Infections of intracardiac devices Adolf W. Karchmer, MD a,b, * , David L. Longworth, MD c,d,e a Division of Infectious Diseases, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Kennedy-6, Boston, MA 02215, USA b Harvard Medical School, 25 Shattuck Street, Boston, MA 02115, USA c Department of Infectious Diseases, The Cleveland Clinic Foundation, Infectious Diseases Section, 9500 Euclid Avenue, Cleveland, OH 44106, USA d Cleveland Clinic Foundation Health Sciences Center, The Ohio State University School of Medicine, Cleveland, OH, USA e School of Medicine, Pennsylvania State University, PA, USA Prosthetic cardiac valves, pacemakers, and implanted defibrillators are essential to maintain the hemodynamic capacity or electrical integrity of the heart in several cardiac disease states. Severe morbidity and mortality results when these devices become infected. Treatment of infected cardiac devices is particularly complex. Eradication of infection involving an implanted for- eign device often requires removal of the foreign material in conjunction with antimicrobial therapy. If the device is not essential and if removal does not entail significant risks, treatment may be simple. Prosthetic valves and electrophysiologic devices, however, are usually essential for life. Further- more, removal requires replacement with a functioning device, a procedure that is associated with morbidity and potential mortality. This article will address infection of these devices and their treatment. Prosthetic valve endocarditis Frequency and risk The frequency of prosthetic valve endocarditis (PVE) after cardiac sur- gery is not uniform. The rate is highest during the initial three months after surgery, remains high through the sixth month, then declines gradually to a relatively constant rate of 0.3% to 0.6% at 12 months and thereafter [1–3]. Infect Dis Clin N Am 16 (2002) 477–505 * Corresponding author. E-mail address: [email protected] (A.W. Karchmer). 0891-5520/02/$ - see front matter Ó 2002, Elsevier Science (USA). All rights reserved. PII: S 0 8 9 1 - 5 5 2 0 ( 0 1 ) 0 0 0 0 5 - 8
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Infections of intracardiac devices
Adolf W. Karchmer, MDa,b,*,David L. Longworth, MDc,d,e
aDivision of Infectious Diseases, Beth Israel Deaconess Medical Center,
330 Brookline Avenue, Kennedy-6, Boston, MA 02215, USAbHarvard Medical School, 25 Shattuck Street, Boston, MA 02115, USAcDepartment of Infectious Diseases, The Cleveland Clinic Foundation,
Infectious Diseases Section, 9500 Euclid Avenue, Cleveland, OH 44106, USAdCleveland Clinic Foundation Health Sciences Center, The Ohio State University
School of Medicine, Cleveland, OH, USAeSchool of Medicine, Pennsylvania State University, PA, USA
Prosthetic cardiac valves, pacemakers, and implanted defibrillators are
essential to maintain the hemodynamic capacity or electrical integrity of theheart in several cardiac disease states. Severe morbidity and mortality results
when these devices become infected. Treatment of infected cardiac devices is
particularly complex. Eradication of infection involving an implanted for-
eign device often requires removal of the foreign material in conjunction
with antimicrobial therapy. If the device is not essential and if removal does
not entail significant risks, treatment may be simple. Prosthetic valves and
electrophysiologic devices, however, are usually essential for life. Further-
more, removal requires replacement with a functioning device, a procedurethat is associated with morbidity and potential mortality. This article will
address infection of these devices and their treatment.
Prosthetic valve endocarditis
Frequency and risk
The frequency of prosthetic valve endocarditis (PVE) after cardiac sur-
gery is not uniform. The rate is highest during the initial three months aftersurgery, remains high through the sixth month, then declines gradually to a
relatively constant rate of 0.3% to 0.6% at 12 months and thereafter [1–3].
0891-5520/02/$ - see front matter � 2002, Elsevier Science (USA). All rights reserved.
PII: S 0 8 9 1 - 5 5 2 0 ( 0 1 ) 0 0 0 0 5 - 8
When valve recipients are followed actively after surgery, the actuarial esti-
mate of the cumulative rate of PVE ranges from 1.0% to 1.4% at 12 months
and from 3.0% to 5.7% at 60 months [1–6] (Table 1). Although conflictingdata have been reported, recent studies have suggested that infection occurs
with similar frequency on valves implanted at the mitral or aortic site [1,3,6].
Additionally, at the end of the initial postoperative year, the rates of infec-
tion involving mechanical valves and bioprosthetic valves (tissue leaflets) are
similar [6–9]. Alternatively, other studies have indicated that the risk of
infection is greater for mechanical valves compared with bioprosthetic
valves during the initial year after valve replacement, but that over time the
risk of infection for bioprosthetic valves increases so that rates are compar-able for the two valve types 5 years after valve surgery [1,3,5,8].
Pathogenesis
The pathogenesis of infective endocarditis is complex. Anatomic and
hemodynamic abnormalities alter endothelial surfaces resulting in the
deposition of platelet–fibrin thrombi. These platelet–fibrin aggregates are
sites at which organisms with unique surface components acting as adhesins
can attach. Some organisms, most notably Staphylococcus aureus, whichuses surface proteins that recognize and bind to adhesive matrix molecules,
can also adhere to apparently normal or minimally traumatized endothelial
surfaces in the absence of preparatory platelet–fibrin thrombi. Adherent
organisms that survive and proliferate in spite of host defenses give rise to
infective endocarditis. The presence of a foreign body (e.g., a prosthetic
heart valve) at the site of infection introduces other major variables. Initi-
ally, the annulus–prosthesis interface is not endothelialized and acts as a for-
mation site for platelet–fibrin thrombi. Exposed sutures anchoring theprosthesis provide pathways whereby organisms can invade cardiac tissue.
Table 1
Estimated cumulative rate of PVE after valve replacement
% patients with PVE
Study (years valves Initial number valve Months after surgery
implanted) recipients 12 24 48 60
Rutledge et al. (1956–81) [6] 1598 1.4 — — 3.2
Ivert et al. (1975–79) [3] 1465 3.0 — 4.1 —
Calderwood et al. (1975–82) [1] 2608 3.1 — 5.4 5.7
Arvay and Lengyel (1981–85) [5] 912 — — — 4.9
Agnihotri et al. (1970–92) [4] 2413a 1.0 1.5 — 3.0
Glower et al. (1975–95) [2] 1119b — — — 3.0
a Aortic position only.b Carpentier-Edwards bioprosthesis only.
Adapted from Karchmer AW. Infections of prosthetic valves and intravascular devices. In:
Mandell GL, Bennett JE, Dolin R, editors. Principles and practice of infectious diseases, 5th
edition. Philadelphia: Churchill Livingstone; 2000. p. 903–17; with permission.
