7,8]. There were no statistically significant differences observed between the donor, recipient or transplant characteristics of those who did and did not receive a CentriMag device, but the use of pre-transplant VAD was higher in keeping with previous reports [2,15]. The 30-day and 1-year survival were expectedly inferior to the non-CentriMag cohort (p ≪ 0.0001). Twenty-six percent of CentriMag patients recovered allowing explantation and 1-year survival. The time course of recovery to explantation was 5–26 days, a course compatible with a diagnosis of reversible cardiac stunning or acute hibernation within the donor heart. Outcomes for patients bridged to acute retransplantation were poor; 40% died in hospital, 60% within the first year. In an era of severe donor heart shortage, this fuels the debate regarding organ allocation to such a high-risk group when other patients may gain a greater survival advantage from the same organ [16–18].

The rationale for CentriMag support in the setting of PGF is twofold: (1) to maintain cardiac perfusion while avoiding the excessive use of inotropes and vasopressors to allow ventricular recovery and (2) to maintain other end-organ perfusion to prevent development of multisystem failure. As there was no control group in our observational study, we cannot draw any firm conclusion about the impact of CentriMag support on patient survival. However, as these patients were all essentially in cardiogenic shock at the time of device implantation, we anticipate that the survival rate without the use of this or an alternative high-flow device would have been very low.

The Papworth [10], Harefield [6,19], Pittsburgh [5] and Minnesota [20] groups have all published on the use of the Levitronix CentriMag for a variety of indications, including post-cardiotomy cardiogenic shock, bridge to HTx, bridge to decision and PGF. Other studies from New York [7] and Melbourne [8] have reported on several types of mechanical support in the treatment of PGF. Patient survival rates in the former group of studies were 30–65% (with the highest rates in the two studies that reported on bridge to decision only) [19,20], and in the latter group were 30–66%. The 32% 1-year survival rate observed in our cohort was therefore similar to these reports, although the patient population and device strategy differed considerably among the studies. Survival rates for the most comparable subset of PGF patients treated with the CentriMag were 50% in the Pittsburgh, Papworth and Harefield [6] reports, but these were small subsets (10, six and four patients, respectively). In addition, the last two journals report on patients included in our comprehensive national survey.

Patients, who received a VAD for bridging prior to HTx, had significantly better patient survival following CentriMag implant for PGF (p = 0.029). After accounting for multiple comparisons, this p-value would no longer be significant, but the effect may be genuine and requires corroboration from other studies. There are a number of possible explanations. First, it is possible that device cannulae or connections remained accessible during the HTx surgery, and the CentriMag was instituted at a lower threshold of PGD with a higher chance of recovery. Second, it is possible that the very presence of a VAD preoperatively, favourably influences the pulmonary vascular resistance. Third, prior VAD support may have allowed correction of the hostile pro-inflammatory environment of previous end-stage heart failure. These reasons could explain the better prognosis in this subset of patients (1-year survival of 71%), but remain speculative and it should be noted that prior VAD support increases the risk of PGF in some studies [2,15]. None of the risk factors considered were significantly associated with survival while on CentriMag support.

Device support was associated with frequent adverse events and extended ICU and hospital stays. Bleeding was particularly common (70%), reflecting the need for anticoagulation, the presence of the multiple vascular suture lines and an ongoing device-related coagulopathy [21,22]. Almost half the patients required renal support therapy, and two patients suffered embolic strokes and died on support a short time later.

CentriMag VADs were the most common mode of post-transplant support used in the UK during the period of this study, and we are unable to make a direct comparison with other modes of support, such as veno-arterial extracorporeal membrane oxygenation (ECMO). Two studies have reported survival rates of 50% and 53% following ECMO support for primary allograft failure [23,24] compared with 50% 30-day and 32% 1-year survival in this study. However, survival-rates are influenced by many factors, including the threshold at which support was instituted; hence, the results from these single-centre studies cannot be used to compare the efficacy of ECMO with that of CentriMag VADs seen in this comprehensive national study.

There are a number of limitations to our study. Although this represents the largest report to date, numbers remain small and implants occurred over a period of almost 5 years. However, the results encompass the outcome of all cases treated with CentriMag support for PGF in the UK during the study period and so avoid any ‘publication bias’. The occurrence of PGF is not recorded in our national audit, and hence, only those PGF cases that received CentriMag support could be identified. Our study therefore reports that there were no significant differences between the CentriMag and non-CentriMag groups, acknowledging that these may not directly correlate to the PGF and non-PGF groups, but provides the best evidence possible from the available data. A large number of statistical tests were performed, increasing the potential type I error rate; and we did not perform any multivariable analysis because of the limited data available. Data on individual treatment decisions, pre- and post-transplant antibodies, device flows, pulmonary vasodilator drugs and anticoagulation used by centres were not available for analysis. Clinicians in the individual centres made the treatment decisions regarding the use of the device without pre-defined indications or selection criteria, although anecdotally we know that they were typically implanted when patients had a low cardiac index and evolving organ dysfunction despite multiple inotropic drugs and/or could not be weaned from cardiopulmonary bypass at the end of the transplant operation. The lack of a common definition of PGF employed across centres is therefore a limitation in our study, but applies throughout the PGF literature due to the absence of a universally agreed-upon definition. Because CentriMag support was the predominant approach used in the UK to treat severe PGF during the study period, we were unable to compare this strategy with other forms of circulatory support.

In conclusion, the use of the Levitronix CentriMag device in a national cohort of HTx patients with severe PGF resulted in a patient survival rate of 32% at 1-year, and the incidence of adverse events was high. However, without mechanical support most would have died, and, hence, mechanical circulatory support is an important option in this very sick group of patients. Further advances in VAD technology and its clinical application may improve results. However, PGF continues to represent a transplant catastrophe and research must also focus on prevention through better donor assessment, identification of robust non-clinical risk factors in both donor and recipient (e.g., biomarkers predictive of PGF) [25] and improved myocardial protection during the transplant procedure [11].

Acknowledgements

This work was undertaken on behalf of the Steering Group of the UK Cardiothoracic Transplant Audit: Dr NR Banner (chairman) and Mr Andre Simon, Harefield Hospital, Harefield; Professor R.S. Bonser, Queen Elizabeth Hospital, Birmingham; Professor P. Corris, Freeman Hospital, Newcastle; Mr P. Braidley, Northern General Hospital, Sheffield; Mr S. Tsui and Dr J. Parameshwar, Papworth Hospital, Papworth Everard; Professor N. Yonan, Wythenshawe Hospital, Manchester; Dr M. Burch, Great Ormond Street Hospital, London; Mr U. Nkere, Golden Jubilee National Hospital, Glasgow; Dr J. van der Meulen, Royal College of Surgeons of England; Professor D. Collett, NHS Blood and Transplant; and Dr M. Ashton-Key, NCG. The authors thank the transplant and VAD coordinators in the transplant centres for collecting the audit data and Rhiannon Birch and Kerri Barber (NHS Blood and Transplant) for their assistance with additional statistical analysis and revision of the article.