Left Ventricular Assist Device Infection Is Associated With Increased Mortality but Is Not a Contraindication to Transplantation
Background Left ventricular assist devices (LVADs) are increasingly used as a bridge to transplantation. Infection is a frequent and major complication associated with the use of these devices; however, the correlation of infection and outcome has not yet been evaluated in a prospective fashion.
Methods and Results Twenty-five patients (24 male, 1 female) with end-stage cardiac failure and resulting organ dysfunction were included. Patients were bridged with the Novacor N100 portable LVAD (median duration of support, 55 days) and were evaluated prospectively by device surface cultures on explantation, molecular typing of isolates, and correlation of infection with survival to transplant. Twelve (48%) of 25 patients had LVAD infection as defined by recovery of multiple isolates of identical genotype from the device surface. Whereas only 5 (42%) of 12 patients with LVAD infection survived until transplantation, 11 (85%) of 13 patients without infection were successfully transplanted (P<.05). Death of the 7 patients with proven LVAD infection was associated with multiple organ failure or other signs of acute infection.
Conclusions LVAD infection is associated with a significantly decreased survival probability. It does not preclude successful bridging but rather may pose an indication for urgent transplantation.
Recent estimates indicate that 30 000 to 60 000 patients in the United States each year are candidates for heart transplantation or permanent mechanical circulatory support,1 yet increasing demand for donor organs is confronted with decreased availability.2 Consequently, left ventricular assist devices (LVADs) are increasingly used to bridge the time span from end-stage heart failure to transplantation.3 4 Although permanently implanted LVADs may become available in the future as an alternative to transplantation, infectious complications in the present externally driven assist devices have been a major cause of morbidity and mortality in supported patients.5 Factors such as critical patient status, a driveline wound allowing for ascending bacterial colonization, and a large intravascular polymer surface may all contribute to infection. To evaluate the impact of infection on survival, we prospectively collected microbiological specimens from 25 patients supported with the Novacor N100 portable device, cultured the device surfaces on explantation, applied molecular typing methods to determine strain identity, and correlated these findings with survival to transplant as the primary end point.
Between February 1993 and August 1996, 32 patients received a Novacor N100 (Baxter Healthcare Inc, Novacor Division) portable LVAD powered through an electric, vented, and tunneled driveline. The use of this LVAD for circulatory support has been approved by the Ethics Committee of our institution (approval No. LVAS 141292), and informed consent was obtained from all participating patients. Devices were explanted either at the time of transplantation or, under surgically sterile conditions, at a maximum of 1 hour after the patient died. Swabs were taken from 12 predefined inner and outer surface sites of the device, the device pocket, and the conduit grafts and cultured by use of routine methods. Other specimens were obtained on suggestion of infection of a site. Pulsed-field gel electrophoresis (PFGE) was performed as previously described.6 Briefly, bacteria were immobilized in agarose beads, then lysed. Total genomic DNA was restricted with the use of Sma I, and DNA fragments were subjected to clamped homogeneous electric field electrophoresis. PFGE patterns with differences in the band pattern explainable with ≤1 “single genetic event” were considered identical.
Clinical and microbiological characteristics of patients enabled us to classify patients with respect to LVAD infection according to the following definitions: (1) LVAD not infected: sterile explant culture swabs or growth of pathogens different either by conventional or molecular typing and no signs indicative for systemic infection and (2) LVAD infected: bacterial growth of pathogens from ≥2 nonadjacent sites proven identical in PFGE, with or without clinical signs of infection. Kaplan-Meier life table analysis7 was used to calculate event-free survival, and the log-rank test was used to compare survival in different groups.
Of a total of 32 patients, 3 patients are currently supported by an LVAD, and explant culture results obtained according to the study protocol were available in 25 of 29 patients. The mean age was 45 years (95% CI, 40 to 50 years). The underlying diagnosis was ischemic cardiomyopathy (11 cases), idiopathic dilated cardiomyopathy (12), tetralogy of Fallot (1), and postpartum cardiomyopathy (1). All patients were in New York Heart Association class IV. The mean preoperative stay in the Intensive Care Unit was 20 days (CI, 9 to 31 days). Ten patients had interventions <10 days before LVAD implant (coronary artery bypass graft [7 patients], cardiopulmonary resuscitation , and aortic arch replacement ). Most patients had preexisting organ failure (renal [23 patients], hepatic , and pulmonary ). The mean preoperative cardiac index was 1.7 L·min−1·m−2 (CI, 1.5 to 1.9 L·min−1·m−2). Preoperatively, all patients received intravenous inotropic treatment and 6 received temporary extracorporeal circulatory support (IABP, ECMO). Prophylactic antimicrobials were administered only in the perioperative period (cefamandole) or during mechanical ventilation (oral formula containing polymyxin B, tobramycin, and amphotericin B). The mean duration of LVAD support was 85 days (CI, 56 to 114 days).