478 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
After several years in place, the stress of repetitive motion may alter the sur-
face of leaflets in bioprostheses and allow the deposition of infection-prone
platelet–fibrin thrombi. The foreign material also impairs host defenses in its
immediate microenvironment and alters the behavior of adherent infectingmicroorganisms, rendering them more difficult to eradicate with antimicro-
bial agents. Organisms can reach the prosthesis site either by the hematoge-
neous route (from a remote site) or by direct introduction at the time of
surgery. This grossly oversimplified scenario has been reviewed in detail
[10–12].
The clinical events that contribute to the development of PVE markedly
influence the microbiology, time of symptomatic onset, and, potentially, the
pathology of PVE. Accordingly, understanding and recognizing these eventsis relevant to the management of infected patients. Based on the types
of organisms causing PVE that develop during the first two months after
valve surgery, it appears that many of these episodes are nosocomial. Epide-
miologic and microbiologic studies have linked PVE caused by coagulase-
negative staphylococci—the cause of 31% of PVE cases during the initial
postoperative months—to intra-operative contamination [13–16]. Most
cases of PVE occurring in these clusters became symptomatic within 60 days
after surgery; however, occasional patients presented clinically between 3and 13 months postoperatively. The cases with delayed onset illustrate the
potential for slow evolution from early infection to symptomatic PVE.
Early-onset PVE has been linked to bacteremia arising from invasive mon-
itoring and support devices and surgical wound infections. The frequency of
these events engenders the high rates of PVE noted through the initial six
months after surgery. Furthermore, during these early months after valve
surgery, infection of the prosthesis is more likely to be associated with para-
valvular invasion and hemodynamically significant valve dysfunction[17,18].
While not universally indicative of endocarditis, nosocomial bacteremia
or fungemia occurring in a patient with a prosthetic valve constitutes a risk
for subsequent PVE. Fang et al. noted PVE in 18 of 115 patients (16%) with
nosocomial bacteremia. Of the bacteremias associated with subsequent
PVE, 61% arose from intravascular catheters or skin and wound infections
[19]. Bacteremia due to staphylococci and gram-negative bacilli resulted in
55% and 33% of subsequent PVE, respectively. Although the experience issmall, PVE developed in 6 of 19 patients (31%) with bacteremia due to Staph-
ylococcus epidermidis, 4 of 23 (17%) with S. aureus bacteremia, and 6 of 58
(10%) with gram-negative rod bacteremia. In 12 of the 18 patients with PVE,
the inciting bacteremia occurred within 60 days of surgery; however, the
published data do not allow assessment of risk for PVE following bactere-
mia in the early months after surgery versus later bacteremia [19]. Nasser et
al. examined the consequences of nosocomial candidemia in 44 patients with
prosthetic heart valves [20]. Eleven patients (25%) developed PVE. In 7patients (16%) with a mean of 8.1 days of fungemia, coincident PVE was
479A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
evident at the time of presentation (median 149 days, range 8–1240 days
after surgery; only 3 within 60 days postoperatively). These patients gener-
ally lacked an identifiable portal of entry for fungemia, and in retrospect hadnot experienced a complicated recovery after surgery. Among the 37
patients without evident PVE at the time of candidemia, 4 (11%) developed
PVE. Blood cultures had been obtained in these 4 patients between days 4
and 36 postoperatively, but PVE was not noted until days 26, 101, 112, and
690. In contrast to those with PVE on presentation, these 37 patients had
definable portals of entry and had experienced long, complicated recoveries
after surgery. Among these 37 patients, the 4 with subsequent PVE had
more sustained fungemia than those who did not develop endocarditis(mean 14.3 days versus 4.3 days). Patients with prosthetic valves who experi-
ence fungemia are at significant risk of either having (those without an iden-
tifiable portal of entry) or developing PVE. In spite of aggressive antifungal
therapy, patients with nosocomial fungemia remain at risk for the develop-
ment of PVE months or years later.
Microbiology
Many and varied organisms, including Coxiella burnetii, Mycoplasma
hominis, many species of bacteria, Tropheryma whippelii, atypical mycobac-
teria, Legionella dumoffii and Legionella pneumophila, and an array of yeasts
and molds have infected prosthetic heart valves. The vast majority of episo-
des, however, are caused by a relatively small number of organisms
(Table 2). From a microbiological perspective, PVE can be divided into
three periods. The origin of infection among those who become sympto-
matic within 60 days of valve surgery (i.e., those with early PVE) is pre-dominantly nosocomial. Accordingly, the major causes of these infections
are organisms commonly associated with intraoperative contamination and
nosocomial infection. With the exception of an increased frequency of cases
caused by S. epidermidis and other species of coagulase-negative staphylo-
cocci, the organisms causing PVE that presents more than 12 months after
cardiac surgery in generally healthy community-residing patients resemble
the causes of native valve endocarditis; the major pathogens are viridans
streptococci, S. aureus, and enterococci. The HACEK group of organisms(Haemophilus aphrophilus, Actinobacillus actinomycetemcomitans, Cardio-
bacterium hominis, Eikenella spp, and Kingella spp) cause a small but signifi-
cant number of these late PVE cases. During the period from 2 months to 12
months after surgery the organisms causing PVE are a blend of those
encountered in early and late PVE cases.
The coagulase-negative staphylococci causing PVE vary significantly in
relation to the time of onset of infection after cardiac surgery. Those causing
infection during the initial year after valve surgery are S. epidermidis, andalmost 85% are methicillin-resistant. The coagulase-negative staphylococci
are roughly equally divided between S. epidermidis and non-epidermidis
480 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
species after 12 months have elapsed; additionally, only 30% or less of these
isolates are methicillin resistant [1,17,21]. Although the mechanism of methi-
cillin resistance is the same as that in methicillin-resistant (MR) S. aureus,
coagulase-negative staphylococci commonly exhibit heteroresistance, which
makes detection of the MR phenotype more difficult. When initiatingtherapy for coagulase-negative staphylococcal PVE, the organism should
be assumed to be MR until the microbiology laboratory definitively ex-
cludes MR.
Various diphtheroids or Corynebacterium spp, including isolates consis-
tent with the antibiotic-resistant C. jekiem, have caused PVE. Many
diphtheroids are fastidious and difficult to grow and test for antibiotic sus-
ceptibility; however, most isolates are highly susceptible to vancomycin.
Additionally, strains not resistant to gentamicin are killed synergisticallyby the combination of penicillin and gentamicin [22].
Among 270 cases of fungal endocarditis reported from 1965 through
a Includes viridans streptococci, Streptococcus bovis, other non-group A, groupable strepto-
cocci, and Abiotrophia species (nutritionally variant streptococci).b Includes Haemophilus species, Actinobacillus actinomycetemcomitans, Cardiobacterium
hominis, Eikenella species, and Kingella kingae.
Data from Karchmer AW. Infections of prosthetic valves and intravascular devices. In:
Mandell GL, Bennett JE, Dolin R, editors. Principles and practice of infectious diseases, 5th
edition. Philadelphia: Churchill Livingstone; 2000. p. 903–17; and Karchmer AW. Infective
endocarditis. In: Braunwald E, editor. A textbook of cardiovascular medicine. Philadelphia:
W.B. Saunders Company; 2000.
