Drews, Thorsten; Dandel, Michael; Krabatsch, Thomas; Potapov, Evgenij; Stepanenko, Alexander; Hennig, Ewald; Lehmkuhl, Hans Brendan; Pasic, Miralem; Weng, Yuguo; Hetzer, Roland
The first mechanical circulatory support (MCS) device was implanted in 1963 by De Bakey1; several years before, Barnard2 had performed the first heart transplantation. Six years later, Cooley et al.3 implanted a total artificial heart in a 54-year-old man as a bridge to heart transplantation. The initial idea of permanent support was abandoned due to the large number of complications such as infections and thromboembolic events. Heart transplantation gained more and more importance. However, in view of the shortage of donor organs, multiple pulsatile systems, physiologically comparable with the beating heart,4 were developed. With the voluminous first generation of implanted devices, pocket bleeding and infection were still common complications. After the introduction of the second generation of devices, which are smaller and have nonpulsatile flow (DeBakey left ventricular assist device [LVAD] in November 1998 and Jarvik 2000 in April 2000), these complications were reduced.5,6 With increased waiting time for a donor heart, the need of patients for long-term MCS increases. With the development of the third generation of MCS systems, with magnetic bearings (Berlin Heart Incor first implanted in June 2002, DuraHeart in January 2004), further progress has been made in minimizing the complications of thromboembolism and technical failure.7–10
Up until now a total of 1,618 MCS systems, 925 of them pulsatile (Berlin Heart Excor, Berlin Heart, Berlin, Germany), Novacor (WorldHeart, Salt Lake City, UT), CardioWest (SynCardia, Tucson, AZ), Abiomed (Abiomed, Inc., Danvers, MA), HeartMate I (Thoratec, Pleasonton, CA), LionHeart (Arrow-Teleflex, CA), and Bücherl Total Artificial Heart (Berlin Heart, Berlin, Germany) and 693 nonpulsatile (Berlin Heart Incor, Berlin Heart, Berlin, Germany), Levitronix (Levitronix, Inc., Waltham, MA), HeartMate II (Thoratec, Pleasonton, CA), DeBakey LVAD (Micromed, Inc., Huston, TX), HeartWare (HeartWare, Inc., Sydney, Australia), Impella (Abiomed, Inc., Danvers, MA), DuraHeart (Terumo Heart, Inc., Ann Arbor, MI), Jarvik 2000 (Jarvik Heart, Inc., New York, NY), VentrAssist (Ventracor, Chatswood, NSW, Australia), and CorAide (Arrow-Teleflex, CA), have been implanted in our institution. Today, with these developments, the question arises whether all MCS systems can provide the same good long-term results. The embolic and bleeding complications, infections, technical failures, and other complications in 198 patients supported by MCS for >1 year were analyzed.
Patients and Methods
This report is based on 198 patients presenting with catecholamine-dependent terminal heart failure who underwent implantation of a MCS device and were supported by it for at least 1 year.
Eighty-seven patients had a pulsatile system (group A) of the first generation (27 Berlin-Heart Excor biventricular assist device [BVAD], 12 CardioWest, 23 Novacor, 22 Berlin Excor LVAD, 2 LionHeart, and 1 HeartMate I), 43 a nonpulsatile MCS with mechanical bearings (group B) of the second generation (33 HeartMate II, eight Jarvik 2000, and two DeBakey) and 68 a nonpulsatile MCS with magnetic bearings (group C) of the third generation (60 Berlin-Heart Incor, 7 DuraHeart, and 1 VentrAssist). In group A, the causative disease was dilative cardiomyopathy in 62 patients, ischemic cardiomyopathy in 22, and cardiomyopathy (postpartum, toxic, and restrictive) in three. The mean age was 49 ± 13 (7–72) years (p = 0.035). In group B, 20 patients had dilative cardiomyopathy and 23 ischemic cardiomyopathy (p = 0.007). The mean age in this group was 53 ± 12 (26–81) years. In group C, 38 patients had dilatative cardiomyopathy and 30 ischemic heart disease. The mean age was 54 ± 11 (18–72) years.
The Jarvik 2000 and the DeBakey LVAD were implanted mostly in elderly patients, which is the reason why in group B there were significantly more patients with ischemic heart disease (p = 0.007). On the other hand, because of progress in medical therapy leading to aging of the population, with a still limited number of donor hearts, the devices of the second and third generation were implanted in patients of increased age. For this reason, the patients in groups B and C were significantly older than in group A (p = 0.035). In total, there were 178 male and 20 female patients, and there were no differences in gender between the groups (Table 1).
All patients with severe heart failure with the indication for MCS implantation were candidates for long-term support. The postoperative course, potential contraindications, the patients' own wishes, and the lack of donor organs contributed to the decision for long-term use.
