Etiologic diagnosis of infective endocarditis by broad-range polymerase chain reaction: a 3-year experience.

Abstract

We analyzed surgically resected endocardial specimens from 49 patients by broad-range PCR. PCR results were compared with (1) results of previous blood cultures, (2) results of culture and Gram staining of resected specimens, and (3) clinical data (Duke criteria). Molecular analyses of resected specimens and previous blood cultures showed good overall agreement. However, in 18% of patients with sterile blood cultures, bacterial DNA was found in the resected materials. When data from patients with definite or rejected cases of infective endocarditis (IE) were included, the sensitivity, specificity, and positive and negative predictive values of broad-range PCR were 82.6%, 100%, 100%, and 76.5%, respectively, overall, and 94.1%, 100%, 100%, and 90%, for cases of native valve endocarditis. The sensitivity, specificity, and positive and negative predictive values of culture of resected specimens from patients with native valve endocarditis were 17.6%, 88.9%, 75%, and 36.4%. We recommend broad-range PCR of surgically resected endocardial material in cases of possible IE, in cases of suspected IE in which blood cultures are sterile, and in cases in which organisms grow in blood cultures but only Duke minor criteria are met. We propose to add molecular techniques to the pathologic criteria of the Duke classification scheme.

Full text

Page 1

Broad-Range PCR in Infective Endocarditis • CID 2003:37 (15 July) • 167
M A J O R A R T I C L E
Etiologic Diagnosis of Infective Endocarditis
by Broad-Range Polymerase Chain Reaction:
A 3-Year Experience
Philipp Peter Bosshard,1,aAndreas Kronenberg,2,aReinhard Zbinden,1Christian Ruef,2Erik Christian Bo ¨ttger,1
and Martin Altwegg1
1Institute of Medical Microbiology, University of Zurich, and
of Zurich, Zurich, Switzerland
2Division of Infectious Diseases and Hospital Epidemiology, University Hospital
We analyzed surgically resected endocardial specimens from 49 patients by broad-range PCR. PCR results
were compared with (1) results of previous blood cultures, (2) results of culture and Gram staining of resected
specimens, and (3) clinical data (Duke criteria). Molecular analyses of resected specimens and previous blood
cultures showed good overall agreement. However, in 18% of patients with sterile blood cultures, bacterial
DNA was found in the resected materials. When data from patients with definite or rejected cases of infective
endocarditis (IE) were included, the sensitivity, specificity, and positive and negative predictivevaluesofbroad-
range PCR were 82.6%, 100%, 100%, and 76.5%, respectively, overall, and 94.1%, 100%, 100%, and 90%, for
cases of native valve endocarditis. The sensitivity, specificity, and positive and negative predictive values of
culture of resected specimens from patients with native valve endocarditis were 17.6%, 88.9%, 75%, and 36.4%.
We recommend broad-range PCR of surgically resected endocardial material in cases of possible IE, in cases
of suspected IE in which blood cultures are sterile, and in cases in which organisms grow in blood cultures
but only Duke minor criteria are met. We propose to add molecular techniques to the pathologic criteria of
the Duke classification scheme.
Infective endocarditis (IE), infection of the endocar-
dial surface, is a rare but severe disease. In Europe,
the incidence is ∼6–7 cases/100,000 person-years [1,
2], and the overall mortality rate is 20%–25%, de-
pending on the causative microorganism and thepres-
ence of complications and comorbidities [3]. IE is
usually diagnosed according to the Duke scheme, by
clinical or pathologic criteria (i.e., histologically con-
firmed pathologic lesions or demonstration of a caus-
Received 20 November 2002; accepted 3 March 2003; electronically published
9 July 2003.
Financial support: University of Zurich.
aP.P.B. and A.K. contributed equally to this study.
Reprints or correspondence: Dr. Philipp P. Bosshard, Institute of Medical
Microbiology, University of Zu ¨rich, Gloriastrasse 30, CH-8028 Zu ¨rich, Switzerland
(philboss@immv.unizh.ch).
Clinical Infectious Diseases 2003;37:167–72
? 2003 by the Infectious Diseases Society of America. All rights reserved.
1058-4838/2003/3702-0002$15.00
ative microorganism by culture or histologic exami-
nation in a sample from a vegetation, an embolism,
or an intracardiac abscess) [4, 5]. The Duke clinical
criteria include major criteria (i.e., blood culture pos-
itive for IE, echocardiographic findings consistent
with IE, and valvular regurgitation) and minor criteria
(i.e., predisposition, fever, vascular and immunologic
phenomena, and growth of organism in bloodcultures
but failure to meet a major criterion). In large series,
blood cultures were sterile in 2.5%–31% of cases of
IE [6]. Blood cultures often are sterile as a result of
previous antibiotic therapy [1] or becauseof fastidious
and difficult-to-cultivate microorganisms such as HA-
CEK group bacteria, Bartonella species, Coxiella bur-
netii, Legionella species, Mycoplasma species, and Tro-
pheryma whipplei [7, 8]. In such cases, the etiologic
agent may be identified by serologic and/or immuno-
histologic techniques [9]. Recent case reports have
shown the usefulness of broad-range amplification of

