|Home | About | Journals | Submit | Contact Us | Français|
We sought to analyze the indications and outcome of extracorporeal membrane oxygenation (ECMO) for early primary graft failure and determine its impact on long-term graft function and rejection risk.
Early post-operative graft failure requiring ECMO can complicate heart transplantation.
A retrospective review of all children requiring ECMO in the early period after transplantation from 1990 to 2007 was undertaken.
Twenty-eight (9%) of 310 children who underwent transplantation for cardiomyopathy (n = 5) or congenital heart disease (n = 23) required ECMO support. The total ischemic time was significantly longer for ECMO-rescued recipients compared with our overall transplantation population (276 ± 86 min vs. 242 ± 70 min, p < 0.01). The indication for transplantation, for ECMO support, and the timing of cannulation had no impact on survival. Hyperacute rejection was uncommon. Fifteen children were successfully weaned off ECMO and discharged alive (54%). Mean duration of ECMO was 2.8 days for survivors (median 3 days) compared with 4.8 days for nonsurvivors (median 5 days). There was 100% 3-year survival in the ECMO survivor group, with 13 patients (46%) currently alive at a mean follow-up of 8.1 ± 3.8 years. The graft function was preserved (shortening fraction 36 ± 7%), despite an increased number of early rejection episodes (1.7 ± 1.6 vs. 0.7 ± 1.3, overall transplant population, p < 0.05) and hemodynamically comprising rejection episodes (1.3 ± 1.9 vs. 0.7 ± 1.3, overall transplant population, p < 0.05).
Overall survival was 54%, with all patients surviving to at least 3 years after undergoing transplantation. None of the children requiring >4 days of ECMO support survived. Despite an increased number of early and hemodynamically compromising rejections, the long-term graft function is similar to our overall transplantation population.
Heart transplantation in children with end-stage heart failure secondary to cardiomyopathy or failed palliation of congenital heart disease (CHD) is a good option with improving outcomes (1). One of the most common complications in the immediate period after transplantation is early graft failure. Graft failure can result from long ischemic time, inadequate myocardial preservation at time of procurement, hyperacute rejection, or poor adaptation of the graft to the recipient’s hemodynamic environment (2). Either as a consequence of left heart failure or as a result of single ventricle physiology, pulmonary vascular resistance (PVR) in many pediatric recipients is increased, resulting in the risk of right ventricular failure after transplantation.
Measures aimed at decreasing PVR after transplantation include the use of inhaled nitric oxide as well as medications with pulmonary vasodilator effects, such as prostacyclin, isoproterenol, and milrinone (3). Graft ventricular function is also commonly supported post-operatively with inotropes (e.g., dopamine, dobutamine, low-dose epinephrine, or milrinone). However, despite these interventions, ventricular failure may persist, and mechanical circulatory support becomes necessary. Extracorporeal membrane oxygenation (ECMO) is widely used for post-cardiotomy low cardiac output syndrome (LCOS) in children and is occasionally required after pediatric heart transplantation (4–6). The objectives of this study were as follows: 1) to describe the indications and outcome of ECMO for early primary graft failure after heart transplantation in children; 2) to identify markers predictive of ECMO survival in this patient population; and 3) to determine the impact of early ECMO on long-term graft function and rejection risk.
We reviewed our institutional heart transplant database from 1990 to 2007. After approval by our institutional review board, all children requiring ECMO in the post-operative transplantation period were included in the study. The course after transplantation for all eligible patients was reviewed. Our institutional transplantation methodology has been previously described but will be reviewed here (7). Donor hearts are procured in standard fashion, with great vessel lengths harvested as determined by recipient anatomy. Roe’s solution is used for donor cardioplegia. One of the ECMO-rescued recipients who survived underwent transplantation from a nonbeating heart donor. Organ procurement in the setting of donation after cardiac death has been previously described by our group (8). All of the other patients underwent transplantation from standard brain-dead, heart-beating donors. Moreover, there were no changes in our procurement practices throughout the duration of this study period.
Milrinone, dopamine, isoproterenol, and occasionally epinephrine are commonly used for inotropic support in the immediate post-transplantation period. Nitric oxide is systematically used in patients with an elevated PVR documented before transplantation (PVR ≥5 Wood units/m2). Triple immunosuppression with methylprednisolone for 48 h, low-dose cyclosporine, and azathioprine is used perioperatively. Induction therapy with methylprednisolone and antithymocite globulin (Thymoglobulin, Genzyme Corp., Cambridge, Massachusetts) was used until 1998, at which time Thymoglobulin was replaced by antithymocyte globulin (American Medical Resources, Nashville, Tennessee). Low-dose cyclosporine (target level 40 to 70 ng/ml) and azathioprine were administered throughout induction. At discharge, maintenance immunosuppression consisted of cyclosporine (target level 175 to 225 ng/ml) and mycophenolate mofetil (target level 2 to 4 μg/ml) (9). In patients with an open chest or dependent on mechanical support (ECMO), induction therapy was withheld until chest closure and/or decannulation was achieved because of the increased risk of infection.
