Recovery of the native heart in any patient dying of dilated cardiomyopathy is a rare event. Most studies in adults have reported that <10% of hearts supported with an left ventricular assist device (LVAD) recover. There have been some exceptions. Birks et al.1 in 2006 reported recovery of 11 of 15 patients with LVAD support for months accompanied by maximal medical therapy and the beta-agonist, clenbuterol. This group was also successful in weaning from a continuous flow pump (HeartMate II Thermo Cardiosystems, Inc., Woburn, MA) in nonischemic cardiomyopathy in older patients ranging in age from 16 to 58 years.2 In addition, Frazier et al.3 also demonstrated LVAD bridge to recovery as an alternative treatment in 28 older patients aged 5–57 years.
In infants and children, most reports have focused on bridge to transplant with “long-term” mechanical circulatory support devices, and the number of infants has been limited. In the Pediatric Heart Transplant Study, a collaboration of 23 centers in North America, 99 ventricular assist device patients with a mean age of 13.3 years had a 77% rate of surviving to transplantation, and five patients (5%) experienced native heart recovery.4 In a recent comparison of extracorporeal membrane oxygenation (ECMO) and VAD support, 1 of 21 patients with a mean age of 4.1 years recovered after support with the Berlin Heart (Berlin Heart GmbH, Berlin, Germany). In 2006, Stiller et al.5 reported on 73 children ranging in age from infancy (2 days) to 17 years. They successfully weaned 12.5% of the patients (4 of 32) from the Berlin Heart EXCOR (Berlin Heart). Of note, all four children who were weaned were <1 year of age. Thus, even in children who might be presumed to have a better chance to recover than adults because of the shorter duration of disease, the “better tissues,” and relative stem cell abundance, there has been a low recovery rate.
Since 2004, we have prioritized recovery as a goal in infants and children treated with LVADs. We have used devices that have been available in our center. Thus, we have used the Berlin Heart since 2000, and in this study, from 2004 until early 2008, the MEDOS Heart (MEDOS Medizintechnik GmbH, Aachen, Germany) in 2004 and 2005, and the Jostra centrifugal pump (Maquet, Solna, Switzerland) since 2008. Our experience with the use of these three devices in dilated cardiomyopathies constitutes the basis for this report.
This retrospective review was approved by the institutional review board, and the requirement for patient consent was waived. Eleven consecutive patients with dilated cardiomyopathies and treated with LVAD support were included in the study that began in 2004 and ended in 2009. There has been no change in the selection criteria since the start of this series. Patients were considered for LVAD implantation only when they had signs of early multiple organ failure (lungs, liver, and kidney) on medical and inotropic therapy, had severely dilated hearts with minimal systolic function, and were felt to be in imminent danger of dying. Three different devices were used: two pneumatic pulsatile pumps, the Berlin and MEDOS hearts, and one centrifugal pump, Jostra. All pumps were connected in the same configuration: an inflow cannula in the apex of the left ventricle and an outflow cannula in the aorta.
Anticoagulation has been described elsewhere in detail.6 We used the same protocol in >375 patients with ventricular assist devices and total artificial hearts starting in 1993. Thromboelastography (TEG) and platelet aggregation studies are the most important tests. The therapeutic goals are to keep the coagulation index measured by TEG in the normocoagulable range with heparin and to use aspirin to inhibit platelet reactivity to ADP, epinephrine, and arachidonic acid, while preserving the response to collagen in platelet aggregation studies. Heparin, dipyridamole, and aspirin were used. We started with high-dose dipyridamole (0.1 mg/kg/h IV) immediately postimplantation. Once bleeding is minimal (<1 ml/kg/h), heparin was started in low dosage (10 U/kg/h) and slowly titrated up to obtain a coagulation index by TEG in the normocoagulable range. Once the platelet count is ≥150,000/ml, and bleeding continues to be minimal to nil, we started aspirin at 10 mg/kg/d and titrated up each day to the previously noted platelet aggregation study goals. Thus, for the duration of support, the patients are on heparin that is titrated using the TEG coagulation index and aspirin that is titrated to the platelet aggregation studies while dipyridamole is kept fairly constant.
Devices are run with the aim of left ventricular decompression as evidenced by decreased ventricular size on echo and by aortic valve closure throughout the cardiac cycle.
