Diagnosis of infectious endocarditis in patients undergoing valve surgery

The American Journal of Medicine (2005) 118, 230 –238

CLINICAL RESEARCH STUDY

Diagnosis of infectious endocarditis in patients undergoing valve surgery Gilbert Greub,a Hubert Lepidi,a Clarisse Rovery,a Jean-Paul Casalta,a Gilbert Habib,b Frédéric Collard,c Pierre-Edouard Fournier,a and Didier Raoulta a

Unité des Rickettsies, Faculté de Médecine, Université de la Méditerranée, Marseille, France; Service de cardiologie, and cService de chirurgie cardiaque, Hôpital de la Timone, Marseille, France.

b

KEYWORDS: Endocarditis; Etiological diagnosis; valve analysis; histopathology; polymerase chain reaction

PURPOSE: Histologic examination of valve samples is considered as the gold standard for the diagnosis of infectious endocarditis. Molecular tools are also very promising for patients with negative-culture endocarditis. Thus, we studied the contribution of valvular histology, culture, and 16S rRNA PCR amplification plus sequencing to the diagnosis of infectious endocarditis in patients undergoing valve surgery. SUBJECTS AND METHODS: We performed culture, histological examination, and broad-range PCR amplification plus sequencing on valve samples taken from 127 patients with infectious endocarditis and from 118 patients without endocarditis. The sensitivity and specificity of these tests for the diagnosis of endocarditis in patients undergoing valve surgery were studied. RESULTS: The sensitivity of PCR was of 61% (64/105) whereas that of histological examination was of 63% (62/98) and that of valve culture was of only 13% (14/105). All 68 positive PCR results considered reliable according to an interpretation scheme were from patients with infectious endocarditis, resulting in a 100% (118/118) specificity of the interpreted molecular approach. The specificity of histology was also of 100% (118/118) when using stringent criteria (ie, presence of vegetation, microorganisms, and/or valvular inflammation with mainly polymorphonuclear cells). PCR identified an etiological agent in 38% (5/13) of definite blood culture-negative infectious endocarditis. CONCLUSION: We show that valvular histology with stringent criteria is the gold standard for the diagnosis of infectious endocarditis. Broad-range amplification of 16S rRNA gene is indicated for infectious endocarditis of unknown etiology, whereas valve culture is of limited sensitivity. © 2005 Elsevier Inc. All rights reserved.

Bacteremia is a common feature of infectious endocarditis, and blood cultures are considered one of the most important tools to make a diagnosis of endocarditis.1 However, blood cultures fail to isolate an etiologic agent in 2.5 to 31% of cases.2– 4 Culture-negative endocarditis is often Requests for reprints should be addressed to Gilbert Greub, Institute of Microbiology, Faculty of Biology and Medicine, Lausanne, Switzerland. E-mail: [email protected]

0002-9343/$ -see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.amjmed.2004.12.014

associated with antibiotic use within the previous 2 weeks,5 or less frequently caused by intracellular pathogen that are not detected using standard culture approaches.6,7 Use of alternative laboratory diagnostic methods are thus needed, including pathologic examination of valvular samples and immunohistology,8 serology,9 and molecular tools (reviewed in 10,11). Histologic examination of valve samples is generally considered as the gold standard and as a criterion of definite

Greub et al

Diagnosis of Infectious Endocarditis

endocarditis.8,11–14 Molecular tools also appear promising. In one study, positive valve PCR results were obtained for 315 and 1916 patients with negative-culture endocarditis. Nevertheless, the validity of a PCR result may be impaired by false negative and false positive results. Thus, cautious interpretation of PCR results is mandatory. Although contamination is a well documented limitation of PCR based molecular diagnostic approaches,17 no criteria for interpretation of PCR results have been established and the prevalence of different types of PCR contamination remained undefined. Here, we compare the results of histological examination, culture and PCR amplification and nucleic acid sequencing on valve samples taken from 127 patients with endocarditis and from 118 patients without endocarditis. Their respective contribution to the diagnosis of infectious endocarditis in patients undergoing valve surgery was analysed.

