Left ventricular assist devices (LVAD) provide circulatory support on a temporary basis to patients awaiting heart transplantation. Increasingly, interest has focused on the use of LVADs as a means of entirely avoiding heart transplantation. However, the use of implantable LVADs, either temporary or permanent, often results in device complications. Device complications include hemorrhage, thromboembolism, arrhythmias, organ system failure, and infection. 1–3 Infection associated with LVADS is a major source of morbidity and mortality, and generally occurs in 25–50% of patients. 3,4–12
Classification systems for LVAD infections have been proposed 9–11 but have not yet gained widespread use. Multiple heart transplant centers have reported individual experiences with LVAD infections. Recently, a large retrospective case series of infected LVADs, which underwent removal, was reported. 12 To date, however, no prospective evaluations of LVAD infections have been published and no systematic review of the literature has been undertaken. We present three cases of LVAD infections seen at our institution and review the literature to assess clinical parameters associated with device infection, causative organisms, treatment modalities, and patient outcomes.
After obtaining Institutional Review Board approval, all cases of LVAD device infections at the Medical College of Ohio over a 10 year period of time (1990–2000) were reviewed. Left ventricular assist device infection was defined as the existence of one of the following: (1) presence of bacteria on a gram stain and/or culture from a surgically removed device; (2) positive blood cultures in the presence of an LVAD; and (3) bacteria on gram stain and/or culture from the surrounding structures (i.e., drive line exit site or pocket site). Three patients at the Medical College of Ohio met this definition of LVAD infection.
A review of the literature was undertaken. The English language literature was reviewed through a computer generated search using MEDLINE. The following search terms were used: heart assist devices, infection, and expansion of these terms. Cases of LVAD infection were selected for inclusion in this analysis using the criteria listed above if sufficient demographic information was available to allow identification of individual patients. Pediatric cases (age ≤ 13 years) were excluded. A total of 46 cases were identified from the literature.
A 50 year old woman with anticardiolipin antibody syndrome, morbid obesity, hypertension, and diabetes mellitus, sustained a massive anterolateral myocardial infarction. Five weeks after the infarction, a left ventricular assist device (HeartMate) was inserted. Approximately 25 weeks later, the patient developed a fever without localizing signs or symptoms. Blood and urine cultures were obtained, and two blood cultures were positive for enterococcus. The patient was treated with ampicillin and gentamicin for 6 weeks. Six weeks later, the abdominal incision began to drain serosanguinous material, which was culture positive for enterococcus. The patient was retreated for 6 weeks with ampicillin and gentamicin to which the organism was sensitive. Approximately 8 weeks after discontinuation of antibiotics, the patient again developed a discharge from the abdominal incision, which grew enterococcus, and blood cultures were again positive for enterococcus. Ampicillin alone was begun due to the increasing hearing loss experienced by the patient from the aminoglycoside. Two weeks later, the patient underwent successful cardiac transplantation and LVAD removal. The device was not cultured, but postoperative ampicillin alone was given for 6 weeks. The patient did relatively well posttransplantation without recurrence of infection for 2 years until she died suddenly and unexpectedly at home.
A previously healthy 45 year old man sustained an acute myocardial infarction with cardiogenic shock and left ventricular assist device (HeartMate) placement. Approximately 22 weeks after insertion of the LVAD, the patient developed pain around the drive line site, with surrounding erythema and purulent discharge. A culture of the drainage grew normal skin flora. The patient was treated with 2 weeks of ciprofloxacin and vancomycin with resolution of the erythema, tenderness, and purulent discharge. Approximately 5 days after the antibiotics were discontinued, culture negative purulent drainage again developed from the drive line. Vancomycin was restarted with a subsequent decrease in tenderness and purulent discharge. Approximately 8 weeks later, while still receiving vancomycin, the patient again developed tenderness and purulent drainage from the drive line that grew Klebsiella pneumoniae, sensitive to ciprofloxacin; this was added to the vancomycin and both antibiotics were continued until the patient was successfully transplanted 4 weeks later and the LVAD was removed. LVAD cultures did not reveal any growth. Six weeks of vancomycin and ciprofloxacin was given postoperatively, and the patient was without recurrence of infection.