Choice of Device
The Berlin Heart Excor, a paracorporeal system, was the first device used for long-term support. It can be implanted for left ventricular support only or, in severely ill patients with multiorgan failure, for biventricular support, which is the most important indication for this pump today. The paracorporeal positioning of the pumps reduces patients' quality of life, so that implantable devices have gained increasing importance for long-term support. Although the CardioWest is an implantable total artificial heart, the Novacor and the Heart Mate I are assist devices for the left ventricle only. Because of their pulsatile technique, they were very voluminous, making the preparation of a subcutaneous pocket necessary. The Lion Heart, the first left ventricular pulsatile system for permanent support only, was used in the years 2000–2004. Because it required a complex surgical procedure in patients belonging to the high-risk group, its use has since been abandoned.
Although pulsatile devices were used in the earlier period, the nonpulsatile systems have gained more importance during the past few years. The left ventricular assist devices implanted were systems for full left ventricular support, in contrast to the Jarvik 2000 LVAD, which is a device that only partially supports left ventricular function. One of the most frequently used nonpulsatile devices worldwide for left ventricular support is the HeartMate II, implanted since May 2006 at the Deutsches Herzzentrum Berlin. The Berlin Heart Incor and the DuraHeart are both nonpulsatile devices of the third generation. The Berlin Heart Incor, an axial flow pump, was implanted for the first time on June 16, 2002, and the DuraHeart, a centrifugal device, has been used since January 15, 2004. Although the left ventricular unloading and arterial flow are practically identical in these pumps, a milder anticoagulation regimen was adopted for the centrifugal devices and for the Heart Mate I and II due to the rough inner structure.
Today, if the candidate needs biventricular support, the Berlin Heart Excor or CardioWest is implanted. Using nonpulsatile devices, the HeartWare, is an alternative. For left ventricular support only, several devices are available. Nonpulsatile devices are preferred because of their compact construction, lesser surgical invasiveness, and potentially fewer long-term complications. The anticoagulation management is adapted to the device implanted (Table 2).
Data Collection and Statistical Evaluation
The postoperative course and the late outcome in these outpatients (time on device, time of hospitalization, rehospitalization, infections, bleeding, embolic complications, organ failure, and technical failure) were analyzed.
All data analyses were performed using SPSS, version 17.0 (SPSS, Inc., Chicago, IL). To analyze differences between the three groups in metric data, the Kruskal-Wallis test for categorical data, the χ2 test for Crosstabs, and for comparison of survival distribution the log-rank test was used. Differences between two groups in corresponding measurements were analyzed by the Mann-Whitney U test for two independent samples. A value of p < 0.05 was considered to indicate statistical significance.
Of a total of 1,618 MCS systems implanted in our center, 198 were applied in patients who were supported for >1 year. The mean time on support of the total group was ∼2 years (690 ± 321 [366–1,875] days, Table 3). The longest time of support was observed in three patients on Novacor (1,875, 1,815, and 1,811 days), in five patients on Berlin Heart Excor (LVAD: 1,836; 1,588; 1,508; and 1,400 days and BVAD: 1,609 days), and in one patient on Berlin Heart Incor (1,815 days). In total, 63 patients (32%) were supported for >2 years, 21 (11%) for >3 years, nine (5%) for >4 years, and two patients (1%) for >5 years.
In group A (first generation, pulsatile MCS), the mean time of support was 674 ± 366 (366–1,875) days (Table 3). Twenty-two patients (25%) were supported in this group for >2 years, 10 (11%) for >3 years, 7 (8%) for >4 years, and two patients (2%) for >5 years. In group B (second generation, nonpulsatile MCS), the patients were, with a mean of 606 ± 195 (373–1,225) days, not longer on MCS. In this group, 10 patients (23%) were supported for >2 years and two (5%) for >3 years. In group C (third generation, nonpulsatile MCS with magnetic bearings), the patients were, with a duration of 765 ± 310 (367–1,851) days, significantly longer on the device (p = 0.004). In this group, 31 patients (46%) were supported for >2 years, nine patients (13%) for >3 years, and two patients (3%) for >4 years (Figure 1).
