| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2002;74:2051-2063
© 2002 The Society of Thoracic Surgeons
a Thoracic and Cardiovascular Surgery, Kaufman Center For Heart FailureCleveland, Ohio , USA
b Department of Biostatistics and Epidemiology, The Cleveland Clinic Foundation, Cleveland, Ohio, USA
* Address reprint requests to Dr Navia, Department of Thoracic and Cardiovascular Surgery, The Cleveland Clinic Foundation, 9500 Euclid Ave, Desk F25, ClevelandOH44195, USA.
e-mail: naviaj{at}ccf.org
Presented at the Thirty-eighth Annual Meeting of The Society of Thoracic Surgeons, Fort Lauderdale, FL, Jan 28–30, 2002.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
METHODS: From December 1991 until January 2002, 264 patients received 277 LVADs for BTTx. We analyzed temporal trends in pre-LVAD patient factors and device-specific time-related complications.
RESULTS: Survival to transplant was 69%. Adverse event analysis demonstrated a high risk of infections (0.56, 1.28, and 1.88 per patient at 30 days and 3 and 6 months). HeartMate devices were more prone to infection than Novacor devices (p < 0.0001). Cerebral infarctions occurred less commonly than infections (0.15, 0.25, 0.30 at 30 days and 3 and 6 months), were more common in Novacor than HeartMate (p = 0.0001), and were decreased by the new Novacor Vascutek conduit (p = 0.07), but these were still slightly higher than the HeartMate (p = 0.04). Device failures occurred in 21 instances (all but one were in HeartMate devices [p = 0.04 vs Novacor]), but have significantly decreased (p < 0.0001) in HeartMate since 1998.
CONCLUSIONS: Infections and device durability limit the chronic use of the HeartMate device, but device failures are decreasing. Novacor has fewer problems with infection and durability, and the new Vascutek conduit will reduce, but not eliminate, strokes.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
The past decade has witnessed tremendous advances in the two FDA-approved implantable LVAD systems (Thoratec HeartMate, Pleasanton, CA, originally TCI, Woburn, MA; and Worldheart Novacor, Ottawa, Canada, formerly Baxter-Novacor, Oakland, CA). The original pneumatically driven HeartMate (1000IP) was rede-signed to a portable, battery-powered version (HeartMate VE). The Novacor, which was always electrically powered, replaced the bulky external console with a portable controller and batteries. Both devices then af-forded the patient the freedom of hospital discharge and extended periods of tether-free existence, which greatly improved patient QOL. Furthermore, the Novacor redesigned the inflow valve assembly and changed the inflow graft to a gelatin-coated knitted material (Vascutek; Sulzer Carbomedics, Austin, TX).
Recently, the first randomized trial of permanent LVAD implant (HeartMate VE) compared with optimal medical therapy (the REMATCH trial) was published [4]. The trial showed a statistically significant increase in survival for the LVAD-treated patients but was accompanied with frequent device-related complications. Therefore, we analyzed our LVAD BTTx experience in regards to: (1) morbidity and mortality; (2) temporal trends; (3) device type; and (4) risk factors for death. The purpose of the study was to determine whether BTTx experience is similar to the REMATCH results. In particular, we focused on three major morbidities during device support (infection, stroke, and device failure), all of which have implications for patient QOL and cost of therapy. We did not attempt to assess QOL or cost directly in this report.
|
Patients and methods |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
Patient characteristics, pre-LVAD medical and temporary ventricular support, and LVAD device use are presented in Table 1.
|
Our experience began with the pneumatic HeartMate, but rapidly switched over to the HeartMate VE (which was only intermittently available when investigational). The Novacor device was implanted almost exclusively from 1996 until 1998, but has not been used since that time. From that point on, the HeartMate VE was used exclusively.