A 54 year old man with idiopathic dilated cardiomyopathy and penicillin allergy developed decompensated congestive heart failure unresponsive to medication and underwent LVAD (HeartMate) placement. Seven days later, he developed culture-negative drainage around a JP drain and the drive line site and was treated with empiric vancomycin and ceftazidime for 1 week. One of the two blood cultures was positive for Micrococcus and diphtheroids. Four weeks later, the patient developed bleeding around the drive line site, fever, and leukocytosis. Blood cultures were positive for nutritionally deficient streptococcus and Staphylococcus epidermidis and drainage from the drive line site grew methicillin sensitive Staphylococcus aureus. Vancomycin and gentamicin were given for 2 weeks with clinical improvement. Two weeks after the antibiotics were discontinued, the patient was again febrile with increasing purulent drainage from the drive line site. A CT scan showed fluid and air around the drive line connecting to the aorta. Vancomycin and gentamycin were restarted and 3 of 4 blood cultures grew enterococcus and S. aureus. The LVAD site drainage also grew methicillin sensitive S. aureus. The LVAD was replaced and S. aureus and enterococci were cultured from the components. Vancomycin was continued for 4 weeks, and gentamicin was continued for 2 weeks. Four days after discontinuing the vancomycin, the patient developed a S. aureus bacteremia. Vancomycin was restarted and continued for 8 weeks until successful cardiac transplantation.
A total of 46 cases, including 3 that are reported for the first time, were reviewed. 1,6–8,13–20 Not all patient variables were recorded in each case. Cases are referenced in Table 1 and are reported in this article by case number. Only 12 cases reported gender. Of these, there were 8 men and 4 women. The age range was 14–55 years, with an average age of 45.6 years. Only nine cases listed underlying medical illnesses: diabetes mellitus (cases 1, 4), hyperbilirubinemia (case 5), ventricular tachycardia (case 3), anticardiolipin antibody syndrome (case 1), morbid obesity (cases 1, 4, 28), sleep apnea (case 28), exfoliative dermatitis (case 27), and endocarditis with thrombotic obstruction (case 46). The reason for LVAD placement was listed in 18 cases. Cardiogenic shock was the primary reason for LVAD placement (cases 1, 2, 4, 5, 8, 9, 47), followed by idiopathic cardiomyopathy (cases 3, 10, 13, 15, 25, 27, 28) and congestive heart failure (cases 7, 12, 14, 26). In 44 cases, the type of LVAD placed was specified. The following models were used: TCI HeartMate (n = 29), Novacor (n = 10), Thoratec (n = 3), Sans-Thoratec (n = 1), and Jarvik (n = 1). All LVADs were placed as bridges to transplantation. The total duration of LVAD placement was an average of 131days (range, 15–444 days; median, 75 days).
The number of days from placement of LVAD to diagnosis of the first LVAD infection was specified in 25 cases and averaged 65.6 days (range, 6–403 days; median, 31 days). Eighty five percent (85%) of LVAD infections occurred in a device left in for greater than 2 weeks. Only 30% of the infections occurred in devices left in for greater than 6 weeks.
The following signs and symptoms of LVAD infection were commonly reported: fever (n = 14), leukocytosis (n = 7), drainage from exit site (n = 7), and abnormality around the exit site such as bleeding (n = 4), pain (n = 3), erythema (n = 3), necrosis of skin or wound (n = 2), and cachexia (n = 1). In seven cases, mechanical device malfunction occurred before the onset of diagnosed infection. Examples of malfunction included hematoma or bleeding within the device (cases 1, 3, 5, 6, 10), inlet obstruction (case 35), and outflow rupture (case 36). Per the case definition, all 46 cases had evidence of bacteremia or fungemia, or bacterial or fungal infection on the device, or recovery of the organism from the surrounding structures.