231

This study, approved by the ethical committee of Marseilles University, included cardiac valves excised at Marseilles University Hospitals between April 1, 1994 and September 30, 2002 from patients with definite and possible endocarditis, as determined prior to valve surgery using the modified Duke criteria.14 Patients without endocarditis who underwent valve replacement for hemodynamic reasons and who were without clinical signs or symptoms of infection and with negative laboratory inflammatory markers prior to surgery were used as controls. Valves for which no frozen material was available for PCR testing or for histological examination were excluded. Demographic data, nature of the valve, clinical presentation, and echocardiographic findings were prospectively recorded using a standardized questionnaire. When available, sera were tested for the presence of rheumatoid factor (Rapitex RF, Dade Behring Inc., Newark, NJ) and for antibodies against intracellular bacteria, including Coxiella burnetii, Bartonella spp., and Mycoplasma pneumoniae.10 Blood cultures were performed using the Bactec (Becton-Dickinson, Sparks, MD) blood cultures automated system.18

inflammation with a predominance of polymorphonuclear cells were present, “compatible lesions” when only a nonspecific valvular inflammation was present, and “no suggestive lesions” when there was a complete absence of inflammation; or in case of inflammation, when degenerative valvular changes, myxoid changes or rheumatic valvular disease were present11 (Figure 1). DNA extraction, PCR and sequencing technologies evolved during the study period. DNA was extracted using different kits, including Chelex Resin (BioRad, Hercules, Ca), QIAamp DNA Mini-Kit (Quiagen, Courtaboeuf, France), MagNAPure LC DNA Isolation Kit III (Roche Biochemicals, Mannheim, Germany), or FastDNA kit (Bio101, Carlsbad, CA, USA). PCR amplification and sequencing were performed as described,21 using primers mentioned in Table 1. To decrease the likelihood of DNA contamination, DNA extraction, preparation of the PCR master mixture, and amplification were carried out in different rooms. In presence of a negative PCR result in the presence of definitive endocarditis using clinical criteria, additional PCR assays were performed. As a consequence, the person performing the PCR was not blinded to the clinical information. Positive PCR products were sequenced, and sequences were compared to those available in GenBank.22 A PCR positive valve sample taken from a patient with proven Staphylococcus aureus endocarditis was used as a positive control, whereas sterile water and PCR negative valve samples taken from patients without endocarditis were used as negative controls. Positive PCR results were classified as reliable or nonreliable according to the results of the positive and negative controls and to previous or concomitant PCR plus sequencing results. Reliable positive PCR results were considered as certain when direct microbiological examination of valve tissues was compatible with the PCR result, and possible in all other cases (Figure 2). In the presence of a possible positive result, we performed additional PCR targeting a second gene, and when negative a third gene, with primers specific for the genus (Table 1). When the PCR was positive and the sequence gave the same result, it was reclassified as certain. When both PCR targeting a second and a third gene were negative, the result was reclassified as negative (Figure 2).

Valve examination

Statistical analysis

A part of each valve sample was inoculated onto blood and chocolate agar (bioMérieux, Marcy-l’Etoile, France). Valves taken from patients with blood culture negative endocarditis were also inoculated onto different cell lines, as described.18,19 Isolated strains were identified using a phenotypic and, when needed, a genotypic approach.20 All valves were examined by the same pathologist (Hubert Lepidi), who was provided with limited clinical information. Histologic findings were classified as “typical lesions” when vegetations, microorganisms, and/or valvular

Using Chi-squared and Mann-Whitney tests, we compared demographic, clinical, and laboratory data according to the presence of endocarditis. Sensitivity, specificity and positive and negative predictive values of each method for the diagnosis of endocarditis were assessed using the modified Duke criteria,14 which were calculated prior to valve surgery for histology, and after surgery (including pathologic criteria of definite endocarditis) for culture and PCR. STATA software (version 7.0, Stata Corporation, College Station, TX) was used for all analyses.