A standardized ECMO circuit was used with appropriately sized cannulas according to the patient size (6). No other types of mechanical support devices were used during the course of this study. Transthoracic cannulation through an open chest was used in all patients via the right atrium and ascending aorta. Pump flow was regulated according to systemic perfusion, blood pressure tracing, and systemic and mixed venous oxygenation. Inotropic and ventilatory supports were weaned as tolerated during ECMO support. In patients with poor left ventricular function, venting of the left atrium or transcatheter atrial balloon septostomy was performed for left-sided decompression whenever required. Left atrial decompression was indicated in the setting of inadequate left ventricular decompression, resulting in increased left atrial pressure and/or pulmonary edema. Activated clotting time was maintained between 180 and 220 s with heparin infusion. Hemofiltration was used as required depending on the renal function and fluid status. Graft function recovery was regularly assessed by transthoracic echocardiography. Increased inotropic and ventilatory support was used for weaning off ECMO. Chest closure was usually delayed until 24 to 48 h after decannulation.
Diagnosis of acute graft rejection in our institution is based on clinical presentation, echocardiographic or electrocardiographic findings, hemodynamics at time of catheterization, and/or endomyocardial biopsy evidence of rejection (10–12). Hemodynamically significant or compromising rejection is defined as a rejection episode for which the patient required intravenous inotropic support while undergoing treatment for rejection. Hyperacute graft rejection is defined as occurring during the immediate post-operative course after transplantation. Early rejection is defined as an acute graft rejection episode occurring during the first year after transplantation, and very early rejection as occurring during the first month after transplantation. Treatment of acute graft rejection includes anti–T-cell antibodies (antithymocyte globulin or OKT3) for 7 to 10 days in combination with steroids (4 doses) and the administration of intravenous immunoglobulin per institutional protocol (12). Transplant coronary artery disease is diagnosed by angiography or intravascular ultrasound and is defined as any luminal irregularities or stenosis that varied from the patient’s previous angiography or intimal thickening ≥3 mm according to the Stanford classification (13,14).
Outcome was analyzed in 2 groups: 1) those successfully weaned off of ECMO and discharged alive (survivor group); versus 2) those who died while on ECMO or in the post-decannulation period (nonsurvivor group). The long-term outcome of the ECMO survivors was compared with our overall transplant population.
Data are presented as mean ± SD. When the population does not follow a normal distribution, median values are given. Intergroup comparison was conducted with an unpaired Student t test. Comparison of proportion was made with the Fisher exact test. A p value ≤0.05 was considered statistically significant. Statistical analysis was performed by use of the GraphPad Prism software (GraphPad Software Inc., La Jolla, California). Graphs are represented with mean and SDs.
From 1990 to 2007, 310 children underwent heart transplantation at our institution, and 28 children who underwent transplantation (9%) were placed on ECMO for post-operative primary graft failure. These 28 children underwent transplantation for cardiomyopathy (n = 5) or complex CHD (n = 23) and required ECMO support for LCOS within 48 h of their transplantation. The cardiomyopathy group included 4 patients with dilated cardiomyopathy and one with restrictive cardiomyopathy. The CHD group comprised 19 infants with hypoplastic left heart syndrome variant and 4 children with other forms of complex CHD. Six of the CHD patients had previous failed corrective or palliative heart surgery, whereas 17 were referred for transplantation as the primary therapeutic option. Patients’ ages ranged between 1 week and 19.5 years (mean 1.3 ± 3.7 years, median 0.3 years). Twenty-four (86%) patients were infants (<1 year) at the time of transplantation. Weight ranged from 3 to 70 kg (mean 7.8 ± 12.5 kg, median 4.5 kg). This population did not follow a normal distribution because of a preponderance of infant recipients.
Three of the 28 patients were on ECMO for cardiac failure before transplantation (1 survivor, 33%). Extracorporeal membrane oxygenation was started in the operating room because of the inability to wean from cardiopulmonary bypass in 16 patients (8 survivors, 50%), whereas the remaining 12 required ECMO in the first 48 h after transplantation (7 survivors, 58%) and were cannulated in the cardiac intensive care unit. Four patients had cardiac arrest before ECMO cannulation (4 survivors, 100%) and of those, 3 developed subsequent neurological deficits.