Page 2

168 • CID 2003:37 (15 July) • Bosshard et al.
16S rDNA (referred to as “broad-range PCR”) for identifi-
cation of the causative agent of IE [10–14]. However, only a
few studies so far have addressed the utility of nucleic acid
amplification techniques for diagnosis of IE in a systematic
fashion [15, 16].
For 14 years, we have used broad-range PCR as part of our
microbiological workup procedure for endocardial specimens
[15]. We here report a retrospective study that included 49
patients who were referred for endocardial surgery during Jan-
uary 1998 through March 2001. Our main focus was to com-
pare the results of broad-range PCR with those of standard
microbiological procedures, to evaluate the clinical utility of
broad-range PCR for the diagnosis of IE, and thus to suggest
possible guidelines for the use of broad-range PCR in clinical
practice.
PATIENTS AND METHODS
Patients.
tients (
(University Hospital of Zurich) who had been referred for en-
docardial surgery and from whom specimens had been sub-
mitted to the Institute of Medical Microbiology (University of
Zurich) were included in this study. Clinical and pathologic
data, including blood culture results for up to 6 months before
surgery, were collected fromthemedicalrecords,andcaseswere
classified according to the modified Duke criteria [5]. Forty-
two patients (86%) were male, and 7 (14%) were female; the
mean age (?SD) was 55 years (?17 years; range, 10–83years).
Twenty-two patients had definite IE and 13 patients had pos-
sible IE according to the modified Duke criteria [5]; in 14
patients, surgerywasperformedbecauseofvalvulardysfunction
(IE was excluded on the basis of the Duke criteria). Eighteen
patients (37%) had prosthetic valves before surgery. A total of
63 specimens were obtained and analyzed, 52 of which were
valves (38 aortic valves and 14 mitral valves) and 11 of which
were of other materials, such as aortic vessels (
from the aortic valve ( ), and a vegetation on a pacemakern p 3
( ). In 39patients, bloodsampleswereobtainedforculturen p 1
before surgery. For these patients, on average 6.9 bloodcultures
(range, 1–25 cultures; each blood culture consisted of 1 aerobic
and 1 anaerobic bottle) were done. Eleven additional patients
previously had had IE and underwent surgery because of val-
vular dysfunction (anti-infective therapy had been completed
between 8 days and 8.5 years before surgery).
Cultures.
Cultures were performed as described elsewhere
[15]. Briefly, aerobic and anaerobic blood cultures(BacT/Alert)
were incubated at 37?C for at least 6 days. If growth was found
in blood cultures, aliquots weresubculturedonColumbiasheep
blood agar (Difco) and on brucella sheep blood agar (Becton
Dickinson BBL). Single blood cultures positive for Propioni-
From January 1998 through March 2001, all pa-
) from a tertiary health care center in Zurichn p 49
), swabsn p 7