All patients receiving HeartMate devices were given aspirin only (325 mg) for anticoagulation. In some patients, it was held due to gastritis. On occasion, HeartMate patients were treated with warfarin for other reasons) (eg, atrial fibrillation or deep vein thrombosis). The Novacor patients were managed with heparin (after the perioperative bleeding stopped), aspirin, and frequently other antiplatelet agents before transition to chronic warfarin, with a target International Normalized Ratio of 2.5 to 3.5.
Data analysis
Temporal trends
Temporal trends for discrete variables were identified by logistic regression using a set of power, inverse, and logarithmic transformation of LVAD implant time. The transformation(s) of time selected was guided by decile analysis, in which we sought the best linearized transform for prevalence expressed on the logit scale. For continuous variables, linear regression was employed.
Time-related complications
Three types of complications during LVAD support were considered: (1) infections, (2) neurologic events, and (3) LVAD failure. The major causes of device failure are inflow conduit bleeding, percutaneous driveline fracture, and failure of the electrical system or mechanical system. We considered the situation a device failure if the HeartMate VE system failed and the patient had to be actuated using the pneumatic console. Although this was not a "catastrophic" device failure, as defined by the Circulatory Support Task Force (Bethesda Conference 1995) [5], we considered it a failure because the V-E patient had to be readmitted and could not be discharged, and there was always a major change in quality of life (ie, sense of security) for the patient and family. Infections were subcategorized as (a) blood stream, (b) pump pocket, (c) driveline, and (d) intradevice. Neurologic events were subcategorized as (a) cerebral bleed or (b) cerebral infarct (presumed embolic). We made no attempt to determine the source of the cerebral infarct. For instance, the embolic source could have been device-related thromboemboli, septic emboli, or emboli from the patients’ left ventricle, left atrium, or vascular system.
A description of the statistical methodology including competing risks, analysis of benefits of bridging, and presentation are provided in Appendix 1.
|
Results |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
The crux of the issue now is what adverse events are attributed to chronic mechanical circulatory support, and how those adverse events can be mitigated.
In the REMATCH trial, the frequency of serious adverse events (SAEs), predominantly infection, bleeding, and device failures, was 2.35 times as high as the medically treated group. Based on the BTTx experience, a high rate of SAEs could be predicted. In this report and a recent extensive review, infection was the most common SAE, especially in the HeartMate device [20, 21]. In particular, the pneumatic HeartMate pumps had a high risk for pump pocket infection. This may reflect that these were our earliest implants and the driveline tunnel was typically only 2 to 4 inches long in most patients. The driveline exit site was 8 to 12 inches away from the pump pocket in the VE and Novacor devices. Eventually, the pneumatic HeartMate driveline was modified to allow a long tunnel. A localized driveline infection is less likely to track along a well-healed long tunnel and infect the pump pocket. Nevertheless, in our opinion, infection remains the largest unresolved obstacle to more widespread use of permanent LVAD support [19]. It is a major source of morbidity and mortality, and accounted for 41% of the LVAD patient deaths in REMATCH [4]. More aggressive investigation of modified percutaneous drivelines (such as silver-impregnated coverings or a bone-mounted pedestal), implantable systems with transcutaneous energy transmission systems, or partially implanted systems (such as a compliance chamber instead of a vent line with a small driveline, such as are used in axial flow pumps) should reduce device infection. Also, intraabdominal implantation, anecdotally, may decrease the risk of infection.
Device failure was another SAE in REMATCH and our BTTx experience. Whereas we demonstrated a 17% risk of HeartMate failure by 12 months, there were none in REMATCH. However, by 24 months, the probability of device failure was 35% in the REMATCH study. We are encouraged that the frequency of device failure is less common in our more recent HeartMate experience, but few of our patients have been on support for 1 year or greater. As expected from bench testing, reliability and durability of the Novacor was much better, with only one failure in our experience, and that was more than 1 year after implant.