ECMO was required for right ventricular failure in 9 patients (5 survivors, 56%) and for biventricular failure in 15 (7 survivors, 47%). In the remaining 4 patients, the type of ventricular failure could not be clearly determined from review of the records. Only 1 patient was diagnosed with hyperacute rejection as the cause of primary graft failure. This patient’s graft function did not improve with standard rejection treatment, and the patient subsequently died from hemorrhagic complications 7 days after initiation of ECMO support. One survivor required 4 days of ECMO support, from which he was successfully weaned, but then he developed hemodynamically significant rejection at 6 days after transplantation and required inotropic support. This patient had complete recovery of graft function after treatment of this rejection episode and was eventually discharged home.
When comparing the survivor group with the nonsurvivor group, we found no significant difference in the time on the waiting list, age at transplantation, weight of the recipient, donor/recipient weight ratio, total ischemic time, sex, or blood type distribution between the 2 groups (Table 1). The proportion of infants <1 year of age was not significantly different between the 2 groups (87% of survivors were infants vs. 85% of nonsurvivors, p = 0.64). There was no statistical difference in the survival rate for the cardiomyopathy group (2 of 5, 40%) compared with the CHD group (13 of 23, 56%, p = 0.42). Among the CHD group, the survival rate was also not statistically different for infants with hypoplastic left heart syndrome (11 of 19, 58%) compared with patients with other congenital heart defects (2 of 5, 40%, p = 0.41).
There was no significant difference in the proportion of patients with previous heart surgery and sternotomy before transplantation between the survivors (n = 3, 20%) and the nonsurvivors (n = 3, 23%, p = 0.6). The donors’ ejection fraction was significantly lower in the survivor group (65 ± 9%, survivors vs. 71 ± 3%, nonsurvivors, p = 0.004), and although the difference in shortening fraction did not reach a statistically significant level, there was a trend toward lower shortening fraction in the survivor group (32 ± 5%, survivors vs. 41 ± 4%, nonsurvivors, p = 0.08). There was a trend toward greater peak creatinine (1.3 ± 0.7 mmol/l, survivors vs. 1.7 ± 1 mmol/l, nonsurvivors, p = 0.1) and lactate (10.1 ± 4.5 mmol/l, survivors vs. 13.4 ± 5.7 mmol/l, nonsurvivors, p = 0.19) while on ECMO in the nonsurvivor group, but these levels did not reach statistical significance.
Fifteen children were successfully weaned off of ECMO and discharged (54%), whereas 13 patients died (46%) of persistent heart failure (n = 9), diffuse bleeding (n = 3), and/or sepsis (n = 4) at a mean time of 9 days (median 7 days) after transplantation. Among the deceased patients, 7 were unable to be decannulated and died at a mean time of 5 days (median 6 days) from the initiation of ECMO support. Six patients were decannulated and died at a mean time of 14 days (median 10 days) from the initiation of ECMO support and 10 days (median 6 days) after ECMO decannulation. The cause of death in this subset of patients was mostly related to persistent cardiac failure with end-organ dysfunction and subsequent cardiac arrest. One patient was decannulated after 4 days, experienced persistent heart failure and multiorgan dysfunction (hepatic and renal failure with the need for dialysis), and could not be resuscitated after cardiac arrest that occurred 37 days after transplantation. Mean duration of ECMO was 2.8 days for survivors (median 3 days) versus 4.7 days (median 5 days) for nonsurvivors (p < 0.01), with all survivors having been on ECMO no longer than 4 days (Fig. 1). Mean duration of ECMO in nonsurvivors weaned off of ECMO was 4 days (median 4.5 days) compared with 5.3 days (median 7 days) for the nonsurvivors unable to be decannulated (p < 0.05).
When comparing the recipients requiring ECMO support for early post-operative graft failure (survivors and nonsurvivors combined) with our overall transplant population, we found that there was a significant difference with a longer total ischemic time (276 ± 86 min vs. 242 ± 70 min, p = 0.008) (Fig. 2), a younger age (1.3 ± 3.7 years vs. 4.8 ± 6.4 years, p = 0.002), and a smaller weight at transplantation (7.8 ± 12.6 kg vs. 16.8 ± 18.4 kg, p = 0.0001) for the ECMO-rescued patients (Table 2). Indications for transplantation (cardiomyopathy vs. congenital heart disease, p = 0.1), the waiting time on the list (81 ± 107 days vs. 77 ± 97 days, p = 0.4), and the donor/recipient weight ratio (1.9 ± 1 kg vs. 1.8 ± 0.6 kg, p = 0.38) were not different between these 2 groups of patients.