Weaning protocols have been discussed for many years and remain controversial. Our protocol was developed in 2004 and used consistently since then. It has medical and hemodynamic function components. When feasible, once on device support, maximal medical therapy was instituted using carvedilol, lisinopril, spironolactone, lasix, and levocarnitine. A 24-hour course of dobutamine at 5 μg/kg/min was given before any weaning trial. These trials were instituted once we saw reverse remodeling with significant reduction in ventricular size and increase in left ventricular wall thickening. The weaning trials were multidisciplinary and included a device engineer to manage the blood pump, a pediatric cardiologist to do the transthoracic echo, and a cardiac surgeon. During these sessions, a stepwise, timed reduction in device flow was done, typically reducing flow by one-half then completely. Heparin (50–100/kg) was administered before the trial. In addition to the echo, central venous pressure and arterial blood pressure were monitored. Maintenance of stable hemodynamics with a fractional shortening of at least 20% and not more than trivial mitral regurgitation for 5 minutes off pump was considered justification for a second weaning trial, and a second successful trial was generally followed by a third trial in the operating room that, if successful, would lead to explantation. All explants were done on normothermic cardiopulmonary bypass with a gently beating heart.
After our initial experiences with recovery, we stopped listing patients for transplantation because they all had significant signs of recovery and we believed recovery to be a better alternative. Before this and in other etiologies, patients were listed for transplant at about 1 week postimplantation.
All 11 patients were female with a mean age of 1.7 years (range: 0.08–6, median: 0.83 years), a mean weight of 9.8 kg (3.6–22 kg), a mean body surface area (BSA) of 0.48 m2 (0.24–0.82 m2), and a mean time from initial diagnosis to LVAD implantation of 88 days (range: 7–180 days, median: 26 days). Only two had positive viral serology, one coxsakie 4 and one adenovirus. All but one was on mechanical ventilation; three were on ECMO support; four were treated with three or more heart failure medications such as carvedilol, lisinopril, spironolactone, levocarnitine, and lasix; and all 11 were on intravenous inotropic support, averaging 2.4 inotropes per patient. Echo findings included a mean fractional shortening of 9% (5%–14%), a mean left ventricular end diastolic dimension Z score of +5.5 (1.6–7.3), and 2+ (moderate) mitral regurgitation. Subjectively, these ventricles were severely dilated and thin walled and seemed on short axis to “rock and roll” and not contract. The decision to implant was made only when all other means of supporting the patient were failing.
All patients received LVADs that were placed using a left ventricular inflow cannula and an aortic outflow cannula. One of the patients was temporarily supported with a Levitronix (Waltham, MA) centrifugal pump before being transferred to a pulsatile device. One patient, the youngest at implantation (2 weeks old), developed right ventricular failure after 2 days of LVAD support and had 5 days of support with a centrifugal right ventricular assist device followed by removal and 10 more days of LVAD support before successful recovery. We used eight pneumatic pulsatile LVADs, three MEDOS and five Berlin Excor hearts, and three Jostra centrifugal pumps.
The time from diagnosis of heart failure to device implantation averaged 88 days, but there was a wide variation (range: 7–570 days, median: 16 days) (Table 1). The mean number of days of support was 12.7 days (range: 6–22 days, median: 10 days). Eight patients recovered and are living from 0.6–6 years postexplant.
There were three deaths: two fatal embolic events, one stroke, and one superior mesenteric embolus. The other death was in a patient who suffered severe anoxic brain damage during a preimplant cardiac arrest. One of these patients (visceral embolus) had demonstrated recovery on two weaning trials but died the evening before scheduled explantation of a massive superior mesenteric artery embolus. Neither of the other two had “recovery,” and their hearts revealed endocardial fibroelastosis in one case and myocardial hypertrophy in the other. It is notable that the only deaths in this series were in patients with a diagnosis to implant interval of >100 days, whereas the patients who recovered had durations of <30 days.
Thromboembolism, bleeding requiring reoperation, and infection were the most common complications (Table 2). There were three other patients with emboli, thus 5 of 11 (45%) patients had an embolic event. One patient had liver and kidney infarcts that caused no long-term problem. A second patient had seizures secondary to multiple small cerebral infarcts noted on magnetic resonance imaging. This patient was treated initially for 5 days with ECMO. There has been no long-term neurological deficit 4 years later. The third patient, 6 years old, had a left middle cerebral artery embolus with right upper and lower extremity paresis and expressive aphasia. She has recovered except for some dysfunction of the right hand. In summary, there were three strokes (27%) and two visceral emboli (18%).
Five patients (45%) had seven bleeding episodes requiring reoperation. One patient had previously been on ECMO. Bleeding in the three ECMO patients was one of the indications for LVAD placement, and two of the three patients had no bleeding post-LVAD. All patients recovered from these episodes. Three patients had pneumonia with each episode being successfully treated.