A particularly devastating complication is permanent stroke. Whereas infections may be suppressed or cleared, and device failures may be managed by pump exchange or use of the pneumatic driver for HeartMate VE failure, a stroke patient may pose a difficult dilemma. Some patients may become permanently impaired, reducing their QOL (and that of their caregivers), and these BTTx patients may no longer be considered candidates for transplantation. Despite intense anticoagulation, the original Novacor we used, with a woven inflow graft, had an unacceptably high stroke rate. Fortunately, the Vascutek conduits reduced the overall stroke rate, especially beyond 30 days, but overall, the Novacor Vascutek stroke rate was still higher than for HeartMate devices. In the REMATCH LVAD group, the rate of ischemic stroke was 10% [4]. Projecting from our limited number of Novacor Vascutek patients, the rate of stroke would most likely be higher in a trial of permanent implants.
Currently, the Novacor is being studied in a nonrandomized trial (INTREPID) for permanent implant. REMATCH clearly demonstrated, in a randomized trial, the dismal prognosis and QOL of patients with end-stage heart failure [4]. In our opinion, it may be unnecessary and perhaps unethical to randomize patients like this again. Instead, we should focus on improving the outcome of device support. Based on our observations in BTTx, we would expect the INTREPID trial would show a much higher reliability and durability, and lower rate of infection, than the REMATCH trial. Whether the risk of stroke will be similar to REMATCH remains to be determined. The INTREPID trial should use Novacor pumps with Vascutek conduits. However, the QOL and cost of therapy implications of better durability and lower infections should be significant.
A host of new mechanical circulatory support devices are beginning or will soon begin clinical use. Most will start with BTTx use, but the Lionheart LVAD and Abiocor total artificial heart began with permanent implants. The wisdom of this decision will be tested as patients with serious adverse events, to be expected in this early phase of the technology, are exposed to prolonged support and the opportunities to improve device design in the clinical laboratory of BTTx are limited. Also, for pumps that are being used for BTTx trials, permanent implants should be postponed until the device-related SAEs are lower in the BTTx experience.
Limitations
There are several inherent limitations when comparing BTTx with permanent LVAD use outcomes. First, the BTTx group is generally younger (55 ± 11 years in our experience vs 66 ± 9.1 for REMATCH) [4]. The older patients in the nontransplant trial may have more comorbidities, and have a more difficult time recovering and rehabilitating from LVAD implant. On the other hand, the BTTx group are generally more acutely ill, with 77% of our patients on an intraaortic balloon pump, 56% intubated, and 22% on other circulatory support devices, with a median 5-day intensive care unit stay before LVAD implant. These acutely decompensated patients are more prone to early multiple organ failure, and may account for the high early risk of bloodstream infection. Most importantly, the BTTx temporary experience should underestimate late device failure/durability issues. Extended ex vivo bench testing probably gives the best estimate of late device failure. For instance, in REMATCH, device failures did not occur until beyond 12 months.
In conclusion, our BTTx experience predicted a significant rate of device-related infections, failures, and cerebral embolic events. These SAEs varied by device, and were modified by device changes. The same SAEs appeared in REMATCH and limit the success of permanent LVAD use with the technology employed in that trial.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
|
Appendix 1 |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
In addition, each event was considered as a modulated renewal process to determine if "complication begets complication." [8]
Competing risks
The earliest occurrence after first LVAD placement of one of three mutually exclusive outcomes was identified: (1) death, (2) transplant, and (3) removal of device after recovery. The common interval of analysis was either the interval between first LVAD placement and the earliest occurrence of one of these outcomes, or the duration by December 31, 2001 of being alive with the LVAD in place (censored).
Freedom from each event was estimated by the nonparametric product-limit method (formula 4.4.2 of Andersen and associates [9]). Variances of the estimates were based on the Greenwood formula (formula 4.4.19 of Andersen and associates [9]). Asymmetric confidence limits were calculated with the use of these variances and formula 6E-6 in Kirklin and Barratt-Boyes [10]. The instantaneous risk (hazard function) for each competing event was estimated by a parametric method (available on the Internet at http://www.clevelandclinic.org/heartcenter/hazard) that resolved the number of hazard phases, identified the shape of the hazard function, and estimated its values [6]. The width of the confidence limits for estimates calculated from the resulting equations was consistent with that for nonparametric estimates.