The most common sites of infection were bloodstream infections (n = 21), followed closely by drive line site infections (n = 19) (Table 2). Wound infections (n = 12) and LVAD device infections (n = 14) were seen more commonly than pocket infections (n = 7). A wound infection consisted of erythema, swelling, drainage around a drive line exit site or midsternal or abdominal incision site, and/or a positive culture from a wound. A pocket infection consisted of drainage, erythema, swelling and/or a positive culture of an organism from the material in the tissue pocket in which the LVAD is contained. Nearly half of all the 46 cases had evidence of infection at more than one site. Only two cases had bloodstream infection only without evidence of infection at another site (cases 14, 22).
Staphylococcus aureus was recovered in the majority of cases (n = 24) (Table 3). The next most common group of organisms were gram negative rods (Pseudomonas, Klebsiella, Escherichia coli, Enterobacter, Proteus) (n = 9). Other commonly recovered organisms were Candida species (n = 8), enterococcus species (n = 7), and S. epidermidis or coagulase negative Staphylococcus (n = 7). Four of the seven S. epidermidis infections involved the drive line. Antibiotic sensitivities were not reported for the vast majority of bacteria isolated.
Treatment of LVAD infections involved primarily surgical intervention with ancillary antibiotics. Surgical therapy was specified in 16 of 46 cases. Antibiotic therapy was reported in 17 cases, which may or may not have specified the surgical therapy. In six cases, surgical treatment was the only mode of treatment specified (cases 11, 14, 22, 31, 35, 38). The organisms involved were mainly S. aureus, enterococcus species, and one case of Pseudomonas (case 35). Four of those six patients had successful transplantation without any posttransplantation complications. In the other two patients, one was awaiting transplantation (case 10) and the other (case 35) outcome was not specified. The surgical treatments used were drive line site or incisional debridement and drainage and/or irrigation (cases 11, 25, 38); replacement of the LVAD (case 31); exploration of the pump pocket (case 14); removal of expanded polytetrafluoroethylene pericardial substitute (ePTFE) (case 22); explantation (case 35).
In another seven cases, only antibiotic therapy was used and no surgical treatment was specified (cases 2, 7, 15–20). The organisms involved in these cases were gram negative rods, Candida, S. aureus, and coagulase negative Staphylococcus. The primary antibiotic therapy was vancomycin (duration varied from 14 to 16 days) and amphotericin (42–87 days). All seven of the patients treated with antibiotics only were successfully transplanted.
In another 10 cases, the treatment consisted of both antibiotics and surgery (cases 1, 3–6, 8–10, 18, 21]. Organisms involved here were mainly S. aureus, coagulase negative Staphylococcus, and enterococcus species. The predominate antibiotic used was vancomycin, with duration ranging from 28 to 270 days. The surgical therapy consisted of the following: drive line site or incisional debridement, drainage, or irrigation (cases 1, 4, 5, 9, 21); replacement of LVAD (cases 3, 8, 10); and exploration of pump pocket (cases 1, 3, 6, 9, 21). Eight of these 10 patients were successfully transplanted. One of these patients died (case 8), and one was awaiting transplant at the time of the case report (case 10).
Treatment duration for intravenous antibiotics were extremely variable and, when specified (n = 14), often ranged from 3 days to 16 weeks. The majority of cases were treated with continuous antibiotics until transplant. Seven were treated with antibiotics after transplantation (range, 2–7weeks). In 28 cases, antibiotic therapy was not given posttransplantation or was not specified.