bacterium acnes or coagulase-negative staphylococci were rated
as contaminated. Parts of endocardial specimens were ground,
prepared for Gram staining, and cultured aerobically and an-
aerobically on Columbia sheep blood agar, chocolate agar (GC
II Agar base with 1% hemoglobin and 1% IsoVitaleX; BBL),
and Brucella sheep blood agar (BBL) for 2–3 days at 37?C. In
addition, pieces of valve tissue were inoculated into thiogly-
colate medium (BBL) and incubated for 10 days.
DNA extraction and broad-range PCR.
and broad-range PCR were performed essentially as described
elsewhere [15]. In brief, small pieces of tissue were digested
with proteinase K in the presence of 0.5% SDS. DNA was
purified with the QIAamp Blood Kit (Qiagen), resulting in a
sample volume of 100 mL. One and 5 mL of the DNA extract
were used for amplification in a total volume of 50 mL con-
taining 1.25 U of AmpliTaq DNA polymerase, LD (Applied
Biosystems). Part of the 16S rDNA (positions 10–806, accord-
ing to Escherichia coli numbering [17]) was amplified with
primers BAK11w (5?-AGTTTGATC[A/C]TGGCTCAG) and
BAK2 (5?-GGACTAC[A/C/T]AGGGTATCTAAT) for 40 cycles.
Insulin-like growth factor II amplification with primers IGF-
P2 (5?-CTTGGACTTTGAGTCA) and IGF-P3 (5?-GGTCGTG-
CCAATTACA) was used as inhibition control.
The amplicons were examined by high-resolution polyacryl-
amide gel electrophoresis (CleanGel; Pharmacia)inwhichsilver
staining was used. Positive samples were recognized by a single
prominent band of appropriate size; negative samples were
those that generated a characteristic banding pattern with sev-
eral bands of smaller size (i.e., 100–200 bp instead of 800 bp)
and low intensity. Bands of proper size were excised and ream-
plified (positions 10–533, according to E. coli numbering) with
primers BAK11w and BAK533r (5?-TTACCGCGGCTGCTGG-
CAC) for 40 cycles using 1.25 U of Taq DNA polymerase(Amer-
shamBiosciences)[15].Forreamplification,apieceoftheexcised
band was added directly to the reamplificationmixwithoutprior
purification of the DNA. To prevent carry-over contamination,
we used physically separate rooms for sample preparation,setup
of PCR mixes, nucleic acid amplification, and analysis of PCR
products. In addition, several negative and positive controls for
DNA extraction and amplification were used.
Occasionally, contaminating DNA was observed bothinneg-
ative controls and in patient samples. Although the contami-
nating DNA produced a band that was similar in size to that
seen in positive samples when high-resolution polyacrylamide
gel electrophoresis was done, this was always less prominent
than in the case of true-positive samples. Sequence analyses
demonstrated that this reagent contamination derived from
waterborne Pseudomonas species.
Sequence analysis.
The nucleic acid sequence of PCR-
amplified gene fragments was determined using automatic
DNA sequencing (ABI Prism 310 Genetic Analyzer [Applied
DNA extraction