The preliminary goal in all patients on a device is discharge from hospital. In group A, 95% of patients (83/87) could be discharged, in groups B and C 100%. In the mean, discharge home was 2 months (67 ± 67 [9–545] days) after device implantation (Table 4). Patients on pulsatile devices needed 3 months (94 ± 84 [18–545] days) and, therefore, significantly longer until first discharge than patients on the second or third generation of MCS (p < 0.05). The mean time spent at home was ∼1.5 years (530 ± 336 [0–1,709] days), corresponding to 77% of the total time on the device (Table 3). The patients with devices of the first generation (group A) were at home for a significantly shorter time (462 ± 367 [0–1,688] days) than patients of groups B and C (p = 0.001). In the mean, 2.8 readmissions were necessary per year (Table 5). The patients with HeartMate II had 1.7 readmissions/patient/yr and, thus, significantly fewer rehospitalizations than the patients on Berlin Heart Excor BVAD and LVAD (p = 0.024 and p = 0.021). Reasons for rehospitalization were mostly coagulation disorders (INR not in target range) (0.39 readmissions/patient/yr), followed by wound infection at the driveline exit (0.33 readmissions/patient/yr), technical failure (0.22 readmissions/patient/yr), and cerebral embolism (0.17 readmissions/patient/yr) (Table 5). The patients on the second generation of MCS (group B) had significantly fewer rehospitalizations due to coagulation disorders (0.15 readmissions/patient/yr; p = 0.021) or wound infections (0.24 readmissions/patient/yr, p = 0.017) than patients on pulsatile devices (Table 4).
The last rehospitalization took place in the mean 51 ± 69 (1–298) days before heart transplantation, weaning from the device or death and did not differ between the three groups (Table 4).
Bleeding and Embolic Complications
Although device-dependent anticoagulation management has been introduced, bleeding and embolic complications occurred in all groups of patients. In 20 patients (10%), a stroke; in two patients (1%), a prolonged ischemic neurologic deficit; and in 27 patients (14%), a transitory ischemic attack was observed. In total, 0.17 readmissions/patient/yr due to cerebral embolism were necessary (Table 5). Therefore, embolic complications were significantly more frequent in patients with pulsatile devices (0.69 readmissions/patient/yr; p = 0.017). For patients with nonpulsatile devices of the second and third generation of MCS (groups B and C), cerebral embolism occurrence was 0.24 and 0.32/patient/yr, respectively. Cerebral bleeding occurred in six patients (3%), corresponding to 0.019/patient/yr in all groups, whereas in patients with pulsatile devices, (group A) it occurred in 0.01 patients/yr, in group B in 0.05 patients/yr, and in group C in 0.01 patients/yr (without significance).
Gastrointestinal bleeding occurred in 0.1 patients/yr and epistaxis in 0.021 patients/yr. As standard therapy, in cases where the INR is within the target range, the administration of acetylsalicylic acid has been definitively stopped.
Local wound infection at the driveline exit is one of the most important potential complications in patients on MCS. In the total group of patients, 62% (123/198) (Table 5) had wound infections in the mean 1 year after device implantation (366 ± 152 [range, 14–1,554] days). This corresponds to a potential risk of 0.17/patient/yr, whereas the patients in group A had a significantly higher risk of wound infection (p = 0.006). Most infections were caused by Staphylococcus aureus, followed by Corynebacterium species, Pseudomonas aeruginosa, methicillin-resistant Staphylococcus aureus (MRSA), and Enterococcus faecalis (Figure 2). Antibiotic therapy was administered early in the case of local infection, even if no signs of systemic inflammation were seen.
Complications caused by the device were observed with all types of devices (Table 5). These complications were due to thrombotic deposits in the device, pump and controller failure, damaged cannula, and broken transcutaneous leads. In the mean, 32% of patients had technical failure, whereas patients on HeartMate II and on Berlin Heart Incor had significantly fewer complications (p < 0.05). Technical failure was seen more often with biventricular paracorporeal devices (Berlin Heart Excor); this system is especially constructed to allow pump exchange in case of deposits and cannula shortening in case of damage.
All potential types of technical failure have been observed: 12 patients with the Berlin Heart Excor needed pump exchange, in 10 due to deposits, and in two due to mechanical valve failure. One patient had a lacerated transcutaneous cannula, and in one patient, the driving system smoked, caused by water damage. With the CardioWest, one patient had lung edema due to pump failure and another patient needed cannula repair due to air leakage. Damaged transcutaneous electrical leads were observed in patients on DuraHeart, Berlin Heart Incor, and on Novacor. In seven patients, a bearing problem occurred. For this reason, five patients needed device exchange (2 Novacor, 1 Berlin Heart Incor, 1 DuraHeart, and 1 VentrAssist), and in two patients (Berlin Heart Incor), high urgent heart transplantation became necessary.
Survival and Causes of Death
The 1-year survival of the patients on MCS of the first generation (group A) was 71%, the 2-year survival 58%, and the 3-year survival 46% (Figure 1). The survival of the patients on second and third generation of MCS (groups B and C) at 1 year was 81% and 82% respectively, and at 2 years 65% and 62%, respectively, and at 3 years 49% for the patients in group C. There were no statistically significant differences (p = 0.39). In the first group, 32% of patients died (mean time until death: 791 ± 485 [366–1,855] days), and the causes of death were mostly septicemia (54%) followed by intracranial bleeding (14%), stroke (14%), carcinoma (7%), and others (11%) (Figure 3). Nineteen percent of patient in group B died (mean time: 600 ± 246 [376–1,103] days), and the causes were septicemia (63%), intracranial bleeding (25%), and pump thrombosis (12%). In the third group (group C), 25% of patients died (mean time: 690 ± 242 [369–1,283] days); causes of death were septicemia (41%), followed by intracranial bleeding (29%), suicide (12%), pump thrombosis (6%), carcinoma (6%), and right heart failure due to ventricular fibrillation (6%). There were no statistically significant differences.