Neurological complications were found in 4 of the 15 (27%) survivors, 2 patients remaining with permanent neurological impairment and 2 with transient and resolved seizures. All of the patients with neurological disability were emergently cannulated in the cardiac intensive care unit, and 3 of these patients had cardiac arrest before cannulation. Overall, there was 100% 3-year survival for the patients successfully weaned off of ECMO and discharged. Two of the survivors died at 4.3 and 7.8 years after transplantation, respectively: 1 from chronic rejection and graft failure and the other from transplant coronary artery disease. Thirteen patients (46%) are still currently alive at a mean follow-up of 8.1 ± 3.8 years; 3 are >10 years and 10 are >5 years after transplantation.
The graft function in surviving patients is normal, with a mean shortening fraction by echocardiography of 36 ± 7% and no patient with a shortening fraction <25%. The mean ejection fraction obtained by cardiac catheterization was 65 ± 10% and the cardiac index by thermodilution was 4.5 ± 1.3 l/min/m2. The mean number of very early graft rejection episodes was not different for the ECMO-rescued patients compared with our overall transplant population (0.6 ± 0.6 vs. 0.4 ± 0.6, p = 0.1). The ECMO survivors had an increased number of early rejection episodes (1.7 ± 1.6 vs. 1.0 ± 1.4, p = 0.03) and an increased number of hemodynamically compromising rejections (1.3 ± 1.9 vs. 0.7 ± 1.3, p = 0.04) compared with the overall transplant population (Table 3, Fig. 3). However, there was no difference in the mean overall number of acute cellular rejection episodes between the survivors and the overall transplant population (2.9 ± 3 vs. 2.1 ± 2.7, respectively, p = 0.18). The incidence of transplant coronary artery disease was not increased in the ECMO survivors (2 of 15, 13% compared with 16% in our overall transplant population, p = NS).
Early primary graft failure after heart transplantation in children is associated with significant rates of mortality and morbidity. Extracorporeal membrane oxygenation is widely used and is well established to support circulatory function in children with post-cardiotomy LCOS (4,15). On the basis of this experience, ECMO has become a reasonable option for early graft failure in children after cardiac transplantation (4,6,16–19). Ventricular assist devices (VADs) are better suited for patients expected to require long-term support; therefore, ECMO remains the most commonly used method of mechanical circulatory support after cardiac surgery or in the post-transplantation period of pediatric patients. This article describes the early and late outcome of ECMO support for early cardiac graft failure in children and represents the largest pediatric experience to date.
In our study, primary post-operative graft failure necessitating ECMO was not uncommon (9%). The majority of patients supported with ECMO in this study were infant recipients (86% <1 year of age), reflecting the usual distribution of pediatric heart transplantation recipients, with the majority of heart transplantations occurring in the <1-yearold group (1). Nevertheless, pediatric heart recipients requiring ECMO support for early graft failure were significantly younger (p < 0.01) and had a lower weight at transplantation (p < 0.001) compared with our overall transplantation population. Moreover, their total ischemic time was significantly longer (p < 0.01), indicating that harvesting technique and time are crucial and that a longer ischemic time is a major risk factor for graft dysfunction.
The indication for ECMO cannulation did not influence the outcome, and the survival rate was not different in patients presenting with biventricular compared with isolated right ventricular failure. In addition, the survival rate was not significantly different for patients with cardiomyopathy compared with patients with CHD. Poor donor graft function has been implicated as a cause of graft dysfunction after transplantation, but this was not considered different in our study, with no difference in the shortening fraction obtained by M-mode echocardiography between ECMO survivors and nonsurvivors. Although the ejection fraction, considered less accurate compared with the shortening fraction when obtained by M-mode echocardiography, was statistically lower in the survivor group, the mean and median values remain within the normal limit, rendering this difference difficult to interpret.
The timing of ECMO cannulation was also not predictive of outcome. Survival was not significantly different between patients started on ECMO in the operating room for failure to wean from cardiopulmonary bypass compared with those cannulated in the first 48 h after transplantation for hemodynamic instability or cardiac arrest in the cardiac intensive care unit. This is in contradiction to the previous report by Galantowicz and Stolar (19), who reported no chance of survival if the fresh cardiac allograft could not support the patient after cardiopulmonary bypass. Moreover, cardiac arrest before ECMO cannulation was not a negative predictive factor for survival in our series, and all patients who had cardiac arrest survived.