Page 3

Broad-Range PCR in Infective Endocarditis • CID 2003:37 (15 July) • 169
Table 1.
culture and microscopic examination of endocardial specimens.
Results of broad-range PCR compared with results of
Specimen status
Results of
broad-range PCR,
no. of specimens
Positive
(n p 25)
Negative
(n p 38)
Culture-positive,
microscopy-negative
Culture-negative,
microscopy-positive
Culture-positive,
microscopy-positive
Culture-negative,
microscopy-negative
3a
3b
7c
1d
00
1534
aIn 1 case, results were contradictory;viridansstreptococciwererecovered
on culture, but broad-range PCR yielded a sequence with 100% identity with
3 different Staphylococcus species (S. arlettae, S. capitis, and S. capprae).
bTwo samples showed growth of Propionibacterium acnes (samples were
from 2 swabs of the aortic valve from the same patient). Coagulase-negative
staphylococci grew on culture of a sample from another patient after enrich-
ment in liquid medium. These 2 patients did not have any other clinical or
pathologic evidence of infective endocarditis, and therefore the PCR results
were considered to be true negative.
cIn all 7 microscopy-positive samples, PCR results were congruent (i.e.,
both diagnostic procedures revealed gram-positive cocci).
dThe results of both PCR and microscopic examination of a second sample
from the same patient were positive.
Table 2.
compared with cultures of previously drawn blood samples.
Results of broad-range PCR in endocardial specimens
Blood culture results
No. of
patients
(n p 49)
Results of
broad-range PCR,
no. of patients
Positive
(n p 23)
Negative
(n p 26)
Positive for any organism
Fulfilled Duke major criteriaa
Did not fulfill Duke major cri-
teria for IE but did fulfill
Duke minor criteria
No organisms grown on culture
Not done
22
18
19
17
3
1b
42
3d
1
2c
17
10
14
9
NOTE.
aMicroorganism typical of IE found in at least 2 separate blood cultures or
microorganism consistent with IE found persistently in blood cultures
bPatient had definite IE, according to Duke criteria (Staphylococcus aureus
was found in 7 of 7 blood cultures). Results of PCR were negative, although
antibiotic therapy was started only 4 days before surgery (false-negativePCR).
cIncludes 1 patient with coagulase-negative staphylococci in 3 of 10 blood
cultures and 1 patient with Propionibacterium acnes in 7 of 21 blood cultures.
PCR results were considered to be false negative in both cases.
dPCR results were considered to be true positive; for more details, see
Results.
IE, infective endocarditis.
Biosystems] and ALF automatic DNA sequencer [Pharmacia]).
Database analysis was done in a 2-step procedure. A first search
was performed with the FASTA algorithm of the GCG Wis-
consin software package (Accelrys). All positions showing dif-
ferences from the best scoring reference sequence were visually
inspected in the electropherogram, and the sequence was cor-
rected if necessary. ThenasecondsearchwasdoneusingBLAST
(http://www.ncbi.nlm.nih.gov/blast). Undetermined nucleo-
tides in the determinedsequenceorthereferencesequencewere
counted as matches. For identification to the genus or species
level, the following criteria were used: 0–2 mismatches with the
sequence entry of an established species was interpreted as spe-
cies identification, and 0–2 mismatches with the sequenceentry
of an unclassified species or 12 mismatches with the highest-
score entry was interpreted as genus identification.
RESULTS
Broad-range PCR versus culture and microscopic examination
ofresectedspecimens.
In2ofthe63specimensweinvestigated,
PCR was inhibited (in these patients, cultures of previously ob-
tained blood samples were sterile). The results of broad-range
PCR were positive in 25 specimens from 23 patients, and 15 of
these specimens were culture and microscopy negative (table 1).
Five sequences were identified at the genus level, and 20 se-
quences could be assigned to a species. Three samples were cul-
ture positive and PCR negative; in these cases, the cultureresults
were considered to be false positive (table 1).
Broad-range PCR versus blood culture.
from whom blood samples were obtained for culture before
surgery (mean number of blood cultures, 6.9; range, 1–25 cul-
tures), 22 had at least 1 positive blood culture (mean number
of positive blood cultures, 5.8; range, 1–19 cultures) (table 2).
The results of PCR in samples from 19 of these 22 patients
were positive (table 3). For 13 patients (68%), both procedures
resulted in identification of the same species; for 6 (32%) of
the 19 patients, PCR yielded a more precise result (e.g., Strep-
tococcus salivarius instead of viridans streptococcus, or an un-
classified oral Streptococcus speciesinsteadofStreptococcusoralis
or Streptococcus sanguis). In the latter case, the obtained se-
quenceexactlymatchedthatofanunclassifiedoralStreptococcus
species sequence but exhibited !98% homology to the most
closely related established species, S. sanguis. The 3 patients for
whom blood culture results were positive but PCR results were
negative had definite IE, according to the Duke criteria; the
PCR results were considered to be false negative (table 2).
For 3 patients, the results of PCR were positive, even though
the results of blood cultures (at least 2 pairs), microscopic
examination, and culture of endocardial specimens were neg-
ative (table 2). In all 3 cases, the results of PCR were considered
to be true positive. For 2 of these patients, IE was diagnosed
as “possible,” according to the Duke criteria; PCR found Hae-
mophilus aphrophilus in one case and Streptococcus bovis in the
other. Both patients were receiving antibiotics at the time that
Of the 39 patients