Methods

232

The American Journal of Medicine, Vol 118, No 3, March 2005 (77%) were classified prior to surgery as definite , and 29 (23%) as possible. Among the 29 possible cases of endocarditis, 7 valves (24%) fulfilled definite pathologic criteria for endocarditis (ie “typical lesions”) and were reclassified as definite endocarditis. Characteristics of these 245 excised valves are presented in Table 2.

Histologic examination

Figure 1 (A) Typical lesions: section of valve from a case of staphylococcal endocarditis. There is extensive destruction of valve tissues with a large vegetation (*) and dense inflammatory infiltration with numerous neutrophils (arrowhead) (hematoxylineosin-saffron, original magnification X100). (B) Compatible lesions: degenerative valvular lesions with dense inflammatory infiltrates, composed mainly of macrophages and lymphocytes, and degenerative changes (hematoxylin-eosin-saffron, original magnification X400). (C) No suggestive lesions: degenerative valvular lesions with fibrosis (arrowhead) and calcifications (*). Note the absence of inflammatory infiltrates in valvular tissues (hematoxylin-eosin-saffron, original magnification X40).

Results Patients and samples A total of 245 heart valves from 238 patients were analyzed. Of these 245 valves, 127 were obtained from patients with endocarditis and 118 from patients without endocarditis. Among the 127 episodes of endocarditis, 98

Among the 98 patients with definite endocarditis according to the modified Duke criteria prior to valve surgery, 62 (63%) were histologically classified as “typical lesions”, 11 (11%) as “compatible lesions”, and 25 (26%) as “no suggestive lesions” whereas among the 29 patients with possible endocarditis, 7 (24%) were histologically classified as “typical lesions”, 10 (35%) as “compatible lesions”, and 12 (41%) as “no suggestive lesions”. Thus, when only the presence of “typical lesions” was considered a pathologic criterion of infectious endocarditis, the sensitivity of histology for the diagnostis of infectious endocarditis was 63% (62/98), (Table 3). Sensitivity was not influenced by antibiotic therapy prior to surgery (6/11 versus 56/87, P ⫽ .52). Sensitivity was higher for native valves (49/67 ⫽ 73%) than for prosthetic valves (13/31 ⫽ 42%, P ⫽ .003), but there was no difference between cases of blood culture-positive (55/85 ⫽ 65%) and blood culture-negative endocarditis (7/13 ⫽ 54%, P ⫽ .45). The specificity of histology with “typical lesions” for the diagnosis of infectious endocarditis was 100% (118/118). When the presence of any evidence of inflammation was considered a pathologic criterion of infectious endocarditis, i.e. classifying both “typical lesions” and “compatible lesions” at histology as positive, the sensitivity increased to 74.2% (72/97) but the specificity decreased to 93.5% (113/118).

Valve culture A microorganism was isolated by culture from 14 (13%) of 105 valves taken from patients with definite endocarditis. Isolated microorganisms included S. aureus (4 cases), coagulase negative Staphylococci (2), Streptococci spp. (3), Enterococcus sp. (1), and 1 each, respectively, of Enterobacter cloacae, Acremonium sp., Aspergillus sp., and Candida albicans.Valve culture was congruent with blood cultures in 11 instance and provided an etiological diagnosis in 2 cases, which were both confirmed by PCR plus sequencing (Streptococcus sp. and Aspergillus sp.). In a third case, coagulase-negative Staphylococci was considered to be a contaminant. Among the 118 patients without endocarditis, valve culture was positive in 2 patients (Escherichia coliand coagulase-negative Staphylococci). The specificity of valve culture was thus 98% (116/118) (Table 3).