481A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
positive in patients with PVE caused by Candida spp or C. neoformans,
whereas blood cultures are often negative when PVE is caused by other
fungi. In the latter circumstance, cultures and microscopic examination ofvegetation recovered at cardiac surgery or embolized peripherally may be
needed to establish a microbiologic diagnosis.
Although a rare cause of endocarditis in the United States, C. burnetii is
an important cause of PVE in regions where Q fever is endemic. Among 229
episodes of C. burnetii endocarditis diagnosed between 1985 and 1998, 157
sue, resulting in abscess formation or valve dehiscence, and occasionally
affects the valve orifice directly, resulting in either regurgitation or stenosis
[17,27–31]. Extension of infection through the aortic valve annulus maycause pericarditis, and invasion of the membranous portion of the interven-
tricular septum may disrupt the conduction system and cause various forms
of heart block [27,28,31,32]. Annulus invasion and myocardial abscess were
noted at surgery or autopsy in 42% and 14%, respectively, of 85 patients
with mechanical valve endocarditis [15,30]. Ismail et al. noted annulus inva-
sion and valve dehiscence in 82% of 41 patients with mechanical valve infec-
tion [29]. With bioprosthetic valve infection, especially that occurring within
the initial postoperative year, invasion into and beyond the annulus hasbeen reported in 36% to 54% of cases [17,33,34]. Among patients with PVE
treated surgically, Lytle et al. noted that infection invaded paravalvular tis-
sue in 43 of 54 (79%) mechanical valves and 37 of 90 (41%) bioprosthetic
valve infections [18]. In addition, Lytle et al. reported that invasion was
common among patients with mechanical valve endocarditis regardless of
time of onset, whereas among patients with bioprosthetic valve infection
invasive disease was more common among those with onset during the initi-
al year after valve replacement (15 of 18, 83%) versus those with later onsetinfection (22 of 71, 31%) [18]. Infection involving the leaflets of bioprosthetic
valves can result in perforation and tears with regurgitation, bulky vegeta-
tions can cause stenosis, and after infection has been cured, delayed leaflet
stiffening also can result in stenosis [34,35].
Clinical and laboratory manifestations
The clinical and laboratory features of PVE are similar to those encoun-
tered in patients with native valve endocarditis [36]. When PVE presents dur-
ing the early postoperative period, the subtle and time-dependent signs of
endocarditis may be absent or masked by complications of surgery. The high
frequency of paravalvular invasion infection in patients with PVE, especiallyin those with aortic valve infection during the initial postoperative year,
results in a higher frequency of new or changing regurgitant murmurs,
482 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
congestive heart failure, persistent fever in spite of optimal antimicrobial
therapy, and new electrocardiographic conduction disturbances than is
encountered with native valve endocarditis [17,37]. Almost 40% of patients
experience clinically apparent systemic emboli; 20% to 40% experience stroke,central nervous system hemorrhage, or neurologic complications [29,38,39].
Unless antibiotics have been administered recently, blood cultures will be
positive in 90% of patients with PVE, and multiple cultures obtained inde-
pendently over time will be positive (persistent bacteremia). When dealing
with blood cultures yielding coagulase-negative staphylococci or diphther-
oids (organisms that often contaminate cultures), persistent bacteremia
helps to distinguish patients with PVE. Although polyclonal coagulase-
negative staphylococcal endocarditis has been reported, this is unusual, andusing molecular techniques to demonstrate clonality among sporadic coagu-
lase-negative staphylococci in blood cultures obtained from patients with
prosthetic valves is also suggestive of PVE rather than contamination [40,41].
Occasionally, patients with PVE who have not received antibiotics re-
cently will have persistently negative blood cultures. These cases are cau-
sed by fastidious microorganisms, for example, Legionella spp, C. burnetii,
M. hominis, Bartonella spp, atypical mycobacteria, and fungi other than
Candida spp (particularly molds). Special blood culture, techniques, serologictests (antigen or antibody detection), and, when possible, culture, micro-
scopic, andmolecular analysis of vegetationmay facilitate making a diagnosis
in these cases [42].
Transthoracic echocardiography (TTE) and transesophageal echocardio-
graphy (TEE) using biplane or multiplanar transducers that allow continu-
ous wave, pulse-wave Doppler, and color-flow imaging are essential for the
diagnosis and management of PVE [43,44]. Together the two approaches
provide optimal imaging of prostheses in the aortic, mitral, and tricuspidpositions. TEE is superior to TTE when imaging a mitral valve prosthesis.
Although TTE and TEE are complimentary, TEE is more sensitive without
loss of specificity for the diagnosis of PVE than is TTE (82% to 96% versus
17% to 36%) regardless of prosthesis type or anatomic position (Table 3)
[43–47]. Additionally, TEE is also superior to TTE for detecting paravalvu-
lar abscesses, fistulae, and paraprosthetic leaks—intracardiac complications
that have a major impact on management strategy [48]. The negative predic-
tive value of a single TEE in a patient with suspected PVE is 86% to 94%[46,49]. However, if PVE is strongly suspected in a patient with negative
TEE, the study should be repeated. Although a repeat TEE may provide
diagnostic evidence of PVE, a second negative study does not exclude PVE
in the face of strong clinical evidence [49].
Diagnosis
The diagnosis of PVE requires a high index of suspicion, knowledge ofthe subtle signs of endocarditis, obtaining 3 or 4 blood cultures prior to the
483A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
administration of antibiotics, and full echocardiographic assessment. The
Duke criteria for the diagnosis of endocarditis provide a systematic
approach to evaluating patients with suspected PVE [50]. In three studies
of pathologically confirmed PVE, retrospective application of the Duke cri-
teria classified 76% to 79% of patients as definite and 21% to 24% as possibleendocarditis [51–53]. Although a significant number of patients were not
assessed by TEE, the diagnosis of PVE was erroneously rejected in only 1
of 118 patients. The importance of TEE evaluation in the diagnosis of PVE
was examined by Roe et al. [54]. Among 34 patients with prosthetic heart
valves and suspected endocarditis, the classification of 13 (38%) was changed
when the assessment included TEE data versus when only TTE findings were
used; 11 were reclassified from possible to definite and 2 from rejected to
possible. The positive predictive value of a TEE for PVE was 89%. The TEEcan be negative in patients with PVE; in the diagnosis of PVE a negative
TEE should never override strong clinical evidence of endocarditis [49,54].
