ISSN: 1524-4539
Copyright © 2010 American Heart Association. All rights reserved. Print ISSN: 0009-7322. Online
72514
Circulation is published by the American Heart Association. 7272 Greenville Avenue, Dallas, TX
DOI: 10.1161/CIRCULATIONAHA.108.773598
2010;121;1141-1152 Circulation
Bernard D. Prendergast and Pilar Tornos
Surgery for Infective Endocarditis: Who and When?
http://circ.ahajournals.org/cgi/content/full/121/9/1141
located on the World Wide Web at:
The online version of this article, along with updated information and services, is
http://www.lww.com/reprints
Reprints: Information about reprints can be found online at
journalpermissions@lww.com
410-528-8550. E-mail:
Fax:Kluwer Health, 351 West Camden Street, Baltimore, MD 21202-2436. Phone: 410-528-4050.
Permissions: Permissions & Rights Desk, Lippincott Williams & Wilkins, a division of Wolters
http://circ.ahajournals.org/subscriptions/
Subscriptions: Information about subscribing to Circulation is online at
by on June 3, 2011 circ.ahajournals.orgDownloaded from
Surgery for Infective Endocarditis
Who and When?
Bernard D. Prendergast, DM, FRCP; Pilar Tornos, MD, FESC
Infective endocarditis (IE) remains a dangerous conditionwith unchanging incidence and a mortality approaching
30% at 1 year.1,2 Surgery is potentially lifesaving3 and is
required in 25% to 50% of cases during acute infection and
20% to 40% during convalescence.4–7 Operative procedures
are often technically difficult and associated with high risk,
not least because patients are frequently extremely sick with
multisystem disease. Nevertheless, indications for surgery are
clear in many patients, and international guidelines8,9 provide
strong recommendations that are applicable for the majority.
These guidelines are not supported by robust clinical evi-
dence, however, and clinical decision making is often ham-
pered by diverse considerations, including advancing age of
the overall patient cohort, the presence of extracardiac com-
plications or preexistent comorbidity, prior antibiotic therapy
of varying duration, and the availability of appropriate
surgical expertise. In this article, we review the evidence base
that supports current clinical practice and attempt to provide
recommendations in areas where doubt persists.
The Role of Surgery
The role of surgery in active IE has expanded progressively
since early reports of successful outcome.10 Subsequent
declines in mortality may be attributed to a variety of
improvements in management, although expeditious surgery
in carefully selected patients has played a major role. Con-
temporary data in Europe indicate that surgery is now
undertaken in approximately 50% of patients with IE; the
most frequent indications are congestive heart failure (60%),
refractory sepsis (40%), embolic complications (18%), and
vegetation size (48%), with a combination of these factors
being present in many patients.11
Overall surgical mortality in active IE is 6% to 25%, with
long-term survival rates of approximately 70% in most
series.4–7,12–31 Assessment of the impact of surgery on out-
come is difficult, because the patients referred are frequently
those with severe complications related to virulent organisms.
Conversely, the most ill patients (frequently the elderly, with
attendant comorbidity) are often deemed unfit for surgery. In
general terms, prognosis is better if surgery is undertaken
early, before cardiac tissue destruction and deterioration in
the overall condition of the patient increase the hazards of
intervention. Final outcome has never been related to the
duration and intensity of prior antibiotic treatment, and
surgery should not be delayed when clearly indicated in the
vain hope that a sterile operative field can be achieved.
Numerous series have attempted to identify variables
predictive of early and late mortality,4–6,12–25 and these are
summarized in Table 1. However, interpretation is hampered
by the heterogeneous nature of the patients studied and the
outcome measures used. Specifically, the vast majority of
studies have been single-center, retrospective series enrolling
patients with both native and prosthetic valve IE. Further-
more, analysis is inherently biased given the selection of
patients for surgery who have an anticipated poor outcome
but acceptable operative risk. Although surgery may be
recommended and commonly performed for indications such
as embolic complications or persistent infection, it should be
recognized that no definitive proof exists of improved out-
come in these situations (in contrast to congestive heart
failure due to valvular regurgitation; see below). More recent
investigations using sophisticated propensity scoring models
have yielded conflicting results on the benefits of sur-
gery,15,22–25 and a future need exists for high-quality prospec-
tive assessment. Even with the help of these data, decision
making is frequently difficult, and overall management is
highly dependent on the experience of the surgical team (as
well as the individual surgeon) and a strong interaction with
cardiology and microbiology colleagues.
Surgical Management
Preoperative Considerations
Transthoracic and transesophageal echocardiography are now
ubiquitous, and their utility in the assessment of complica-
tions of IE is clearly recognized. Transesophageal imaging
has superior sensitivity and specificity and is recommended
in virtually all patients in whom surgery is contemplated.