Functional recovery was documented with serial echo studies. Immediately in the operating room, decompression of the left ventricle through the apical cannula led to significant size reduction of the ventricular chamber and decreased global wall thickening. Fractional shortening postexplantation rose from 9% preimplantation to outpatient values of 22% (range: 18%–30%) (p < 0.001), and the left ventricular end diastolic diameter dropped from +5 (1.6–7.3) to +1 (0.09–3.7) (p < 0.001). None of these patients have been reconsidered for transplantation, and all have been stable since explantation for durations of 0.6–6 years.
The competing outcomes analysis shown in Figure 1 demonstrates graphically that the earliest recovery was at 6 days postimplantation and that all recoveries occurred by 22 days of support. Seventy-five percent of the recoveries were during the first 16 days of support. The first death was at 7 days and the third at 21 days.
It is curious that all 11 patients were female. We have no explanation. These patients were young (1.7 years) and small with a mean weight of 9.8 kg and BSA of 0.48 m2, and the durations of their illnesses were relatively short (mean: 88 days). We assume that most were sick from viral myocarditis, but we were able to document this in only two cases. They were uniform in presentation with severely dilated nearly noncontractile left ventricles that were failing to respond to oral and intravenous therapy and in three cases failing on ECMO. In addition, the ventricles uniformly reverse remodeled, with decreasing intracavitary dimensions and symmetrical wall thickening. The one exception was the patient with an irreversibly damaged intraventricular septum that ultimately required resection after being weaned from LVAD support. She survived. Only one patient required a temporary right ventricular support. One of the three patients who died had suitable recovery criteria for explantation of her LVAD, whereas in the other two patients, recovery was still inadequate at the time support was withdrawn. Thus, the recovery rate in the patients who died seemed to be 33%, and all of these presented for implantation at >100 days from onset of the cardiomyopathy. In contrast, all of those who survived presented at <30 days from the initial diagnosis. This suggests that longer duration of dilated cardiomyopathy was associated with poorer prognosis. But, if we look at the causes of death, two from emboli and one from preimplant anoxic brain injury, or if we look at the duration of support for the patients who died (7–21 days), it is not clear why longer duration of the cardiomyopathy is associated with higher mortality with LVAD support. One might guess that longer duration of congestive heart failure is associated with hypercoagulability that increases the risk of thromboembolism because two of the deaths were embolic. The other conclusion that might be considered is that transplantation in infants and children with dilated hearts and minimal systolic function for >30 days duration may be a better option than device support.
This fairly homogeneous group of dilated cardiomyopathies with such a high percentage of recovery is provocative. It raises the possibility of virtually eliminating bridge to transplantation in this subset of patients. All infants and children would have a better long-term outcome if native heart recovery were possible rather than facing the uncertainties and complications of transplantation. It is possible that this experience could be reproduced if recovery rather than transplantation were prioritized. Unfortunately, much of this effort worldwide in this age group is exclusively directed toward transplantation.
This “dilated/systolic failure” group of infants and children seems to have the inherent capacity for reverse remodeling. We think there are three major reasons. First, the hearts are small. Second, the time from onset of congestive heart failure to device implantation is short. Third, the collagen skeleton of these hearts is reestablished once the size of the ventricular cavity is returned to normal. The time course of recovery is relatively short; most hearts were recovered in 16 days. It is less likely that this would happen in adults who have much bigger hearts that have been dilated for years. In addition, in reports of adult recovery,1–3 the support times are generally 6 months or more.
Many will doubt the need for LVAD use and continue to adhere to the dogma that ECMO is adequate. This has not been our experience. We believe that direct decompression of the left ventricle is the key to success. This could be accomplished by other approaches than apical venting of the left ventricle, but we contend that there is no better way to decompress the left ventricle with an LVAD than with apical cannulation.