Antimicrobial treatment
Effective therapy for PVE requires identification of the microbial cause,
determination of a bactericidal regimen of proven efficacy, an understanding
of the intracardiac pathology of PVE and its implications for valve replace-
ment, and effective management of extracardiac complications. The com-
plexity of evaluation and treatment suggests that patients should be
hospitalized in centers with experience in treating PVE. It is essential to iso-
late the causative organism. When evaluating patients who present withindolent disease and are hemodynamically stable, antibiotic therapy should
be delayed for 2 or 3 days awaiting the results of blood cultures. This is par-
ticularly important if antibiotics have been administered recently, because if
initial cultures are negative, it allows one to obtain additional cultures with-
out further confounding. Patients who present with fulminant PVE or who
are hemodynamically unstable and may require urgent replacement of the
Table 3
Echocardiography in the diagnosis of PVEa
Number of valves (%)
Prosthesis type and position Transthoracic Transesophageal
a Patients had pathoanatomic or clinically confirmed (Von Reyn criteria) PVE.
Data from Refs. [36,45,46].
484 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
infected prosthetic valve should be treated empirically immediately after
obtaining blood cultures.
Although administered for a longer duration, the antimicrobial therapy
recommended for the treatment of PVE caused by specific organisms, withthe exception of staphylococci, is similar to that used to treat native valve
endocarditis (Table 4). Detailed comments regarding these regimens can
be found elsewhere [17,36,55]. If streptomycin or gentamicin is to be com-
bined with a cell wall active antibiotic to achieve synergistic killing of strep-
tococci or enterococci, the organism must not exhibit high-level resistance to
the selected aminoglycoside (i.e., unable to grow in the presence of 500 lg/mL of gentamicin or 1000 lg/mL of streptomycin). The presence of high-
level resistance precludes a synergistic effect from the respective aminoglyco-side. Gentamicin high-level resistance also generally precludes a synergistic
effect from netilmicin, kanamycin, tobramycin, and amikacin. Optimal ther-
apy for PVE caused by vancomycin-resistant Enterococcus faecium (VREF)
that are also resistant to penicillin and ampicillin and highly resistant to
streptomycin and gentamicin has not been established. Although linezolid
and quinupristin–dalfopristin are often active against VREF, these agents
have not been proven to be effective therapy for PVE. Surgical intervention
during suppressive antimicrobial therapy should be considered for thesepatients.
For optimal treatment of staphylococcal PVE, a multi-drug regimen is
recommended. In vitro data, evidence from animal models, and clinical
experience indicate that rifampin has a unique ability to kill staphylococci
adherent to foreign material and is an important component of regimens
used to treat staphylococcal PVE [17,21,56–60]. Staphylococci have a rela-
tively high intrinsic mutation rate for the gene controlling the site of rifam-
pin action; thus when relatively large populations of staphylococci areexposed to rifampin, selection of rifampin-resistant organisms is common.
To prevent the emergence of rifampin-resistant organisms, the regimens
selected for the treatment of staphylococcal PVE should contain two anti-
biotics that are known to be active against the staphylococcal isolate in
addition to rifampin. All three antimicrobials can be started at the same
time. If two additional anti-staphylococcal agents have not been identified,
treatment with a single anti-staphylococcal agent should be administered for
3 to 5 days before beginning rifampin. This approach may reduce the totalnumber of staphylococci and thus diminish the probability that a rifampin-
resistant subpopulation will emerge.
The regimens recommended for the treatment of staphylococcal PVE are
not based on the species of the isolate—S. aureus or coagulase-negative
staphylococci—but rather upon susceptibility of the isolate to methicillin.
For isolates that are susceptible to gentamicin, it is an effective second
anti-staphylococcal antimicrobial. For strains resistant to gentamicin and
other aminoglycosides, a fluoroquinolone to which the strain is highly sus-ceptible should be considered [57,60].
485A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
Table
4
Recommended
antibiotictherapyforpatients
withPVE
Infectingorganism
Antibiotic
Dose
androute
a
Duration
(wk)
Comments
1.Penicillin-susceptible
viridansstreptococci,
Streptococcusbovis,
andother
streptococci
(penicillinMIC
£0.1
lg/m
L)
A.PenicillinG
plus
gentamicin
18–24millionunitsIV
dailyin
divided
dosesq4h
1mg/kgIM
orIV
q8h
6 2
Mayomitaminoglycoside
when
potentialfornephrotoxicity
isincreased
B.Ceftriaxoneplus
gentamicin
2gIV
orIM
daily
assingle
dose
1mg/kgIM
orIV
q8h
6 2
Canbeusedin
patients
withnon-immediate
penicillinallergy.
Intramuscularceftriaxoneispainful.
Cephapirin
2gIV
q4hcanbe
substitutedforceftriaxone.
C.Vancomycinb
15mg/kgIV
q12h
6Use
forpatients
withim
mediate
orseverepenicillinorcephalosporinallergy.
Infuse
dosesover
1hourto
avoid
histaminerelease
reaction(redmansyndrome).
2.Relativelypenicillin-resistant
streptococci(penicillinMIC
>0.2
lg/m
L)
A.PenicillinG
plus
24–30millionunitsIV
dailyin
divided
dosesq4h
6Preferred
fornutritionallyvariant
(pridoxal-orcysteine-requiring)
streptococci(A
biotrophia
spp).
gentamicin
1mg/kgIM
orIV
q8h
4
3.Enterococci(invitro
evaluationforMIC
topenicillinandvancomycin,
b-lactamase
production,
andhigh-level
resistance
togentamicin
and
streptomycinrequired)
A.PenicillinG
plus
gentamicin
24–30millionunitsIV
dailyin
divided
dosesq4h
1mg/kgIV
orIM
q8h
6 6
Streptomycincanbeusedinstead
ofgentamicin
ifstreptomycinhigh-level
resistance
isnotpresent.If
high-level
resistance
togentamicin
butnotto
streptomycinisdetected,streptomycin
ispreferred.
486 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
B.Ampicillinplus
gentamicin
2gIV
q4h
6
Samedose
asnotedabove
6
C.Vancomycinbplus
gentamicin
15mg/kgIV
q12h
Samedose
asnotedabove
6 6
Use
forpatients
withpenicillinallergy.
Donotuse
cephalosporins.
4.Staphylococci
methicillin-susceptible
(assumepenicillinresistance)
A.Nafcillinoroxacillin
plusgentamicin
plusrifampin
c
2gIV
q4h
1mg/kgIV
orIM
q8h
300mgpoq8h
6–8
2 6–8
Penicillin18–24millionunitsdailyin
divided
dosesq4hcanbeusedinsteadofnafcillin
oroxacillin
ifstrainsdonotproduce
b-lactamase
andpenicillinMIC
is£0
.1lg
/mL.
Cephapirin
2gm
IVq4hcanbeused
inlieu
ofnafcillin/oxacillin
forpatients
withnon-immediate
allergyto
penicillins.
Use
gentamicin
duringinitialtw
oweeks.
See
textforalternatesforgentamicin.
Forpatients
withim
mediate
penicillinallergy,
use
regim
en5.