Cardiac catheterization is unnecessary for diagnosis and
anatomic characterization. Coronary angiography should be
From the Departments of Cardiology (B.D.P.), The John Radcliffe Hospital, Oxford, United Kingdom and Hopital Vall D’Hebron (P.T.), Barcelona,
Spain.
Correspondence to Dr B.D. Prendergast, Consultant Cardiologist, The John Radcliffe Hospital, Headley Way, Headington, Oxford, OX3 9DU, United
Kingdom. E-mail bernard.prendergast@orh.nhs.uk
(Circulation. 2010;121:1141-1152.)
© 2010 American Heart Association, Inc.
Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.108.773598
1141
Valvular Heart Disease: Changing Concepts in
Disease Management
by on June 3, 2011 circ.ahajournals.orgDownloaded from
Table 1. Key Articles Assessing Surgical Outcome in IE
First Author
and
Reference
Citation Year Study Design
Statistical
Methods Setting
No. of
Subjects
Patient
Characteristics
In-Hospital
Mortality,
%
Mean
Follow-Up,
y
Long-
Term
Survival
Rate, % Summary of Findings
Jault4 1997 Retrospective
single-center
surgical
cohort study
Multivariate
logistic
regression
analysis
France 247 Native valve IE
alone; surgery
100%
7.6
(Surgical
series)
6 71 Predictors of operative mortality:
age, insidious illness, CHF.
Long-term survival good except
for neurological complications
and mitral valve IE.
Castillo5 2000 Prospective
single-center
cohort study
Simple
group
comparisons
Spain 138 Native valve IE
69%, PVE
31%; surgery
51%
Overall 21;
surgical
21,
medical 20
(P�NS)
10 71 Early surgery associated with
good long-term results and no
increase in mortality. Not a
comparative study.
Alexiou12 2000 Retrospective
single-center
surgical
cohort study
Multivariate
logistic
regression
analysis
UK 118 Native valve IE
70%, PVE
30%; surgery
100%
7.6
(Surgical
series)
10 73 Predictors of operative mortality:
CHF, impaired LV function.
Predictors of recurrence: PVE.
Predictors of late mortality:
myocardial invasion,
reoperation. Predictors of poor
long-term survival:
coagulase-negative
staphylococcus, annular
abscess, long ICU stay.
Wallace13 2002 Retrospective
single-center
cohort study
Multivariate
logistic
regression
analysis
UK 208 Native valve IE
68%, PVE
32%; surgery
52%
Overall 18;
impact of
surgery
not
reported
0.5 73 Duration of illness, age, gender,
site of infection, organism, and
LV function did not predict
outcome. Abnormal white cell
count, raised creatinine, �2
Duke criteria, or visible
vegetation conferred poor
prognosis.
Hasbun14 2003 Retrospective
multicenter
cohort study
Multivariate
logistic
regression
analysis
USA 513 Native valve IE
alone; surgery
45%
Not
reported
0.5 74 Mortality associated with
comorbidity, abnormal mental
status, CHF, nonstreptococcal
IE, or medical therapy.
Prognostic classification
proposed.
Vikram15 2003 Retrospective
multicenter
cohort study
Propensity
analysis
USA 513 Native valve IE
alone; surgery
45%
Not
reported
0.5 74 Valve surgery associated with
reduced mortality after
adjustment for baseline
variables and propensity scores.
Benefits of surgery greatest in
patients with CHF.
Habib16 2005 Retrospective
multicenter
cohort study
Multivariate
logistic
regression
analysis
France 104 PVE alone;
surgery 49%
Overall 21;
surgical
17,
medical 25
(P�NS)
2.7 62 Predictors of in-hospital
mortality: CHF, S aureus.
Predictors of long-term
mortality: early PVE,
comorbidity, CHF,
staphylococcal infection, new
prosthetic dehiscence. Mortality
reduced by surgery in high-risk
subgroups with staphylococcal
infection and complicated PVE.
Delahaye17 2007 Prospective
multicenter
population-
based survey
Multivariate
logistic
regression
analysis
France 559 Native valve IE
85%, PVE
15%; surgery
47%
Overall 17;
surgical
14,
medical 19
(P�NS)
Not
reported
Not
reported
Predictors of mortality: CHF,
immunosuppression, insulin
dependent DM, left-sided IE,
septic shock, coma, cerebral
hemorrhage, high C-reactive
protein.