In all three groups, patients underwent heart transplantation. In the first group, 51% of patients received an organ transplant (after a mean waiting time of 608 ± 281 [374–1,811] days), in the second group 9% (mean waiting time: 455 ± 86 [373–576] days), and in the third group 32% of patients (mean waiting time: 642 ± 235 [367–1,154] days).
Only in group A could patients be weaned from the device (2%). At present, 15% of patients in group A, 72% in group B, and 32% in group C are still on support.
So far at the Deutsches Herzzentrum Berlin, 1,618 MCS systems, 925 of them pulsatile and 693 nonpulsatile devices, have been implanted. In 198 patients, the systems were used for >1 year, and 194 patients could be discharged from hospital to live with their families. Progress in medical therapy has led to increasing aging of the population, while the number of donor hearts remains limited. For this reason, long-term MCS is becoming of increasing importance.
In this study, the time on MCS, the rehospitalizations, the embolic and bleeding complications, infections, technical failure, and other complications of all 198 patients with MCS for >1 year were analyzed. Three groups of patients, according to the generations of devices available, were studied. Thanks to experience with the devices of the first generation, the need for less blood traumatic devices resulting in lower anticoagulation could be shown. Additionally, smaller sizes of MCS may reduce the perioperative risk and the later risk of pocket infections. Therefore, smaller devices were developed (second and third generation). In an earlier publication,11 the advantages of nonpulsatile devices were demonstrated. Elderly patients showed an increasing long-term survival using this new technology. Whether this applies to all patients on long-term MCS has been the current question.
In this study, although the patients with pulsatile devices (group A, first generation of MCS) were significantly younger, the results were compared. The mean time on support was ∼2 years, whereas patients with MCS of the third generation had the longest support time (p = 0.004).
The number of patients weaned from the device was low. The main reason may be that ventricular recovery is achieved during the first months on MCS.12 In earlier reports, the potential advantages of pulsatile devices in terms of better unloading of the left ventricle and reduction in the transpulmonary gradient were disproved,13 and the efficacy of nonpulsatile devices, especially of the HeartMate II, has been shown in several studies.14–16
In the mean, 2.8 rehospitalizations/patient/yr were necessary. The high number of rehospitalizations was caused primarily by coagulation disorders, wound infections, technical failure, and cerebral embolism. Cerebral embolism was rarely lethal, but infections leading to septicemia, cerebral bleedings, and pump thrombosis were the most important reason for death. Patients on HeartMate II had significantly fewer readmissions. This may be explained by the lesser need of anticoagulation leading to fewer anticoagulation disorders and by the more flexible transcutaneous cannulas leading to less wound infection at the exit side.
One of the greatest advantages of rotary pumps is their smaller size, which requires a less invasive surgical procedure and less intensive anticoagulation, allowing faster recovery. Nevertheless, an antecedent study showed that nonpulsatile devices have significantly more technical failure.11 In this study, we found that technical failure occurred with all devices. Further technical improvements are still important. Not only noninvasive power transfer through the intact skin—as used for the LionHeart—but also improved pump technology leading to less thrombosis formation is necessary.17,18
Limitations of the Study
The patients in group A were younger, and 45% (39/87) had a biventricular device. The devices in this group were implanted in an earlier period (1994–2009). The implantation of the nonpulsatile devices was later, between 1999 and 2009. The HeartMate II has been implanted since 2006 only. As 27 patients are still being supported by this device, future analysis may show better results in the group of patients with devices of the second generation. Additionally, the learning curve during recent years in the care of these patients on MCS may have contributed to the better results in patients with devices of the second and third generation of device.
In the past, the implantation of a MCS has been reserved for potential candidates for later heart transplantation only. This study aimed to show the differences in and challenges of the different assist devices available on the market for long-term use.
It was shown that the patients on all 12 different devices—pulsatile and nonpulsatile—still have limited life expectancy and need frequent rehospitalizations. The most critical risk factors are infections, coagulation disorders (including embolic and bleeding complications), and technical failures.
The product life cycle of these important lifesaving devices is still in the introductory phase. Further technical improvements are necessary to allow growth on the market and broader acceptance by patients and physicians. Close outpatient monitoring and support are crucial to ensure good long-term results.
The authors thank Ms. Anne Gale, Editor in the Life Sciences, for editorial assistance.