Page 4

170 • CID 2003:37 (15 July) • Bosshard et al.
Table 3.
with infective endocarditis for whom both methods had positive results.
Organisms identified by blood culture and broad-range PCR in samples from 19 patients
Bacterial group, patientIdentification by broad-range PCR Identification by blood culture
Viridans streptococci
1
2
3
4
5
6 and 7
8
9 and 10
11 and 12
Enterococci
13 and 14
Staphylococci, 15–18
Haemophilus species, 19
Streptococcus australis
Streptococcus bovis
S. bovis/gallolyticus
Streptococcus gordonii
S. mitis
Streptococcus mutans
Streptococcus salivarius
Unclassified oral streptococcus
Unclassified oral streptococcus
Streptococcus mitis
S. bovis
S. bovis
S. gordonii
S. mitis
S. mutans
Viridans streptococcus
S. mitis
Streptococcus oralis/sanguis
Enterococcus faecalis
Staphylococcus aureus
Haemophilus parainfluenzae/paraphrophilus
E. faecalis
S. aureus
H. parainfluenzae
blood samples (10 and 3 pairs) were obtained for culture. In
the third patient, T. whipplei was identifiedbyPCR.Thispatient
underwent surgery for aortic insufficiency. The valvular path-
ology, described as “lymphoplasmacellular inflammation with
formationof granulomasandfibrosis,”wasinterpretedassterile
endocarditis in a patient with seronegative spondylarthritis.
However, 2 different PCRs specific for T. whipplei (targeting a
different part of the 16S rDNA and the ribosomal intergenic
spacer region [18, 19]) were performed on aortic material,
confirming the results of broad-range PCR. These findings
make it unlikely that laboratory contamination occurred, be-
cause for each PCR, a specific positive control is used that only
includes the region of interest. From these data, we concluded
that T. whipplei most probably was the true cause of IE in this
case (true-positive PCR result). The patient has not yet shown
other manifestations of Whipple disease. Cases of Whipple en-
docarditis in the absence of overt gastrointestinal disease [11]
or without other signs of Whipple disease have been reported
elsewhere [10, 20].
Sensitivity, specificity, and positive and negative predictive
values of assays.
According to the Duke criteria, which in-
clude clinical and pathologic features, 22 patients were consid-
ered to havedefiniteIE; 13patients,possibleIE;and14patients,
no (“rejected”) IE. The patient with Whipple endocarditis was
considered to have definite IE, resulting in the inclusion of 23
patients with definite IE and 13 patients with rejected IE in
our study. Broad-range PCR results were positive for 19 of 23
patients with definite IE, 4 of 13 patients with possible IE, and
0 of 13 patients with rejected IE (table 4).
Definite and rejected cases of IE were used as the reference
standard for calculating the sensitivity and specificity of broad-
range PCR. Overall, sensitivitywas82.6%,specificitywas100%,
positive predictive value was 100%, and negative predictive
value was 76.5%. Six patients with definite IE, 8 patients with
possibleIE,and4patientswithrejectedIEhadprostheticvalves.
Broad-range PCR was positive in samples from 16 of 17 of the
patients with definite native valve endocarditis (NVE) but only
in 3 of 6 of the patients with definite prosthetic valve endo-
carditis (PVE). The sensitivity, specificity, and positiveandneg-
ative predictive values of broad-range PCR were 94.1%, 100%,
100%, and 90%, respectively, when the analysis was limited to
cases of NVE and 50%, 100%, 100%, and 42.9% when the
analysis was limited to cases of PVE.
Organisms grew on culture of resected specimens from 3 of
23 patients with definite IE (3 of 17 patients with NVE), 0 of
13 patients with possible IE, and 2 of 13 patients with rejected
IE (1 of 9 cases with NVE and 1 of 4 cases with PVE). The
sensitivity, specificity, and positive and negative predictive val-
ues for culture of resected specimens were 13%, 84.6%, 60%,
and 35.5%, respectively, overall, and 17.6%, 88.9%, 75%, and
36.4% when the analysis was limited to cases of NVE. The
failure to obtain a positive result for specimens from any of
the patients with PVE makes statistical analysis irrelevant but
indicates the limited use of culture techniques in this context.
Patients with previous IE.
viously had had IE. Adequate anti-infective therapy had been
completed between 8 days and 8.5 years before surgery. The
patients underwent surgery because of valvular dysfunction in
the absence of acute signs of endocarditis. In 6 patients, the
results of PCR were positive. In these cases, diagnosis was made
1, 2, 4, 7, and 23 months before surgery. Two of these patients
were intravenous drug users who had multiple episodes of IE
16 months before surgery. The cultures of blood samples from
both patients were positive for Staphylococcus aureus and Strep-
Eleven additional patients pre-