Thromboembolism in 45% of the patients causing two deaths and a mild residual neurologic deficit in one is the major shortcoming of this therapy. Three of these were strokes (27%) and the other two were visceral, one fatal and one picked up as an incidental finding on computed tomography scan of the abdomen. In the Pediatric Heart Transplantation Database,4 long-term devices had a 13% stroke rate while short term had 35%, and visceral embolism was not mentioned. A recent study by Imamura et al.,7 including 21 children supported by the Berlin Heart LVAD and 21 supported by ECMO, documented a 38% stroke rate in each group. They did not mention visceral thromboembolism. The Stanford experience8 specifically mentions a higher thromboembolic/central nervous system hemorrhage rate than reported in the European literature. They had 5 of 8 (63%) neurologic events related to Berlin Heart support. In addition, they mention one death from “visceral and renal” thrombosis. Their more recent report of 29 patients documents a 55% stroke rate.9
We have followed the same anticoagulation regimen that has been widely published10 in >375 adult and pediatric mechanical circulatory support device patients as well as in the ECMO patients. We have seen very low incidences of thromboembolism in our CardioWest, HeartMate II, and ECMO populations, but we continue to be frustrated in the use of small pumps for small patients. This group of patients has much lower flows and hence more opportunity for stasis within the device or inflow and outflow tubing. We did have one patient with two device change-outs in this group and have had more in some of our other pediatric applications such as congenital postrepair and restrictive cardiomyopathy hearts. The relatively large device foreign body surface area for the body size of these patients may cause proportionally more inflammation and hypercoagulability than we see in adults. In addition, over the course of time, we have been struck by the frequency of low antithrombin III (AT III) levels in device-supported infants and children. This can prevent heparin from working and lead to either thrombosis or, in the face of increased heparin dosing, to hemorrhage. We are now prophylactically treating with an initial dose of AT III and then following levels closely. We have pushed our anticoagulation therapy to the point that 45% of patients have had take back operations for bleeding. This has not caused major long-term complications and may be preferable to under anticoagulation because stroke is so devastating.
Fortunately, infection was an annoyance but not a major problem. Pneumonia was the most common infection and may have been a consequence of pulmonary edema and prolonged intubation.
The return of ventricular function we have witnessed is remarkable. Seventy-five percent of those who recovered were explanted by day 16. Would this have happened without LVAD therapy? In three of the cases that initially failed ECMO, this question is clear. In three of the cases that initially failed ECMO, the answer is clear. We were strongly influenced by the initial intraoperative and subsequent “decompressed” echos that documented reduction in ventricular size and improvement in fractional shortening that was statistically significant. Whether the use of medical therapy helps recovery remains controversial. We were able to use it in this series in only four patients. We did, however, use dobutamine at 5 μg/kg/min for 24 hours before weaning trials. Short-term use of a beta-stimulant causes improved inotropic status and some hypertrophy that may enable the native hearts to optimize function.
In this small experience with infants and children with severe cardiac dilatation and systolic failure implanted with LVADs over a 6-year period, one must be cautious in generalizing. Much more work needs to be done, and other investigators need to validate our findings. We could be accused of being over enthusiastic, but it is true that 73% (8 of 11) did experience long-term recovery of their native hearts.
1. Birks EJ, Tansley PD, Hardy J, et al
: Left ventricular assist device and drug therapy for the reversal of heart failure. N Engl J Med
355: 1873–1884, 2006.
2. Birks, EJ, George RS, Hedger M, et al
: Myocardial recovery from advanced heart failure using the HeartMate II LVAD combined with drug therapy: Early results from prospective study. J Heart Lung Transplant
28: S304, 2009.
3. Frazier OH, La Francesca S, Demirozu ZT, et al
: Long-term survival after left ventricular assist device explantation. J Heart Lung Transplant
28: S304–S305, 2009.
4. Blume ED, Naftel DC, Bastardi HJ, et al
; Pediatric Heart Transplant Study Investigators. Outcomes of children bridged to heart transplantation with ventricular assist devices. Circulation
113: 2313–2319, 2006.
5. Stiller B, Lemmer J, Schubert S, et al
: Management of pediatric patients after implantation of the Berlin Heart Excor ventricular assist device. ASAIO J
52: 497–500, 2006.
6. Copeland JG, Arabia FA, Tsau PH, et al
: Total artificial hearts: bridge to transplantation. Cardiol Clin
21: 110–113, 2003.
7. Imamura M, Dossey AM, Prodhan P, et al
: Bridge to cardiac transplantation in children: Berlin heart versus extracorporeal membrane oxygenation. Ann Thorac Surg
87: 1894–1901, 2009.
8. Malaisrie SC, Pellitier MP, Yun JJ, et al
: Pneumatic paracorporeal ventricular assist device infants and children: Initial stanford experience. J Heart Lung Transplant
27: 173–177, 2008.
9. Stein ML, Robbins, R, Sabati A, et al
: Use of INTERMACS criteria to assess major clinical outcomes in children bridged to heart transplant using mechanical circulatory support. J Heart Lung Transplant
28: S207–S208, 2009.
10. Zimmerman H, Copeland JG, Aquila-Allen LA, Smith RG: Total artificial heart, in Selke FW, del Nido PJ, Swanson SJ (eds),: Sabiston and Spencer Surgery of the Chest
, Philadelphia, Saunders Elsevier, 2010, pp. 1525–1532.