5.Staphylococci
methicillin-resistant
A.Vancomycinbplus
gentamicin
plusrifampin
c
15mg/kgIV
q12h
1mg/kgIV
orIM
q8h
300mgpoq8h
6–8
2 6–8
Use
gentamicin
duringtheinitialtw
o
weeksoftherapy.See
textforalternatives
togentamicin.Donotsubstitute
acephalosporinorim
ipenem
forvancomycin.
6.HACEK
organismsd
A.Ceftriaxone
2gIV
orIM
daily
asasingle
dose
6Cefotaxim
eorother
thirdgeneration
cephalosporinin
comparable
doses
maybeused.
B.Ampicillinplus
gentamicin
2gIV
dailyin
divided
dosesq4h
1mg/kgIV
orIM
q8h
6 4
Testorganism
forb-lactamase
production.
Donotuse
thisregim
en
ifb-lactamase
isproduced.
(continued
onnextpage)
487A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
488 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
For fungal PVE, amphotericin B (0.7 to 1.0 mg/kg/d) is recommended;
larger doses (1.0 to 1.5 mg/kg/d) are recommended for treatment of PVE
caused by molds (e.g., Aspergillus spp). A synergistic effect is often sought
by combining amphotericin B with 5-fluorocytosine (150 mg/kg/d dividedinto 4 doses with adjustments for renal dysfunction). The role of liposomal
formulations of amphotericin B or caspofugin in the treatment of PVE has
not been established. Early surgical intervention is considered a standard
element of treating fungal PVE [23–25]. As a result of the high rate of
relapse of candida PVE, initial therapy is often followed by long-term or
indefinite suppressive therapy with fluconazole (200 to 400 mg orally per
day) [24,25,61].
Surgical management
Murmurs suggesting valve dysfunction, moderate to severe congestive
heart failure due to valve dysfunction, fever for 10 or more days in spite
of appropriate antibiotic therapy, new onset electrocardiographic conduc-
tion abnormalities, or selected findings on echocardiogram identifies compli-
cated PVE with extension of infection into paravalvular tissue or valve
dysfunction. PVE with one or more of these features is not likely to respondto antibiotic therapy alone [37,62,63]. Complicated PVE is more common
with infection of aortic valve prostheses and with infection during the initial
year after valve surgery [37]. Antibiotic therapy combined with valve repla-
cement and cardiac reconstruction results in higher survival rates and fewer
relapses, rehospitalizations for valve surgery, and late endocarditis-related
mortality than does treatment with antibiotics alone in patients with compli-
cated PVE [37,63–65].
Indications for surgical treatment of PVE have been developed step-wiseas the pathology of PVE was defined, surgical techniques were refined, and
the presumed inevitability of recurrent PVE involving the newly implanted
prosthesis was disproven. The indications (see box on following page) are
not absolute and must be implemented with a careful risk–benefit analysis.
Notably, patients with late-onset PVE caused by viridans streptococci,
HACEK, or enterococci and without evidence of paravalvular invasion or
valve dysfunction can be treated successfully with antibiotics alone [17].
Survival rates among PVE patients with moderate to severe congestiveheart failure due to valve dysfunction are dramatically improved if these
patients are treated surgically. Only occasional patients in this group will
survive if treated with antibiotics alone. In contrast, 44% to 64% survive
when treated with antimicrobials plus valve replacement [17,63,66]. If out-
come is to be optimal, surgical intervention to correct valve dysfunction
must be performed before intractable heart failure ensures. There is no
evidence that delaying surgery in the face of progressive hemodynamic
deterioration in order to administer additional antibiotics improves out-come or reduces the frequency of recurrent endocarditis [33,67]. In fact,
489A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
postoperative mortality is proportional to the severity of hemodynamic
impairment at the time of surgery [33,68]. Occasionally, patients with PVE
who have responded to antibiotics develop valve dysfunction but remain
hemodynamically compensated. Surgery can be delayed until late in the
course of antimicrobial therapy in these patients.
Even in the absence of heart failure, mortality rates are high for patients
with paravalvular invasion when they are treated with antibiotics alone. Incontrast, complex reconstructive surgery in these patients has been associa-
ted with survival rates of 80% [64,69,70]. Relapse of PVE after optimal anti-
microbial therapy commonly reflects unrecognized invasive disease. These
patients should be treated with antibiotics plus surgery [37,62].
The outcome of S. aureus PVE is improved by aggressive surgical inter-
vention. When treated with antibiotics alone, 70% of patients with S. aureus
PVE die [71–73]. Intracardiac complications in patients with S. aureus PVE
are associated with a 13.7-fold increase in mortality; however, mortality isreduced 20-fold (OR 0.05, 95% CI: 0.005–0.42; P¼ 0.004) by surgical inter-
vention during antimicrobial therapy [51]. In fact, improved survival with
Indications for cardiac surgery in patients with PVEa
• Moderate to severe congestive heart failure due to valvedysfunction (regurgitant or stenotic)
• Unstable prosthesis• Paravalvular extension of infection, especially with abscessformation or fistula
• Uncontrolled infection (persistent bacteremia) on optimalantimicrobial therapy
• PVE caused by selected microorganisms�fungi�P. aeruginosa�S. aureus�enterococci in the absence of available bactericidaltherapy
�other gram-negative bacilli and microorganisms thatusually require surgery when infecting native valves(e.g., Brucella spp, Coxiella burnetii)
Relapse after optimal therapyLarge (>10 mm) hypermobile vegetationsCulture negative PVE with unexplained persistent (‡10 days)
fever
a Not all indications are absolute (see text)Adapted from Karchmer AW. Infections of prosthetic valves and intravascular
devices. In: Mandell GL, Bennett JE, Dolin R, editors. Principles and practice ofinfectious diseases. New York: Churchill Livingstone; in press.
490 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
antibiotics plus surgical treatment versus antibiotics alone was noted among
both those with and without intracardiac complication. Multiple studies
have suggested that PVE caused by S. aureus is treated most effectively with
antibiotics plus aggressive cardiac surgery [63,72–74].The role of surgical intervention to prevent systemic emboli is unclear.
Patients with native valve endocarditis who have vegetations >10 mm in dia-
meter experience a higher rate of embolic complications than do patients
with smaller vegetations [75]. Although data are not available correlating
vegetation size with arterial emboli in patients with PVE, the overall rates
of embolic events are similar in patients with infection of native and prosthe-
tic heart valves. Furthermore, these rates decrease rapidly with effective anti-
biotic therapy. While it is desirable to prevent emboli to vital organs, it isnot clear that mortality and morbidity are reduced by cardiac surgery in
PVE patients simply because vegetations >10 mm in diameter are detected.
The decision to operate in this situation must be made in the context of an
overall clinical evaluation and integrated plan for therapy rather than on
vegetation size alone.
An optimal outcome for patients with PVE often requires complex recon-
struction of the aortic or mitral valve and the supporting or adjacent tissue.