Devices were infected in 12 of 25 patients, 7 of whom had signs or symptoms of systemic infection. Seven of 12 patients with LVAD infection did not survive until transplantation (Fig 1A⇓); clinical reasons for death were multiple organ failure (MOF) (5 patients) and bleeding from the graft (1 patient) or the anastomosis (1 patient) of the outflow conduit. Eleven of 13 patients without LVAD infection survived until transplantation (Fig 1A⇓); the 2 deaths were attributable to a global preoperative hypoperfusion and to cerebral embolism, respectively. Of the transplanted patients, 2 died due to noninfectious causes (acute right ventricular failure and acute rejection). The remaining 16 patients have been discharged and are current long-term survivors. Analysis of survival to transplant reveals a significant difference between the survival curves (Fig 1A⇓; P<.05). Eight of 9 death events occurred before day 45. Analysis of the time-to-infection interval is shown in Fig 1B⇓ and indicates that at the mean time of LVAD support (86 days), the probability of infection-free status was .49 and decreased to .33 by the end of the study period.
Patients without LVAD infection were younger (mean, 38 years [CI, 32 to 44 years] versus 53 years [CI, 46 to 60 years]), had a longer LVAD support duration (median, 90 days [range, 20 to 335 days] versus 42 days [range, 17 to 158 days]), had less respirator support (median, 3.4% versus 13.2% of LVAD support time), and required surgical device revisions in 30.7% of cases compared with 83.3% of cases in the group of infected patients. Major complications of the clinical course occurred in 5 (38%) of 13 uninfected patients versus 10 (83%) of 12 infected patients and consisted mainly of intrathoracic bleeding complications in most (14 of 15) patients. Major embolic events affecting the central nervous system, mesenteric arteries, and at least one additional organ system occurred in 1 (7.7%) of 13 noninfected patients in contrast to 4 (33.3%) of 12 infected patients.
The principal pathogens recovered from the device surface on explant resulting in infection were Staphylococcus aureus (3 patients), Staphylococcus epidermidis (2), Enterococcus faecalis (3), Enterococcus faecium (1), Corynebacterium species (1), Bacillus cereus (1), and Candida albicans (1). Concomitant blood cultures with clonally identical pathogens were positive in 4 of 12 infected and 2 of 7 deceased patients. The device-infecting strain was demonstrated at the driveline exit site in only 3 of 12 patients; in the remaining patients, other pathogens or commensal strains were found. PFGE analysis is exemplified in Fig 2⇓ with selected isolates. This typing technique enabled us to demonstrate the molecular fingerprint of isolates from different explant culture sites and to compare these fingerprints with those from isolates of other origin. On the other hand, using PFGE, we could demonstrate that a blood culture isolate (patient 8, Staphylococcus hominis) was genetically different compared with isolates with identical antibiogram and biochemical reactions recovered from the device.
Infection is a major and potentially devastating complication in patients undergoing mechanical circulatory support (see review in Reference 5) and is associated with bleeding and thromboembolic events. The advent of implantable LVADs with a lower incidence of infection than that of the totally implantable artificial heart has suggested to some authors that infectious complications may not carry significant impact on overall survival.8 In other studies, the prevalence of infection in LVAD patients has been estimated to be between 23% and 58%,9 10 11 12 13 14 and mortality attributable to LVAD infection ranges between 15% and 44%.4 11 12 15 LVAD infection, however, is difficult to diagnose. MOF, a clinical entity frequently encountered in complicated courses after LVAD implantation, may be caused or aggravated by various factors including infection; however, positive cultures in the course of MOF may be the result rather than the cause of the severe morbidity. To add to the complexity, blood cultures may remain sterile, and direct access to the device in situ for sampling is often not possible. To resolve this dilemma, we prospectively sampled the devices on explantation and performed molecular typing of cultured pathogens. Survival analysis revealed a significant correlation between survival to transplant and absence of LVAD infection. At the time of LVAD implant, our patients were already in advanced, deteriorative condition. Although the correlation between infection and death does not necessarily imply a causal relationship, LVAD infection would not have even been recognized as a factor associated with death in most of the cases in the present study without the information obtained through culture and PFGE. Our study for the first time provides a link between LVAD infection and survival and may thus serve as the basis for the development of future policies to optimally prevent LVAD infection and prioritize patients for timing of surgery.