This result clearly differs from survival rates of 33% to 53% for those patients who experience cardiac arrest and then are placed on ECMO as a bridge to transplantation (18,20–23). This finding suggests that the donor’s heart is more likely to recover after cardiac arrest and subsequent ECMO support than the failing ventricle of a patient who is in need of transplantation. Importantly, however, emergent cannulation was associated with a greater risk of neurologic complications, and all survivors with neurological sequelae were either cannulated emergently or experienced cardiac arrest before cannulation. Therefore, anticipating graft failure with elective cannulation before hemodynamic collapse is likely to improve neurologic outcome.
The mean duration of ECMO support in survivors was significantly less than nonsurvivors, and all survivors were decannulated within 4 days of initiation of support. This finding confirms what has been previously reported in the infant population by our institution (6). In the Mitchell et al. (6) infant study, ECMO support for >4 days was associated with death and increased risk of morbidity, particularly sepsis, bleeding, and neurologic sequelae. Recovery of ventricular function within 8 days of ECMO support has been reported and may become more common as ECMO and other mechanical support technology progress and improve over time.
However, on the basis of our institutional experience described here, consideration could be given to listing for retransplantation or transition to a VAD if the patient is not decannulated within 4 days of initiation of ECMO support (6,18,22). Of course, the clinical situation for each individual patient must be carefully considered when making the decision to relist for transplantation. The already present challenges of shortage of donor supply and increased risk of transplantation after support with mechanical assistance require that the decision to list for retransplantation should be reserved for those patients deemed to be good candidates for this option. Those individuals requiring ECMO after transplantation who have evidence of severe neurologic injury, irreversible organ injury, hyperacute rejection, and infection would not be appropriate candidates for retransplantation. The only other factor that correlated with survival was a lower donor ejection fraction. This finding suggests that the use of marginal donors, especially for the infant population where donor shortage is a problem, may not compromise outcome even if ECMO rescue is necessary in the early post-transplantation period (24).
Overall, 54% of the patients supported with ECMO were successfully weaned and discharged alive, which is comparable to what has been previously reported in the pediatric population (6,16,17). Long-term survival of those surviving to hospital discharge was excellent, with 100% alive at 3 years after transplantation.
Although the number of early (within the first year after transplantation) acute graft rejection episodes was significantly increased in the ECMO survivors compared with our other transplant recipients, there was no difference in the total number of acute graft rejection episodes between these groups. The number of hemodynamic compromising rejections per patient also was significantly increased in the ECMO survivors (1.3 ± 1.9 vs. 0.7 ± 1.2 in the overall transplant population, p < 0.05). In fact, 20% of the ECMO survivors presented with at least 1 episode of hemodynamically significant rejection requiring inotropic support.
The incidence of transplant coronary artery disease in the ECMO survivors (13%) was not increased compared with our overall transplantation population (16%). This finding is somewhat unexpected, because the increased frequency of early and hemodynamic compromising rejection in the ECMO survivors are factors that have been associated with the development of coronary vasculopathy (25,26). This finding is not related to timing of follow-up because transplant coronary vasculopathy in our program is diagnosed at a mean time of 6.4 years after transplantation, and the mean follow-up of the ECMO survivors is 8.1 years. Finally, despite increased frequency of early and severe acute graft rejection in the survivors, the graft function at long-term follow-up is normal.
This study is limited by its descriptive nature and the small sample of patients.
Primary graft failure requiring mechanical circulatory support in the early period after transplantation is not uncommon in children (9%), and a long ischemic time is a major risk factor of graft dysfunction. Pediatric cardiac allografts can be successfully salvaged by ECMO in a reasonable proportion of patients (54%). Although cardiac arrest before ECMO as well as emergent cannulation did not impact survival, these were associated with adverse neurologic outcome. The duration of cannulation was important in our series, with no child in this study surviving ECMO support for >4 days. If clinical recovery seems unlikely and the patient is a good candidate for retransplantation, consideration to relist and/or transition to VAD support is reasonable if >4 days of support is required. Importantly, long-term outcome in those patients supported by ECMO for primary graft failure and surviving to hospital discharge was excellent, with 100% 3-year survival and normal graft function despite an increased number of early rejection and hemodynamic compromising rejection episodes. Therefore, ECMO as a bridge to graft recovery after transplantation can be used without altering the long-term outcome of this challenging group of high-risk post-operative patients.
The authors thank all of the transplant coordinators of the Children’s Hospital Heart Institute for their hard work with the patients and Sam Schofield for his help with the pediatric heart transplant database.