(Continued)
1142 Circulation March 9, 2010
by on June 3, 2011 circ.ahajournals.orgDownloaded from
Table 1. Continued
First Author
and
Reference
Citation Year Study Design
Statistical
Methods Setting
No. of
Subjects
Patient
Characteristics
In-Hospital
Mortality,
%
Mean
Follow-Up,
y
Long-
Term
Survival
Rate, % Summary of Findings
San
Román18
2007 Prospective
multicenter
cohort study
Multivariate
logistic
regression
analysis
Spain 317 Native valve IE
64%, PVE
36%; surgery
28%
Overall 21 Not
reported
Not
reported
Predictors of high risk:
interhospital transfer, AV block,
acute onset, CHF, periannular
complications, S aureus
infection.
Revilla19 2007 Prospective
multicenter
cohort study
Multivariate
logistic
regression
analysis
Spain 508 Native valve IE
66%, PVE
34%; surgery
100%
Overall 36;
native
valve 32,
PVE 45
(surgical
series)
Not
reported
Not
reported
Poor clinical outcome after
urgent surgery. Persistent
infection and renal failure
associated with higher mortality.
Hill20 2007 Prospective
single-center
cohort study
Multivariate
logistic
regression
analysis
Belgium 193 Native valve IE
66%, PVE
34%; surgery
63%
Not
reported
0.5 78
Overall
(26 if CI
to
surgery)
Predictors of mortality: age, S
aureus, CI to surgery (present in
50% of deaths).
Remadi21 2007 Prospective
multicenter
cohort study
Multivariate
logistic
regression
analysis
France 116 S aureus IE
alone; native
valve IE 83%,
PVE 17%;
surgery 47%
Overall 26;
surgical
16,
medical 34
(P�0.05)
3 57 Predictors of mortality:
comorbidity, CHF, severe sepsis,
PVE, major neurological events.
Early surgery associated with
improved outcome.
Wang6 2007 Prospective
multicenter
cohort study
Multivariate
logistic
regression
analysis
Global
registry
556 PVE alone;
surgery 49%
Overall 23;
impact of
surgery
not
reported
Not
reported
Not
reported
Predictors of in-hospital
mortality: age,
healthcare-associated infection,
S aureus infection, CHF, stroke,
intracardiac abscess, persistent
bacteremia.
Aksoy22 2007 Prospective
single-center
cohort study
Propensity
score
matching:
logistic
regression
analysis
USA 426 Native valve IE
69%, PVE
19%, “other”
12%; surgery
29%
Overall 17;
left-sided
IE: surgical
12,
medical 18
5 Surgical
�48,
medical
�28
Factors associated with surgical
treatment: age, interhospital
transfer, staphylococcal
infection, CHF, intracardiac
abscess, hemodialysis with IV
catheter. Surgery associated
with long-term benefit. Factors
associated with mortality: DM,
paravalvular infection, indwelling
IV catheter.
Tleyjeh23 2007 Retrospective
single-center
cohort study
Matched
propensity
analysis
USA 546 Native valve IE
alone; surgery
24%
Not
reported
0.5 Surgical
73,
medical
76
No survival benefits associated
with surgery despite correction
for timing and early operative
deaths. Prospective study
recommended.
Tleyjeh24 2008 Retrospective
single-center
cohort study
Propensity
analysis
USA 546 Native valve IE
alone; surgery
24%
Not
reported
0.5 Surgical
73,
medical
76
Strong correlation between
propensity score and timing of
surgery. Individual effect of
each variable difficult to
measure.
Thuny25 2009 Retrospective
single center
observational
cohort study
Propensity
analysis
France 291 Native valve IE
82%, PVE
18%; surgery
100%
Not
reported
6 months 13% Very early surgery (�7 days)
associated with improved
survival (especially in highest
risk patients) but greater
likelihood of relapse or
post-operative valve dysfunction.
CHF indicates congestive heart failure; PVE, prosthetic valve endocarditis; NS, not significant; UK, United Kingdom; LV, left ventricle; ICU, intensive care unit; USA,
United States; CI, contraindication; IV, intravenous; DM, diabetes mellitus; and AV, atrioventricular.
Prendergast and Tornos Surgery for Infective Endocarditis 1143
by on June 3, 2011 circ.ahajournals.orgDownloaded from
considered in men �40 years of age, postmenopausal
women, and those with a history of ischemic heart disease or
an adverse risk factor profile, although caution is necessary in
the presence of large aortic vegetations that may be dislodged
by catheter manipulation. Alternative noninvasive techniques
such as multislice CT or MRI may be used if available.
Oral anticoagulation is associated with adverse outcome in
IE, particularly during the first 2 weeks, when embolic risk is
highest and surgical decisions are usually necessary.32 Anti-
coagulants should be discontinued in this phase if possible.
If essential (for example, in those with prosthetic valve
endocarditis that affects a mechanical valve), then a switch
to intravenous unfractionated heparin is recommended.
Whenever possible, the primary source of infection re-
sponsible for IE should be eliminated before cardiac
surgical intervention.