With appropriate antibiotic therapy and surgical intervention by experi-enced cardiac surgeons, survival rates for patients with complicated PVE
range from 70% to 90% [18,64,69,70,76–79]. These remarkable results sup-
port the concept that—when possible—patients with PVE complicated by
extensive invasion and tissue disruption should undergo surgery in centers
with extensive experience.
The long-term results of surgical treatment of PVE are also encouraging.
Recurrent PVE is noted in only 6% to 15% of patients, and repeat cardiac
surgery for either recrudescent PVE or dysfunction of the newly implantedvalve is required in 18% to 26% [18,33,37,64,70,77,79]. Survival rates 5 years
after surgery for PVE have ranged from 54% to 87% [18,33,70,76,77].
Anticoagulant therapy
The use and specific type (heparin versus warfarin) of anticoagulant ther-
apy in patients with active PVE is controversial [80–84]. Cerebrovascular
accidents were reported in 16 of 133 patients (12%) with mechanical valveinfection who were effectively anticoagulated versus 32 of 76 patients (42%)
who were not anticoagulated. Additionally, hemorrhagic complications
were not increased among those who were anticoagulated [80]. Accordingly,
carefully monitored anticoagulant therapy is recommended for patients with
infected mechanical or bioprosthetic valves when this therapy would be used
routinely in the absence of infection. The authors prefer warfarin and
suggest the International Normalized Ratio (INR) be maintained between
2.5 and 3.0. If intracerebral hemorrhage or another contraindication toanticoagulant therapy is noted, anticoagulation should be reversed. Because
491A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
rifampin accelerates the hepatic clearance of warfarin, patients being treated
for PVE with rifampin require major adjustments in daily warfarin doses
Cardiac pacemakers and implantable cardioverter defibrillators haverevolutionized the treatment of patients with rhythm disorders, and their use
has become increasingly common in clinical practice. It is estimated that, as
of April 1999, there are approximately 180,000 functioning implantable car-
diac defibrillators and 3.25 million permanent pacemakers presently in use
worldwide [85]. The rates of infectious complications involving these devices
have decreased in recent years with improvements in surgical techniques and
with the development of transvenous devices. Nevertheless, infection rates
remain in the range of 1% to 7% [85–89]. With the burgeoning number ofdevices implanted, infections involving them are increasingly encountered
in clinical practice.
Anatomy, pathogenesis, and classification
Most permanent pacemakers and cardiac defibrillator devices presently
in use are transvenous devices. They consist of a generator or defibrillator
implanted in the subcutaneous tissue of the chest wall or abdomen, and elec-
trodes which traverse the soft tissue, enter the subclavian vein, pass through
the right heart, and terminate in the right atrial and ventricular endocar-
dium. Depending on the type of device, some patients with transvenous car-
diac defibrillators may have a mesh defibrillator patch positioned in thesubcutaneous tissue in the left mid or posterior axilllary area. In over 98%
of patients, the current cardioverter–defibrillator systems are completely
transvenous. In the early years of cardiac pacing and cardiodefibrillator
technology, electrodes and mesh defibrillator patches were placed by way
of thoractomy or sternotomy directly onto the epicardial surface of the
heart. Some patients still have these devices in place. The development of
transvenous devices—especially cardioverter defibrillators and combination
devices that serve both pacing and defibrillating functions—represents animportant advance in the field, given the reduced morbidity associated with
their implantation.
Infections involving these devices have traditionally been classified ana-
tomically and by time of onset following implantation. Infection may
involve the generator or debrillator pocket (Fig. 1), the electrodes as they
traverse the soft tissue and venous system or track to the epicardium in older
devices, mesh patches in the subcutaneous tissue or on the surface of the
heart, and valvular or nonvalvular endocardium. Deep endovascular infec-tion may involve the tricuspid valve, the insertion site of the lead tip in the
right atrium or ventricle and, on occasion, thrombi that form around the
492 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
electrodes in the venous system or right atrium (Fig. 2) [90]. Myocardial
abscesses are rare but have been described [90]. Early infection is classified
as infection developing in the first month following implantation; late infec-
tion presents beyond the first month [85,91]. Some studies have further clas-
sified this latter group into those presenting from months 1 to 12 afterimplantation (‘‘late infection’’) versus those presenting beyond one year
(‘‘delayed infection’’) [85].
Several pathogenetic mechanisms contribute to the development of infec-
tion involving these devices. Early infection most often arises from intrao-
perative contamination at the time of device implantation or generator
exchange and is attributable to direct microbial seeding of the device or
pocket. Late infection occasionally develops from primary mechanical ero-
sion of the generator or defibrillator through the skin, with resultant seedingof the pocket. In other cases, chronic smoldering infection in the pocket pro-
duces a thinning of the overlying tissue and ultimately device erosion.
Whichever mechanism occurs, infection may spread from the pocket to
involve the electrodes and ultimately the endocardial surfaces of the heart.
Hematogenous seeding of the device from distant sites of infection has been
reported, but with the exception of S. aureus bacteremia appears to be rela-
tively rare [90,92,93]. In one study of 21 patients with S. aureus bacteremia
occurring a year or more after device placement, 29% developed a subse-quent device infection [94].
Fig. 1. ‘‘Pocket infection’’: Erythematous discoloration of the skin overlying an infected pocket
in a patient’s left chest wall, containing an implantable cardioverter–defibrillator. (Courtesy of
Bruce Wilkoff, MD.)
493A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
Epidemiology and risk factors
A number of studies have examined risk factors associated with the devel-
opment of infections involving these devices. Underlying conditions that
predispose patients to pacemaker-related infections include malnutrition,malignancy, diabetes mellitus, skin disorders, and the use of corticosteroids,
other immunosuppressive agents, or anticoagulants [92,95,96]. Prolonged
sternal wound infections, and diabetes mellitus are risk factors associated
with the development of cardiodefibrillator-associated infections [97,98].
In the largest recent review of infections associated with these devices in
123 patients treated at the Cleveland Clinic Foundation, comorbid condi-
tions were common and included coronary artery disease in 64%, coronaryartery bypass surgery in 32%, diabetes mellitus in 26%, anticoagulation
in 19%, atrial fibrillation in 19%, malignancy in 6%, corticosteroid use in
5%, and hemodialysis in 2% [85].
The timing, route, and location of implantable cardiodefibrillator place-
ment have been associated with the risk of subsequent infection. In one
recent study, infectious complications fell from 16.7% to zero coincident
with an increase in transvenous device placement and with the elimination
of two staged procedures in which electrophysiologic studies performed inthe cardiac catheterization laboratories were followed within hours to days
by the placement of defibrillator devices in the operating room by cardiac
Fig. 2. Implantable cardioverter–defibrillator lead with a large, attached, infected vegetation.
Adjacent to the lead is a large, infected embolus that was removed from the pulmonary artery of
the patient from whom this infected device was extracted. (Courtesy of Bruce Wilkoff, MD.)