Page 5

Broad-Range PCR in Infective Endocarditis • CID 2003:37 (15 July) • 171
Table 4.
with infective endocarditis (IE) according to the Duke classifi-
cation scheme.
Comparisonof cultureandbroad-rangePCRinpatients
Duke category
Total no.
of patients
(n p 60)
No. of patients
with positive results
of culture
No. of
patients with
positive results
of PCR
(n p 60)
Blood
sample
(n p 49)
20a
2
0a
5a
Resected
specimen
(n p 60)
Definite IE
Possible IE
Rejected IE
Previous IEc
23
13
13
11
1
0
3b
0
19
4
0
6
aBlood cultures were done for only 22 patients with definite IE, 4 with
rejected IE, and 10 with previous IE.
bThese 3 patients did not have any other clinical or pathologic evidence of
IE; culture results were considered to be false positive. Cultures were done
for only 13 patients.
cAnti-infective therapy was suspended for a period ranging from 8 days to
8.5 years before surgery.
tococcus dysgalactiae (in 1 patient, Streptococcus pyogenes was
also found). In both cases, histologic examination revealed
chronic inflammation; PCR in resected specimens identified S.
dysgalactiae (isolated in blood cultures from these 2 patients 7
and 23 months before valve replacement).
DISCUSSION
In this study, we used a combination of broad-range PCR of
16S rDNA and nucleic acid sequencing to identify the causative
agent of IE. We analyzed all patients of a tertiary care hospital
who were referred for endocardial surgery during a 3-year pe-
riod and from whom specimens were submitted to our labo-
ratory. In addition to patients with definite or possible IE,
patients with no suspicion for IE and patients with a previous
episode of IE were included in this study.
Our study demonstrates that broad-range PCR is a useful
tool for the analysis of resected heart valves. For the majority
of PCR-positivespecimens, noorganismswerefoundbyculture
on standard media or by microscopic examination. This is
probably because 90% of the patients had received antibiotics
before surgery. PCR results were in strong agreement with the
results of culture of previously obtained blood samples, and
sequencing provided more-accurate results in 33% of cases
(tables 2 and 3). However, imprecise identification by tradi-
tional techniques did not affect management of patients.None-
theless, for epidemiological reasons, a precise identification of
the causative agent is important.
IE usually is diagnosed by the Duke criteria, which havebeen
accepted by most clinicians [4]. The Duke criteria build on the
earlier published Beth Israel criteria [21] by including echo-
cardiographic findings and the isolation of typical IE pathogens
from blood cultures as major criteria for a clinically definite
categorization of IE. Further modification was proposed by Li
et al. [5] and Fournieretal.[22].Onthebasisofourexperience,
we propose that nucleic acid amplification–baseddetectionand
identification of IE pathogens from endocardial specimens
should be included in the Duke criteria (i.e., a microorganism
can be demonstrated by culture, histologic examination, or
molecular-based techniques to fulfill a pathologic criterion and
thus to contribute to the definite diagnosisof IE). Thisproposal
is supported by the fact that PCR is far more sensitive and
specific than culture of resected specimens, a procedure which
is accepted as a pathologic criterion. More recently, Millar et
al. [16] suggested that molecular-based diagnosis of IE using
blood cultures be included as an additional major criterion
within the Duke clinical criteria. In this study, the authors used
PCR to identifyDNAextractedfrombloodculturesandshowed
the feasibility of identifying the causative agent in cases of
blood culture–negative IE [16].
The present study indicates that, if broad-range PCR of re-
sected specimens is included among the pathologic Duke cri-
teria, cases that otherwise were classified as possible IE may be
classified as definite IE, regardless of the result of culture of
previously obtained blood samples. Four of 13 patients with
possible IE in this study would have been classified as having
definite IE on the basis of a positive PCR result. For the other
9 patients in this group, the negative PCR result didnotexclude
IE. The clinical and pathologic investigations performed in this
study did not allow strict definition of whether these patients
had or did not have IE, and therefore those cases wereclassified
as “possible IE.” Blood culture results that fit the Duke minor
criteria often are associated with some uncertainty; PCR results
from endocardial specimens may verify the causative agent and
aid the clinician in interpretation of culture findings. Broad-
range PCR may be helpful for cases of so-called blood culture–
negative IE [11, 23, 24]. In the present study, PCR made it
possible to diagnose and identify the cause of IE for 23% of
the patients with definite or possible blood culture–negativeIE.
The sensitivity, specificity, and positive and negative predic-
tive values of broad-range PCR were calculated to be 82.6%,
100%, 100%, and 76.5%, respectively. When only cases of NVE
were included in the analysis, the sensitivity and negative pre-
dictive value increased to 94.1% and 90%, respectively. In spite
of the excellent specificity, a positive PCR result always has to
be interpreted within the clinical context; even several months
after therapy for IE has been completed, PCR results may still
be positive. The low sensitivity of 50% found when only cases
of PVE are included in the analysis highlights the difficulty of
obtaining the correct piece of specimen—if mechanical pros-
thetic devices were collected, pieces of adjacent tissue had to
be used for digestion and DNA extraction, and this tissue is
not necessarily involved in infection [25].