Intraoperative Management
Imaging
The operating table is an inappropriate place for cardiac
diagnosis, and full anatomic characterization is required
before surgery is undertaken. Perioperative transesophageal
echocardiography is a useful tool to determine the exact
location and extent of infection,33 assist the choice of recon-
struction procedure, validate the surgical result, and guide
perioperative hemodynamic management.
Microbiology
The excised native or prosthetic valve should be sent to the
microbiology laboratory in physiological saline for immedi-
ate culture. A Gram’s stain is useful for immediate diagnosis,
and molecular examination of excised valve tissue may play
a role, particularly in culture-negative patients.34
Surgical Technique
The 2 primary objectives of surgery are control of infection
and reconstruction of cardiac morphology.35 The mode of
surgery (replacement versus repair) or type of prosthesis
used (mechanical versus biological) has no influence on
operative mortality,36 although repair techniques, when
applicable, offer long-term advantages, including a re-
duced risk of late complications (notably, recurrent IE) and
obviation of the need for lifelong anticoagulation.37 Ho-
mografts offer a reduced risk of recurrent infection in aortic
IE,38 although their use remains controversial owing to a
higher risk of late complications.39 Cardiac transplantation
may be considered in extreme cases with recurrent prosthetic
valve endocarditis.40
Indications for Surgery
Tables 2 and 3 present information on indications for and
timing of surgery, and the Figure presents surgical indications
in native valve IE. However, it is important to remember that
no randomized controlled trials are available to guide current
practice. Nor are they likely; such studies would be difficult
to perform owing to the diversity of patients with IE, the
relatively small numbers available for study, and potential
ethical concerns if surgery were to be withheld in control
patients. What available evidence, therefore, guides current
practice and international recommendations?
Congestive Heart Failure Due to
Valvular Regurgitation
Among the complications of IE, congestive heart failure has
the greatest impact on prognosis.14,15 Studies from the 1970s
and 1980s compared medical and surgical treatment of
congestive heart failure complicating IE and demonstrated a
consistent reduction in mortality after surgery, from a range
of 56% to 86% to a range of 11% to 35%.41,42 Although no
correction was made for underlying comorbidity, similar
results were reported in 2 more recent Scandinavian stud-
ies,26,43 in which the best results were obtained with early
intervention within 1 week of presentation. Early valve
surgery in patients with congestive heart failure is associated
with a substantial reduction in mortality compared with
medical therapy alone,5,15,21,22,44 and this scenario is now the
most common and clearest indication for surgery, being
present in 72% of patients who underwent early intervention
in a recent European series.19
Congestive heart failure is usually the result of valvular
regurgitation, which may develop acutely as a result of
perforation of a native valve or bioprosthetic valve leaflet or
rupture of infected mitral chordae. Rarer causes include valve
obstruction by bulky vegetations and sudden intracardiac
shunts from fistulous tracts or prosthetic dehiscence. Acute
aortic regurgitation is poorly tolerated and usually rapidly
progressive. Urgent surgery is indicated regardless of the
status of the infection, particularly when evidence of pulmo-
nary edema or cardiogenic shock exists, or echocardiographic
evidence of rapidly rising left ventricular end-diastolic pres-
sure (premature closure of the mitral valve). Acute mitral
regurgitation may be better tolerated as a consequence of
offloading into the left atrium and pulmonary bed; however,
acute pulmonary edema may develop as a result of the rapid
increase in left atrial pressure, and urgent surgical interven-
tion is frequently required.
Patients with less dramatic presentation may respond well
to initial medical therapy with diuretics and short-term
afterload reduction with vasodilator therapy. No clear evi-
dence exists to guide management strategy in this situation,
and many physicians and surgeons are reluctant to recom-
mend early surgery. Thus, in the absence of other immediate
indications, intervention may be postponed in the short term
to allow a brief period of antibiotic therapy under careful
clinical and echocardiographic observation. Conversely, mild
congestive heart failure at initial presentation may progress
insidiously despite appropriate antibiotic therapy, usually
within the first month of treatment. Delayed surgery in these
circumstances is unacceptable, being associated with a dra-
matic rise in operative mortality45 as a consequence of
progressive cardiac decompensation and exposure of the
patient to secondary risks of the disease and its treatment.
In the occasional patient with well-tolerated valvular re-
gurgitation and no other indication for surgery, medical
management with antibiotics is recommended under strict
clinical and echocardiographic control. Surgery can then be
considered at a later stage after healing of the infection,
depending on tolerance of the valve lesion. This strategy may
be particularly attractive in the elderly comorbid patient who
presents extreme surgical risk and in the young patient for
1144 Circulation March 9, 2010
by on June 3, 2011 circ.ahajournals.orgDownloaded from
whom valve replacement may