The majority of patients (n = 34 of 46, 73%) underwent cardiac transplantation. Of the remaining 12 patients, 4 were awaiting transplant at the time of case report, and 8 died before transplantation. For five of the eight patients (62%) who died before transplantation, the cause of death was multiorgan system failure due to sepsis (cases 8, 28, 32, 33, 46). Four of these eight patients had a gram negative rod infection (cases 28, 32, 33, 46), which directly involved the LVAD in three cases (cases 28, 32, 33). Two of these cases were caused by Pseudomonas infection (cases 28, 33), out of a total of three pseudomonal cases in the entire review. Surgical intervention or antibiotic treatment was not reported in either of these cases.
Secondary LVAD infections that developed after treatment of a primary infection occurred in 14 cases. Of these 14 recurrences, 7 had surgical treatment; of these 7, only 1 patient had surgery for the initial infection (case 5). Two of these seven had exploration of the pump pocket and debridement of the wound (cases 4, 5). Replacement of the LVAD was performed on two patients (cases 3, 8), local wound irrigation in two (cases 38, 39), and LVAD repair with irrigation of the pump pocket in one case 1. The other 7 of the 14 did not have any surgical treatment or treatment was not specified. Seven of the 14 cases specified antibiotic use (cases 1, 3–5, 7, 8, 15). Five cases had both antibiotic therapy and surgery (cases 1, 3–5, 8). Only two recurrent LVAD infections were treated with surgery alone (cases 38,39) without ancillary antibiotic use specified. Nine of 14 (56%) cases of recurrence involved S. aureus.
Left ventricular assist devices are an important bridge to cardiac transplantation for many patients with failed hearts. Due to the shortage of available hearts for transplant, many centers have looked to LVADs as long-term devices to entirely bypass cardiac transplant2. One of the major complications that interferes with both of these goals is development of LVAD device infections.
In this report, we presented three cases of LVAD infection and reviewed the literature. One of the major limitations of this composition was its retrospective nature. Many case reports were incomplete or did not allow full patient identification. However, we did examine the clinical parameters, causative organisms and optimal therapy associated with infections. Interestingly, the mortality associated with LVAD infections was relatively similar (17%) to overall mortality associated with LVAD devices. 17,18,21,22 In fact, in some series, infection rate was examined and found not to be associated with increased mortality overall. 8,17 No predominantly immunosuppressive diseases such as diabetes mellitus were associated with LVAD infection. The vast majority of patients had an idiopathic cause for their heart failure as opposed to an illness requiring immunosuppression. This finding may reflect selection bias for LVAD placement.
Eighty- five percent (85%) of LVAD infections occurred in a device left in for greater than 2 weeks. Only 30% of the infections occurred in devices left in for greater than 6 weeks. Therefore, the majority of infections (55%) recurred between weeks 2 through 6. This period may be a high-risk time for infection to occur. In another study of LVAD infections, 1 42% of patients developed infection (site not specified) by 4 weeks; at 12 weeks, only a 3% difference in infection rate was reported. This window of increased risk of LVAD infection may be due to the changes occurring in the microenvironment of the device. Under appropriate conditions, an extensive exopolysaccharide polymer forms over time on the surface of the LVAD. This polymer may facilitate ligand-receptor interaction and bacterial adhesion to the substrate 23,24 and result in an increased risk of device infection.
Some authors consider infection a contraindication for transplantation. 15 Others have reported a high morbidity and mortality for LVAD infection. 25 However, the majority of patients in this series did well with transplantation even in the presence of infection. No patients died of sepsis due to recurrent infection after transplantation. In fact, only 5 of 46 (10.8%) died due to sepsis before transplantation.
A majority of infections occurred in a functional device. Twenty-two percent of the devices that had infections had a mechanical malfunction before infection. In a case series reported by Holman et al. with 38 patients, 2 of 5 patients with device malfunction developed infection. Device malfunction may predispose the LVAD to infection.