Not surprisingly considering the usual microflora of intravascular prosthesis infections, the primary pathogens observed in our patients with device infection were staphylococci. The prevalence of enterococci in our study may be explained by the high incidence of thromboembolism of small-bowel arteries. Whether enterococcal device colonization is the primary or secondary event, it is striking that major thromboembolic events occurred in 3 of 4 patients with enterococcal device infection compared with none of 8 patients with device infection caused by other pathogens. The observation of a high prevalence of Gram-negative pathogens16 or of fungi14 resulting in LVAD infection observed by other authors could not be confirmed in our study group. Of note, swab cultures obtained from the driveline exit site correlated poorly with organisms on the infected device because PFGE-identical strains from both sites were recovered only in 3 of 12 cases. Therefore, in cases without septicemia but with suspected LVAD infection, treatment must remain empiric and provide good coverage for Gram-positive bacteria.
Patients with device infection in our study tended to suffer from a more complicated course, including prolonged mechanical ventilation, bleeding, and major thromboembolic events. Most deaths occurred in the early phase of support (up to day 45), whereas all but one patient who were supported beyond day 45 survived to transplant. In agreement with a recent report,17 our results also show that even the presence of frank LVAD infection with persistent bacteremia and pus entirely surrounding the device (as demonstrated in two of our patients) is not a contraindication for transplantation because infection in two patients rapidly resolved after transplantation even under immunosuppressive therapy and another three patients were successfully transplanted despite LVAD colonization. The need for scrupulous prevention of hematoma, particularly in the implantation pocket with the subsequent possibility of infection, is emphasized. Furthermore, our data may support implementation of aggressive preventive infection-control measures (eg, superclean air, maximum precautions for prevention of bacteremia episodes, driveline design using impregnated cuffs) in this high-risk patient group.
This work has been supported by a grant from the German Minister for Education and Research (grant No. 01 KI 9450). Prof Proctor is a recipient of the Alexander-von-Humboldt award. We thank U. Ho¨rling and S. Weber for expert technical assistance, Dr R.-J. Fischer for statistical evaluation, and P. Diestelhorst for patient data documentation.
Reprint requests to Mathias Herrmann, MD, Institute of Medical Microbiology, University of Muenster Hospital and Clinics, Domagkstr 10, 48129 Muenster, Germany. E-mail firstname.lastname@example.org.
Presented in part at the 33rd Annual Meeting of the Infectious Diseases Society of America, San Francisco, Calif, September 16-18, 1995.
- Received September 12, 1996.
- Revision received December 13, 1996.
- Accepted December 18, 1996.
- Copyright © 1997 by American Heart Association
Hognes, JR. In: The Artificial Heart: Prototypes, Policies and Patients. Washington, DC: National Academy Press; 1991:1-312.
Annual Report of the US Scientific Registry for Organ Transplantation and the Organ Procurement and Transplantation Network–1990. Washington, DC: US Department of Health and Human Services; 1990.
Pennington DG, Swartz M. Infectious complications associated with ventricular assist device support. In: Lappin-Scott HM, Costerton JW, eds. Microbial Biofilms. Cambridge, UK: Cambridge University Press; 1995:322-326.
Maslow JN, Slutsky AM, Arbeit R. Application of pulsed-field gel electrophoresis to molecular epidemiology. In: Persing DH, Smith TF, Tenover FC, White TJ, eds. Diagnostic Molecular Microbiology: Principles and Applications. Washington, DC: American Society for Microbiology; 1993:563-572.
Ocampo A, Ramasumy N, Morley D, Billingham ME, Portner PM. Patients on left ventricular assist system: hematogenous infection and device retrieval analysis. Presented at the 21st Annual Meeting of the Society for Biomaterials; March 18-22, 1995; San Francisco, Calif.
Fischer S, Trenholme GM, Costanzo MR, Piccione W. Infectious complications of left ventricular assist device placement. In: Program and abstracts of the 33rd annual meeting of the Infectious Disease Society of America, September 16-18, 1995; San Francisco, Calif. Abstract 458.
Schmitt SK, Gordon SM, Vargo RL, McCarthy PM. Risk factors for bloodstream infection during cardiovascular support with the HeartMate left ventricular assist device. In: Program and abstracts of the 33rd annual meeting of the Infectious Disease Society of America; September 16-18, 1995; San Francisco, Calif. Abstract 109.
Messa J, Forman W, Axelrod P, Suh B. Infections complicating cardiac-assist devices. In: Program and abstracts of the 33rd annual meeting of the Infectious Disease Society of America; September 16-18, 1995; San Francisco, Calif. Abstract 103.
Catanese KA, Argenziano M, Moazami N, Gardocki MT, Clavenna MW, Scully BE, Rose EA, Oz MC, Levin HR. Infections in patients with left ventricular assist devices do not preclude successful transplantation. In: Program and abstracts of the 16th annual meeting of the International Society for Heart and Lung Transplantation; March 15-18, 1996; New York, NY. Abstract 067.