494 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
surgeons [88]. In recent years, electrophysiologic studies have been followed
immediately by pacemaker or defibrillator placement in the cardiac catheter-
ization laboratory or in the operating room for patients requiring these
devices.
Microbiology
Numerous studies have examined the microbiology of these infections[85,87,89,91,93,99–105]. In nearly all series, the majority of infections are
caused by staphylococci, including S. aureus and coagulase-negative staphy-
lococci. Some studies have suggested that infections occurring within the
first month following implantation are more likely due to S. aureus, and late
infections are more commonly caused by coagulase-negative staphylococci.
This is likely attributable to the fact that many late infections arise from
mechanical erosion of the device through the skin, resulting in contamina-
tion of the pocket with skin flora. Less common isolates include enterococci,Peptostreptococcus spp, Propionibacterium acnes, micrococcus, and gram-
negative bacilli such as E. coli, and Enterobacter, Serratia, Pseudomonas,
and Klebsiella species. Mycobacterium avium-intracellulare and fungal
pathogens such as Candida spp, Torulopsis glabrata and Aspergillus spp have
been implicated on rare occasion [106–108].
In the most recent series from the Cleveland Clinic, which evaluated 123
patients with infected pacemakers (87) or cardioverter defibrillators (36), the
most common pathogens were coagulase-negative staphylococci (68%),S. aureus (24%), and enteric gram-negative bacilli (17%). Thirteen percent
of infections were polymicrobial. The distribution of pathogens was similar
between infected pacemakers and cardioverter–defibrillators.
Clinical presentation
The clinical presentation of these infections is highly variable, both with
regard to time of onset and clinical severity. In the series by Chua et al., 25%
of infections occurred within the first month, 33% presented late (days
29–364), and 42% were delayed, presenting beyond 1 year [28]. Clinical man-
ifestations in patients from this series are summarized in Table 5 in descend-
ing order of frequency and are consistent with earlier reports [87,98].
Several points merit emphasis. Eighty-five patients (69%) presented withsymptoms localized to the device pocket, including erythema (55%), pain
(55%), a draining sinus (42%), erosion (32%), and warmth (23%). Twenty-
five patients (20%) presented with a combination of local and systemic
symptoms, and thirteen (11%) presented with systemic complaints alone.
Overall, a history of fever was present in only 29%, and was documented
in only 19%. The frequent absence of fever in patients with these infections
has been emphasized by other researchers [98]. Despite the absence of
documented fever in over 80% of patients in the Cleveland Clinic series, bac-teremia was surprisingly common and occurred in one-third of patients.
495A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
Among the 40 patients with bacteremia, fever and systemic symptoms were
absent in 16 (40%). Clinicians must therefore have a high index of suspicion
for bacteremia, even in patients who lack fever and systemic signs of infec-
tion. Blood cultures are indicated in all patients with suspected or proveninfections of these devices. The possibility of cryptic device infection, includ-
ing infection limited to the generator pocket, must be considered in patients
with staphylococcal bacteremia. Additionally, occurrence of pulmonary
emboli (Fig. 2) can provide a clue that an intracardiac lead has become
infected, especially if a chest radiography shows multiple focal infiltrates,
indicating the likelihood of septic emboli.
Some studies confined to the analysis of patients with pacemaker endo-
carditis have reported fever in up to 86% to 91% of patients and chills inup to 75% [99,101]. In a review of 44 cases from Israel, a new or changing
murmur was present in 13%, leukocytosis exceeding 10 · 109/L in 66%, sple-nomegaly in 11%, anemia in 39%, and microscopic hematuria in 59% [99].
Several studies have examined the role of echocardiography in the diag-
nosis of implantable cardiac device infections. In the Cleveland Clinic study,
64 of 123 patients had transthoracic (45), transesophageal (8), or both (11)
echocardiographic procedures performed. Vegetations involving the device
leads or a valve were present in 13 patients, 12 with a pacemaker, and 1 witha cardioverter–defibrillator. In three other studies confined to patients with
proven pacemaker endocarditis, the sensitivity of transthoracic echocardio-
Table 5
Clinical manifestations in 123 patients with infections involving implantable electrophysiologic
cardiac devicesa
Manifestation Number Percent
Pocket pain 68 55
Pocket erythema 68 55
Draining sinus from pocket 52 42
Pocket swelling 44 36
Bacteremia 40 33
Pocket erosion 39 32
History of fever 35 29
Pocket warmth 28 23
Purulent drainage from pocket to sinus 28 23
Chills 27 22
Malaise 26 21
Bacteremia accompanied by systemic symptoms 24 20
Fever on examination 23 19
Anorexia 14 12
Sepsis (fever and tachycardia) 14 12
Tachycardia (>100 beats/min) 10 8
Nausea 10 8
a Includes pacemakers (87) and cardioverter defibrillators (36).
Modified from Chua JD, Wilkoff BL, Lee I, Juratli N, Longworth DL, Gordon SM.
Diagnosis and management of infections involving implantable electrophysiologic cardiac
devices. Ann Int Med 2000;133:604–8, with permission.
496 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
graphy in demonstrating valvular or lead vegetations was 23% to 30%, com-
pared with 91% to 96% with transesophageal echocardiography [100,101,
105]. The authors believe that echocardiography is indicated in all patients
with bacteremic implantable device-related infections and in those whoseclinical findings that suggest the possibility of endovascular infection. Trans-
esophageal echocardiography is superior in sensitivity to the transthoracic
approach.
Management
The optimal management of implantable device-related infections has
been debated in the literature, especially regarding the necessity for deviceremoval. Some authors have advocated a conservative approach consisting
of antibiotic therapy with hardware in place. Pursuit of this strategy has been
motivated by the desire to avoid invasive surgical procedures, especially in
debilitated individuals and in those with epicardial devices. Turkisher et al.
described a single patient in whom a cardiodefibrillator infection was suc-
cessfully managed with a 3-month course of oral fusidic acid and rifampin
[109]. Lee et al. reported successful treatment of 4 patients with implantable
cardiodefibrillator infections with wide debridement of the device pocketfollowed by placement of a closed irrigation system and continuous irriga-
tion with polymyxin, bacitracin, and neomycin solution, along with culture-
directed antimicrobial therapy [110]. In general, reports of success without
removing the entire device have been based on small numbers of patients
or patients with infections limited to the generator pocket [102].
Other authors have recommended complete explantation of infected
devices, but have commented that in rare cases patients may be successfully
treated with partial device removal [91,98,102]. Partial removal is likely tobe sufficient only when infection is clearly limited to the removed compo-
nent, usually the generator unit and immediately adjacent leads. In the
recent large study involving 123 patients at the Cleveland Clinic, complete
removal of the device and all lead material was successful in 117 patients
(95%) [85]. One hundred nineteen received antimicrobial therapy after hard-
ware removal; specific therapy was guided by culture results but was not
standardized with regard to route or duration. Seventy-one patients received
intravenous therapy for a median of 28 days. Forty-three received intrave-nous antibiotics (median 7 days) followed by oral therapy (median 16 days).