Adhesions that form on the device can prolong the actual heart transplantation procedure and increase blood loss, particularly if the patient is maintained on anticoagulation. 20 In light of this finding, some surgeons have favored placing an ePTFE membrane pericardial substitute around the LVAD to reduce fibrous adhesions and calcification. This method also allows easier removal or repair of device should malfunction or infection develop; but this membrane has been reported to increase the incidence of infection. In this series, two patients with ePTFE pericardial substitutes were infected and one of these patients died. In view of this, some authors have recommended aggressive treatment of this membrane infection. 20
Infection with gram negative rods, especially Pseudomonas, had a surprisingly poor outcome. For reasons that are unclear, these gram negative rod infections tended to involve the LVAD device directly. This involvement is associated with a poor prognosis, regardless of the infecting organism. 12,13 Little information was available about the treatment modalities used to treat these infections. However, antibiotic therapy as reported in these cases was associated with a uniformly poor outcome, and aggressive surgical therapy may be warranted.
Speciation of Candida and enterococcus was not reported in most studies. Speciation of these two organisms may be important in light of the rising incidence of vancomycin resistant enterococci and fluconazole resistant Candida species. For example, Enterococcus faecium as opposed to Enterococcus faecalis is more likely to be vancomycin resistant. Candida albicans is usually susceptible to fluconazole. However, other non-albicans Candida can have varying levels of resistance to fluconazole.
Work has been done on alternative placement of drive lines to decrease the incidence of LVAD infections. Several studies have examined placement of drive lines through the iliac crest. 26,27 At least four patients have had drive line site pneumatic tubing stabilized through the iliac crest, and none of these four patients developed infection by means of this method. Further work is needed on novel preventative techniques such as these that may decrease the incidences of infection.
Given the organisms associated with device infection in this series, empiric treatment for patients with LVAD device infection should include antibiotic coverage for S. aureus and gram negative bacilli. Vancomycin and a third generation cephalosporin such as ceftazidime or a fluoroquinolone such as ciprofloxacin would be appropriate. Antifungal coverage should be considered in the initial empiric regimen as well, given the high rate of fungal infections reported in the literature.
It is probably prudent to continue antibiotics at therapeutic levels until transplantation, 6 especially with S. aureus infections. Approximately 30% (14 of 46 cases) of infections were recurrent infections and 56% of those involved S. aureus. Infection even with an organism likely to cause recurrent infection should not be a contraindication for transplantation as the majority of patients in this series did well posttransplantation. A prospective study of LVAD infections would be very useful in the evaluation of risk factors, causative organisms, and optimal therapy.
In conclusion, LVAD infections are common and, overall, associated with a good outcome, except where the infecting organism is a gram negative rod such as Pseudomonas. S. aureus infections tend to recur when antibiotics are stopped and may require continuous suppression until transplantation. 14,16,19
1. Kormos RL, Borovetz HS, Armitage JM, Hardesty RL, Marrone GC, Griffith BP: Evolving experience with mechanical circulatory support. Ann Surg 214: 471–477, 1991.
2. Myers TJ, McGee MG, Zeluff BJ, Radovancevic B, Frazier OH: Frequency and significance of infections in patients receiving prolonged LVAD support. ASAIO Trans 37: M283–M285, 1991.
3. Petri WA Jr: Infections in heart transplant recipients. Clin Infect Dis 18: 141–148, 1994.
4. Burns GL: Infections associated with implanted blood pumps. Int J Artif Organs 16: 771–776, 1993.
5. Masters RG, Hendry PJ, Davies RA, et al: Cardiac transplantation after mechanical circulatory support: A Canadian perspective. Ann Thorac Surg 61: 1734–1739, 1996.
6. McCarthy PM, Schmitt SK, Vargo RL, Gordon S, Keys TF, Hobbs RE: Implantable LVAD infections: Implications for permanent use of the device. Ann Thorac Surg 61: 359–365, 1996.
7. Moroney DA, Vaca KJ: Infectious complications associated with ventricular assist devices. Am J Crit Care 4: 204–211, 1995.