Consequences of the independent, simultaneously operative transition rates (hazard functions) from the category "Alive With LVAD in Place" into each of the event categories were calculated by integrating the parametric equations [11]. These calculations represent a time-related synthesis of the individual events.
Variables examined multivariably for each event are listed in Appendix 2. Multivariable analyses were conducted independently in the multiphase hazard function domain [7] for each of the three competing events to generate parsimonious equations. Both a guided technique of entry of variables into the multivariable models [13] and bootstrap bagging of 1,000 resamplings were used [14]. A p = 0.05 criterion, with 50% bootstrap reliability, was required to retain variables in the model.
Benefit of bridging
To assess the survival benefit of the total program of using LVADs as a BTTx, overall all-cause mortality was analyzed from the time of first LVAD implantation. Thus, total follow-up, including death before transplant and death after transplant, was considered. Nonparametric estimates of survival were obtained by the Kaplan-Meier method, and parametric estimates were obtained by a multiphase hazard function method.
Multivariable analysis focused on pre-LVAD risk factors (Appendix B). It used the variable selection methods described above under "Competing Risks."
Presentation
Regression coefficients are presented plus or minus 1 SE. These are presented rather than hazard ratios because the models, and the underlying data, were not proportional hazards across time. Confidence limits (CL) of parametric and nonparametric estimates are 68%, equivalent to plus or minus 1 SE.
|
Appendix 2 |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
Hemodynamics
Right atrial pressure, pulmonary artery pressures (systolic, diastolic, mean, capillary wedge), cardiac index
Etiology of myopathy
Ischemic, idiopathic dilated
Comorbidity (within 7 days of implantation)
Serum creatinine, total bilirubin, occurrence of ventricular tachycardia or fibrillation, infection
Temporary support pre-LVAD (within 3 days of implantation)
Intraaortic balloon pump, extracorporeal membrane oxygenation, ABIOMED, intubation, inotropic agents
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
I stress this because the early problems, particularly with the HeartMate, were related to the driveline exiting at the 6 o’clock position. Pat McCarthy trained with Phil. When Pat began using the HeartMate pump, he placed it in the extraperitoneal position, which allowed a large amount of blood to collect around a moving foreign body, contaminating the stiff driveline, which exited at the 6 o’clock position in the hypogastrium near the pocket. This positioning has since been changed, both with the pneumatic and with the electrical device. So, I think the problem with infection was much more a surgical issue of contamination with this large bulky device rather than a device issue.
Another interesting fact about the TCI is that the device failures occurred within a short period of time in 1995 and 1996. When the TCI was approved in 1994, additional personnel were hired to manufacture the device. Most of the problems were traced to this expansion. For example, the problem with inflow valve erosion was traced to a new person at Medtronic. Flaws caused by new machinists were corrected in the electrical pump. Now that these problems have been solved, we hope that there will be fewer long-term complications, particularly infection.
The work presented by this group is outstanding, because it shows us that excellent results can be obtained for these terribly sick patients. The overall registry for the HeartMate shows a 70% survival, not including ECMO patients. In the Cleveland Clinic experience, there is a 70% risk with improved 1-year survival posttransplant of nearly 90%, a number most centers meet.
I would like to ask Dr Navia how patients are selected for ECMO. There were 51 patients in this series who were on ECMO, so these were very high-risk patients. How these patients are converted from ECMO to an LVAD would be of interest. With the gravity-fill inlet cannula of the Novacor, it would be interesting to know the duration of follow-up with these patients and if the stroke rate continues over time.
The anticoagulation regimen that we use with the HeartMate has always been aspirin and Persantine. What anticoagulation regimen was used in the Heartmate patients who had cerebral bleeding?
These devices continue to improve. With the limitations of transplant, I think assist devices will soon be used long-term, which was, by the way, their original intent. Thank you for the opportunity to review this paper.