Page 6

172 • CID 2003:37 (15 July) • Bosshard et al.
On the basis of the findings of the present study, we propose
to perform broad-range PCR in resected endocardialspecimens
(1) in cases of possible IE, (2) in cases of bloodculture–negative
suspected IE, and (3) in cases of growth in blood cultures that
fulfills the Duke minor criteria to verify the causative agent of
IE. Because of the limited availability of these highlyspecialized
molecular investigations, specimens should be sent to a labo-
ratory with experience in these techniques. The present data
also indicate that nucleic acid amplification techniques do not
yield additional information in cases of definite IE that fulfill
a Duke major criterion and in which microorganisms grow on
blood cultures or in cases with no evidence of IE. A limitation
of broad-range PCR is that it does not provide information
about antimicrobial susceptibility. Thus, routine culture and
susceptibility testing remain important cornerstones of the mi-
crobiological diagnosis of IE.
References
1. Alestig K, Hogevik H, Olaison L. Infective endocarditis: a diagnostic
and therapeutic challenge for the new millennium. Scand J Infect Dis
2000;32:343–56.
2. Horstkotte D. Endocarditis: epidemiology, diagnosis and treatment. Z
Kardiol 2000;89(Suppl 4):2–11.
3. Mylonakis E, Calderwood SB. Infective endocarditis in adults. N Engl
J Med 2001;345:1318–30.
4. Durack DT, Lukes AS, Bright DK. New criteriafordiagnosisofinfective
endocarditis: utilization of specific echocardiographic findings. Duke
Endocarditis Service. Am J Med 1994;96:200–9.
5. Li JS, Sexton DJ, Mick N, et al. Proposed modifications to the Duke
criteria for the diagnosis of infective endocarditis. Clin Infect Dis
2000;30:633–8.
6. Bayer AS, Scheld WM. Cardiovascular infections. In: Mandell GL,Ben-
nett JE, Dolin R, eds. Mandell, Douglas, and Bennett’s principles and
practice of infectious diseases. 5th ed. Philadelphia: Churchill Living-
stone, 2000:857–902.
7. Brouqui P, Raoult D. Endocarditis due to rare and fastidious bacteria.
Clin Microbiol Rev 2001;14:177–207.
8. Millar BC, Altwegg M, Raoult D, Moore JE. Culture-negative endo-
carditis—causes, diagnosis and treatment. Rev Med Microbiol 2000;
11:59–75.
9. Hoen B, Selton-Suty C, Lacassin F, et al. Infective endocarditis in
patients with negative blood cultures: analysis of 88 cases from a one-
year nationwide survey in France. Clin Infect Dis 1995;20:501–6.
10. Geissdo ¨rfer W, Wittmann I, Seitz G, et al. A case of aortic valve disease
associated with Tropheryma whippelii infection in the absence of other
signs of Whipple’s disease. Infection 2001;29:44–7.
11. Gubler JG, Kuster M, Dutly F, et al. Whipple endocarditis without
overt gastrointestinal disease: report of four cases. Ann Intern Med
1999;131:112–6.