8. Myers TJ, Catanese KA, Vargo RL, Dressler DK: Extended cardiac support with a portable left ventricular assist system in the home. ASAIO J 142: M576–M579, 1996.
9. Hill JD: Infections: Prophylaxis and treatment. Ann Thorac Surg 46: 131–132, 1988.
10. Holman WL, Skinner JL, Waites KB, Benza RL, McGiffin DC, Kirklin JK: Infection during circulatory support with ventricular assist devices. Ann Thorac Surg 68: 711–716, 1999.
11. Gaykowski R, Taylor K, Yates W: Cumulative clinical experience with the symbion J7 TAH. Trans Am Soc Artif Intern Organs 34: 455–459, 1988.
12. Chua JD, Wilkoff BL, Lee I, Juratli N, Longworth DL, Gordon SM: Diagnosis and management of infections involving implantable electrophysiologic cardiac devices. Ann Intern Med 133: 604–608, 2000.
13. Jarvik R, Westaby S, Katsumata T, Pigott D, Evans RD: LVAD power delivery: A percutaneous approach to avoid infection. Ann Thorac Surg 65: 470–473, 1998.
14. Phillips WS, Burton NA, Macmanus Q, Lefrak EA: Surgical complications in bridging to transplantation: The Thermo Cardiosystems LVAD. Ann Thorac Surg 53: 482–486, 1992.
15. Burton NA, Lefrak EA, Macmanus Q, et al: A reliable bridge to cardiac transplantation: The TCI left ventricular assist device. Ann Thorac Surg 55: 1425–1431, 1993.
16. Hravnak M, George E, Kormos RL: Management of chronic left ventricular assist device percutaneous lead insertion sites. J Heart Lung Transplant 12: 856–863, 1993.
17. Parameshwar J, Wallwork J: Left ventricular assist devices: Current status and future applications. Int J Cardiol 62: S23–S27, 1997.
18. Holman WL, Bourge RC, Zorn GL, Brantley LH, Kirklin JK: Use of expanded polytetrafluoroethylene pericardial substitute with ventricular assist devices. Ann Thorac Surg 55: 181–183, 1993.
19. Fischer SA, Trenholme GM. Costanzo MR, Piccione W: Infectious complications in left ventricular assist device recipients. Clin Infect Dis 24: 18–23, 1997.
20. Holman WL, Bourge RC, Spruell RD, Murrah CP, McGiffin DC, Kirklin JK: Ventricular assist devices as a bridge to cardiac transplantation. Ann Surg 225: 695–706, 1997.
21. Koyanagi H, Kitamura M, Nishida H, Hachida M, Endo M, Hasimoto A: Current strategy for severe heart failure with mechanical circulatory support. Artif Organs 19: 766–768, 1995.
22. Griffith BP, Kormos RL, Nastala CJ, Winowich S, Pristas JM: Results of extended bridge to transplantation: window into the future of permanent ventricular assist devices. Ann Thorac Surg 61: 396–398, 1996.
23. Gristina AG, Giridhar G, Gabriel BL, Naylor PT, Myrvic QN: Cell biology and molecular mechanisms in artificial device infections. Int J Artif Organs 16: 755–764, 1993.
24. Gristina AG, Dobbins JJ, Giammara B, Lewis JC, DeVries WC: Biomaterial centered sepsis and the total artificial heart. JAMA 259: 870–874, 1988.
25. Wasler A, Springer WE, Radovancevic B, Myers TJ, Stutts LA, Frazier OH: A comparison between intraperitoneal and extraperitoneal left ventricular assist system placement. ASAIO J 42: M573–M576, 1996.
26. Phillips SJ, Kongthahworn C, Zeff RH, et al: A new left ventricular assist device: Clinical experience in two patients. Trans ASAIO 25: 186–191, 1979.
27. Phillips SJ, Kongthahworn C, Skinner JR, et al: Permanent Left Ventricular Assistance for Outpatients. Tex Heart Inst J 16: 275–279, 1989.