DR MATTHIAS LOEBE (Houston, TX): I would like to get some further information on how you address these three issues of device-related complications: infection, anticoagulation, and device failure. How did this affect your ability to transplant patients? Did you transplant patients with device failures on an emergency basis? Dr Frazier raised the issue of anticoagulation in the Novacor patients. As you well know, in the European experience, the thromboembolic rate has been reduced to less than 10% with a modified inflow graft as well as more delicate ways of anticoagulation. So I would be interested if you have undertaken any attempts to improve your anticoagulation treatment in these patients because your thromboembolic event rate was extremely high.
DR NAVIA: Dr Frazier, thank you very much for your comments. In our experience, there is a clear association with the increasing risk of pump pocket infection and the earliest pneumatic Heartmate implant. This was related to the proximity of the pump and the exit site of the driveline. The driveline tunnel is typically only 2 to 4 inches long and comes out on the left hypogastrium. As soon as the patient starts to mobilize, the driveline starts to loosen around the skin incision, becoming more apt to contamination.
On the other hand, both the Electric Heartmate and the Novacor have a driveline exit site that is 8 to 12 inches away from the pump pocket on the right hypogastrium. Therefore, a localized driveline infection is less likely to tract along the newly healed long tunnel and infect the pump pocket.
Another interesting point to address is that both Heartmates show a greater significant incidence of bloodstream infection than Novacor. Some sources of infection may include preexisting vascular lines, urinary catheter, prolonged ventilatory support, and also malnourishment, which can increase susceptibility to infection. However, there is a growing evidence that LVAD implantation leads to defects in cellular immunity secondary to an aberrant state of T-cell activation. These defects may predispose LVAD recipients to candidal and other systemic infections. In our opinion, infection remains a large and unresolved problem, and is a major source of morbidity and mortality of patients on LVAD support.
In terms of device failure, we have experienced more incidents with Heartmate than Novacor. The most common problem was inflow conduit erosion, driveline leak fractures, and electrical problems. However, since Heartmate made the changes, the incidence of failures decreased dramatically.
In our experience, the magnitude of embolic events was substantially different depending on the devices. Novacor has a significantly higher cumulative number of embolic events per patient than the Heartmate, even though all patients receive an intensive anticoagulation regimen. Heartmate, on the other hand, has a similar extremely low rate of embolic event across time. Since FDA approval was granted, the Novacor inflow conduit has been shorted and changed to a gelatin-seal graft (Vascutek), reducing the incidence of stroke rate, especially beyond 30 days. However, the overall Novacor Vascutek stroke rate was higher than Heartmate.
Temporary mechanical support devices like ECMO or Abiomed have been used to bridge seriously ill patients in profound cardiogenic shock after myocardial infarction or postcardiotomy. In our institution, patients presented in this kind of situation are supported with ECMO for 48 to 72 hours, and this time on support gives us the opportunity to perfuse vital organs and stabilize the patient. Sometimes, the heart recovers and the patient is weaned from the short-term mechanical support. If the heart function has not improved, and the patient is neurologically intact and a candidate for transplantation, then the patient is switched to a more complex support system of either the Heartmate or Novacor until a donor heart is available.
I would like to address Dr. Loebe’s question. First, we do not transplant the patient with a device failure on an emergency basis. We prefer to change or replace the device if there is a need, and we will then perform the heart transplantation in a more stable clinical condition. In terms of anticoagulation therapy during the Novacor support, we start giving heparin a few hours after surgery to achieve a partial thromboplastin time (PTT) of 50 to 70 seconds or an activated clotting time (ACT) of 180 to 200 seconds. Subsequently, warfarin is added to keep the International Normalized Ratio between 2.5 and 3.5, along with the aspirin or dipyridamole. In contrast, patients with either of the Heartmate LVADs do not require anticoagulation therapy unless they have a concomitant condition, such as atrial fibrillation, for which they are given warfarin. Most Heartmate patients are given aspirin, 325 mg a day. Patients with the Novacor device were on full anticoagulation therapy, and the difference of the thrombolism rate experienced by devices may be related to the interior surface of the devices. In our experience, the Novacor originally corrugated Dacron inflow graft frequently showed an extensive thrombus and a poorly adherent pseudointima formation at the time of the LVAD explant. Since the company changed for a vascutek graft, we have never experienced the thrombus formation on the graft, and the thrombolism rate significantly decreases.