12. Jalava J, Kotilainen P, Nikkari S, et al. Use of the polymerase chain
reaction and DNA sequencing for detection of Bartonella quintana in
the aortic valve of a patient withculture-negativeinfectiveendocarditis.
Clin Infect Dis 1995;21:891–6.
13. Wendler D, Mendoza E, Schleiffer T, Zander M, Maier M. Tropheryma
whippelii endocarditis confirmed by polymerase chain reaction. Eur
Heart J 1995;16:424–5.
14. Wilck MB, Wu Y, Howe JG, Crouch JY, Edberg SC. Endocarditiscaused
by culture-negative organisms visible by Brown and Brenn staining:
utility of PCR and DNA sequencing for diagnosis. J Clin Microbiol
2001;39:2025–7.
15. Goldenberger D, Kunzli A, Vogt P, Zbinden R, Altwegg M. Molecular
diagnosis of bacterial endocarditis by broad-range PCR amplification
and direct sequencing. J Clin Microbiol 1997;35:2733–9.
16. Millar B, Moore J, Mallon P, et al. Molecular diagnosis of infective
endocarditis—a new Duke’s criterion. Scand J Infect Dis 2001;33:
673–80.
17. Brosius J, Palmer ML, Kennedy PJ, Noller HF. Complete nucleotide
sequence of a 16S ribosomal RNA gene from Escherichia coli. Proc Natl
Acad Sci USA 1978;75:4801–5.
18. Bra ¨ndle M, Ammann P, Spinas GA, et al. Relapsing Whipple’s disease
presenting with hypopituitarism. Clin Endocrinol (Oxf) 1999;50:
399–403.
19. Hinrikson HP, Dutly F, Nair S, Altwegg M. Detection of three different
types of ‘Tropheryma whippelii’ directly from clinical specimens by
sequencing, single-strandconformationpolymorphism(SSCP)analysis
and type-specific PCR of their 16S–23S ribosomal intergenic spacer
region. Int J Syst Bacteriol 1999;49:1701–6.
20. Celard M, de Gevigney G, Mosnier S, et al. Polymerase chain reaction
analysis for diagnosis of Tropheryma whippelii infective endocarditisin
two patients with nopreviousevidenceof Whipple’sdisease.ClinInfect
Dis 1999;29:1348–9.
21. von Reyn CF, Levy BS, Arbeit RD, Friedland G, Crumpacker CS. In-
fective endocarditis: an analysis based on strict case definitions. Ann
Intern Med 1981;94:505–18.
22. Fournier PE, Casalta JP, Habib G, Messana T, Raoult D. Modification
of the diagnostic criteria proposed by the Duke Endocarditis Service
to permit improved diagnosis of Q fever endocarditis. Am J Med
1996;100:629–33.
23. Raoult D, Fournier PE, Drancourt M, et al. Diagnosis of 22 new cases
of Bartonella endocarditis. Ann Intern Med 1996;125:646–52.
24. Yuasa Y, Yoshiie K, Takasaki T, Yoshida H, Oda H. Retrospectivesurvey
of chronic Q fever in Japan by using PCR to detect Coxiella burnetii
DNA in paraffin-embedded clinical samples. J Clin Microbiol 1996;
34:824–7.
25. Karchmer AW. Infections of prosthetic valves and intravasculardevices.
In: Mandell GL, Bennett JE, Dolin R, eds. Mandell, Douglas, and
Bennett’s principles and practice of infectious diseases. 5th ed. Phila-
delphia: Churchill Livingstone, 2000:903–17.