Five patients received oral therapy alone for a median of 24 days. Among
the 117 patients who had complete device removal, relapse occurred in only
one. By contrast, among the remaining six patients in whom device removal
was incomplete, relapse occurred in three. Chua et al. concluded that com-
plete explantation of all hardware combined with antibiotic therapy is the
optimal method of management of implantable device-related infections.
Data from several recent studies of pacemaker endocarditis strongly supportremoval of the entire system as the preferred therapy [93,99,100–102].
497A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
Among 190 patients with pacemaker endocarditis, 12 of 29 patients (41%)
treated with antibiotics alone died as contrasted with 30 of the 161 patients
(19%) treated with antibiotics plus removal of the entire system [100].Older studies have emphasized the morbidity and potential mortality
associated with device removal, especially in those requiring sternotomy
or thoracotomy [99]. Older traction techniques to remove transvenous
devices were sometimes associated with complications such as tricuspid
valve tear, rupture of the right atrial or ventricular wall, tearing of the
cephalic or subclavian veins, or rupture of the chordae tendineae [86]. In
recent years, the availability of laser sheath technology for extraction of
pacing leads has facilitated nonoperative removal of transvenous devices[111]. In one recent study, use of the Excimer laser sheath to extract pacing
leads was successful in 94% of cases versus 64% using standard non-laser
technology; in patients who failed non-laser standard extractions, 88% were
successfully treated utilizing the laser tool. Life-threatening complications
occurred in 3% of patients treated with standard non-laser extraction, com-
pared with 0% in those treated with the laser Excimer sheath [111]. Patients
with suspected infection of an implanted defibrillator with epicardial patches
are particularly vexing because removal of the entire device requires a thor-acotomy. Bacteremia, fever, and signs of septic pericarditis are clues that
point to epicardial lead infection. Additionally, computed tomography and
a gallium or labeled white blood cell scan can be used to evaluate for infec-
tion involving the deep leads and patches. If infection appears limited to the
generator pocket (i.e., there are neither clinical clues or patch abnormalities
by computed tomography or isotope scan) removal of only the generator
and adjacent lead wires, in combination with antibiotic therapy, may—on
occasion—be sufficient to eradicate infection [102]. Success rates with partialdevice removal are low because the entire device is typically infected.
Few studies have examined the optimal route or duration of antibiotic
treatment for infected electrophysiologic devices. Experience suggests that
those patients with bacteremia and definite or probable endocarditis or peri-
carditis will require long courses of parenteral therapy as advocated for
endocarditis. Patients with infection limited to the generator unit pocket and
no bacteremia can be treated with less intensive regimens (parenteral ther-
apy followed by oral therapy) and wound care for the pocket from whichthe device has been explanted [36].
At the Cleveland Clinic, all patients with suspected infections involving
cardiac devices have at least 2 sets of blood cultures collected. At the time
of device removal, cultures are obtained from the device pocket and from
all lead tips and mesh. Patients with endocarditis receive at least 4 weeks
of culture-directed intravenous antibiotic therapy. Those with infection con-
fined to the device pocket are treated with parenteral therapy with or with-
out subsequent oral therapy; the duration of therapy depends on thepresence or absence of bacteremia and on the clinical response. The optimal
duration of therapy remains controversial and has not been well defined for
498 A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
those individuals who lack bacteremia, fever, and echocardiographic evi-
dence of endocarditis but who nevertheless have positive lead tip cultures.
Because of the concern about focal infection at the right atrial or ventricular
lead tip site in those with positive lead tip cultures, most patients receive 4weeks of parenteral therapy. The optimal antibiotic management of patients
with these infections requires further study.
The timing of device reimplanation has been an area of ongoing contro-
versy. Some authors have recommended reimplanation as early as 36 hours
following explantation in those with infection confined to the device pocket
[112]. Because blood cultures may take longer than 36 hours to become
positive, others have advocated waiting at least several days [85]. In the
Cleveland Clinic study, the median interval from device explantation to re-implantation was 5 days (range 0–68 days). Among the thirteen patients
with endocarditis, reimplantation was successfully performed in those who
required it at a median of 7 days following explantation (range 5–25 days).
Before replacing an intracardiac device in a patient with device-associated
endocarditis, it is prudent to provide sufficient antibiotic therapy to eradi-
cate bacteremia and to sufficiently suppress or eradicate endothelial infec-
tion so that the probability of reinfection of the new device is minimized.
It is important to emphasize that not all patients with electrophysiolo-gic cardiac device-related infections require new devices. In the Cleveland
Clinic study, 18% of patients did not require further device therapy after
re-evaluation of their cardiac status.
Infections of left ventricular assist devices and artificial hearts
As an increasing number of left ventricular assist devices (LVAD) areimplanted, it has become clear that they are subject to serious risk of infec-
tion. In a controlled study of 68 patients followed for up to 2.5 years after
implantation of LVADs for severe heart failure, approximately one-third of
the devices became infected within three months [113]. Infections of the
drive-line tract or pocket occurred at a rate of 0.41 per patient year, and
infection of the pump’s interior or inflow/outflow tracts at a rate of 0.23 per
patient year. This resulted in an overall sepsis rate of 0.6 per patient year.
Moreover, sepsis was the most common cause of death. Sepsis caused41% of the 41 deaths in the LVAD group, but only one (2%) of 54 deaths
among the controls [113]. It is evident that until more effective means to
reduce the high rate and high mortality of infections of LVADs and similar
devices such as artificial hearts are found, their long-term value will be
limited.
As to treatment, removal of the device offers the best chance for cure. De
Jonge et al. [114] reported recently an apparent cure of a case of ‘‘endocar-
ditis’’ in an LVAD by ‘‘valve replacement’’—that is, removal and replace-ment of the mechanical inflow and outflow valves of the device. For various
499A.W. Karchmer, D.L. Longworth / Infect Dis Clin N Am 16 (2002) 477–505
reasons, removal of the device may not be practicable, so many of these
patients have been treated empirically with antibiotics. Data are so far inade-
quate to assess the likelihood of successful long-term suppression—or evenpossibly cure—of such infections by means of antibiotic treatment alone.
However, extensive previous experience with infections of other prosthetic
devices indicates that cure with antibiotics alone is unlikely. Thus, amore pro-
ductive strategy may be suppression of infection and early cardiac transplan-
tation with eradicative therapy administered after heart transplantation.
References
[1] Calderwood SB, Swinski LA, Waternaux CM, Karchmer AW, Buckley MJ. Risk factors
for the development of prosthetic valve endocarditis. Circulation 1985;72:31–7.
[2] Glower DD, Landolfo KP, Cheruvu S, et al. Determinants of 15-year outcome with 1,119
standard Carpentier-Edwards porcine valves. Ann Thorac Surg 1998;66:S44–8.