Both implantable LVADs provide a good hemodynamic support and a good quality of life for patients awaiting heart transplantation. Although Novacor originally posed a higher risk of thromboembolic events, recent design changes have reduced the risk.
Unfortunately, LVADs carry a significant risk of infection, because of preexisting illness, vulnerable percutaneaus connection, and immunologic deficiencies related to prolonged support. Patients who develop driveline infection or device infection may be able to undergo heart transplantation, but the infection remains a serious limitation to widespread long-term LVAD support. Chronic device infection would decrease quality of life, and will increase the overall cost therapy. This may require pump replacement, with the consequent negative impact in clinical outcome. The devices are not perfected yet, in spite of the important device changes performed lately. The technology has to continue to evolve, and the better understanding of the problems and prompted modifications of the devices will produce a more reliable circulatory assist device for permanent implants, as a gold standard alternative to transplantation for older and sicker patients.
|
References |
|---|
|
TopAbstractIntroductionPatients and methodsResultsCommentAcknowledgmentsAppendix 1Appendix 2DiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. Joshi, L. S. Magder, Z. Kon, S. Kallam, M. Kwon, R. Sangrampurkar, R. Pierson, and R. Poston Association between prothrombin activation fragment (F1.2), cerebral ischemia (S-100{beta}) and international normalized ratio (INR) in patients with ventricular assisted devices Interactive CardioVascular and Thoracic Surgery, June 1, 2007; 6(3): 323 - 327. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Pavie, N. Reiss, and S. Aubert Implantation technique of the Novacor left ventricular assist device MMCTS, February 19, 2007; 2007(0219): 1529. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Schenk, A. El-Banayosy, W. Prohaska, L. Arusoglu, M. Morshuis, W. Koester-Eiserfunke, L. Kizner, E. Murray, P. Eichler, R. Koerfer, et al. Heparin-induced thrombocytopenia in patients receiving mechanical circulatory support J. Thorac. Cardiovasc. Surg., June 1, 2006; 131(6): 1373 - 1381. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Schenk, L. Arusoglu, M. Morshuis, K. Minami, P. Sarnowski, R. Koerfer, and A. El-Banayosy Triple Bridge-to-Transplant in a Case of Giant Cell Myocarditis Complicated by Human Leukocyte Antigen Sensitization and Heparin-Induced Thrombocytopenia Type II. Ann. Thorac. Surg., March 1, 2006; 81(3): 1107 - 1109. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Schenk, P. M. McCarthy, E. H. Blackstone, J. Feng, R. C. Starling, J. L. Navia, L. Zhou, K. J. Hoercher, N. G. Smedira, and K. Fukamachi Duration of inotropic support after left ventricular assist device implantation: Risk factors and impact on outcome J. Thorac. Cardiovasc. Surg., February 1, 2006; 131(2): 447 - 454. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. C. Dang, V. K. Topkara, B. T. Kim, M. L. Mercando, J. Kay, and Y. Naka Clinical outcomes in patients with chronic congestive heart failure who undergo left ventricular assist device implantation J. Thorac. Cardiovasc. Surg., November 1, 2005; 130(5): 1302 - 1309. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Doi, L. A. R. Golding, A. L. Massiello, M. W. Kopcak Jr, R. L. Gerhart, S. Schenk, M. Inoue, Y. Ootaki, and K. Fukamachi Preclinical readiness testing of the arrow international CorAide left ventricular assist system Ann. Thorac. Surg., June 1, 2004; 77(6): 2103 - 2110. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||