Greub et al

Diagnosis of Infectious Endocarditis

233

Table 1 Primers used for broad-range 16S rRNA PCR and, according to species identified by sequencing, primers targeting a second gene for confirmation of positive 16S rRNA PCR results. Bacteria

Gene

Eubacterial

rrs

Staphylococcus spp. Streptococcus spp.‡ Enterococcus spp. Streptococcus spp.‡,§ Enterococcus spp.§ Enterobacteriaceae Enterobacteriaceae Mycoplasma hominis Coxiella burnetii Bartonella spp. Bartonella spp. Bartonella spp.#

RpoB RpoB RpoB SOD SOD rpoB fliC fisY IS111 ITS Pap rib 1 rib 2 18S rRNA ITS ap**

Fungi Aspergillus sp.

Forward primer

Reverse primer

536f 5=CAGCAGCCGCGGTAATAC or fD1 5= AGAGTTTGATCCTGGCTCAG* StphF 5=AAACCIATACGGAATTGGTT† StrpF 5=AARYTIGGMCCTGAAGAAAT† StrpF 5=AARYTIGGMCCTGAAGAAAT† d1 5=CCITAYICITAYGAYGCIYTIGARCC† d1 5= CCITAYICITAYGAYGCIYTIGARCC† CM7 5=AACCAGTTCCGCGTTGGCCTGC EcoH1 5=AATACCAACAGCCTCTCGCT¶ MH1F 5=GTGTTGTATCGACAACAG Trans3 5=CAACTGTGTGGAATTGATGA UR1 5=CTTCGTTTCTCTTTCTTCA Pap1 5=CTTTAATGACGACTTCTGTT ZRib1F 5=CGGATATCGGTTGTGTTGAA Zrib2F 5=GCATCAATTGCGTGTTCA NS1 5=GTAGTCATATGCTTGTCTC NS3 5=GCAAGTCTGGTGCCAGCAGCC ITS1 5=TCCGTAGGTGAACCTGCGG apF 5=AGCACCGACTACATCTAC

1050r 5=CACGAGCTGACGACA or rP2 5= ACGGCTACCTTGTTACGACTT* StphR 5=GTTTCATGACTTGGGACGG† StrpR 5=TGIARTTTRTCATCAACCATGTC† StrpR 5=TGIARTTTRTCATCAACCATGTC† d2 5=ARRTARTAIGCRTGYTCCCAIACRTC† d2 5= ARRTARTAIGCRTGYTCCCAIACRTC† CM31b 5=CCTGAACAACACGCTCGGA EcoH2 5=AGAGACAGAACCTGCTGC¶ MH1R 5=GTGTTGTATCGACAA Trans5 5=TTTACATGACGCAATAGCGC UR2 5=CTTCTCTTCACAATTTCAAT pap4 5=CCGAAATCTGAGTAACGG TA ZRib1R 5=CATCAATRTGACCAGAAACCA Zrib2R 5=CCCATTTCATCACCCAAT NS4 5=CTTCCGTCAATTCCTTTAAG NS8 5=TCCGCAGGTTCACCTACGGA ITS2 5=GCTGCGTTCTTCATCGATGC apR 5=GAGATGGTGTTGGTGGC

Described by Mollet et al.34 # Suicide one-step nested PCR.31 *Described by Weisburg et al.32 †I⫽inosine, R⫽purine (AG), Y⫽pyrimidine (CT), M⫽amino (AC). ‡Including S. bovis, Abiotrophia spp. and Granulicatela spp. §Described by Poyart et al.33 ¶Described by Reid et al.35 **Alkaline protease encoding gene, primers described by Raad et al.36

PCR amplification of 16S rRNA gene and sequencing A total of 330 PCR amplifications targeting the 16S rRNA gene were performed on 245 excised valves. A total of 153 PCR amplifications were positive from 114 valvular samples. Of the 114 samples that tested positive by PCR, 68 were classified as reliable (Table 4). Table 5 shows the organisms that were identified by PCR and sequencing. An etiological diagnosis was already obtained prior to surgery by blood culture (n ⫽ 57) or by serology (n ⫽ 5) for 62 of the 68 cases positive by PCR that were reliable. PCR results were congruent with the microbiological diagnosis (culture and/or serology) in 60 (96.8%) of 62 cases. The results of blood culture were discordant with results obtained by PCR amplification and sequencing in 2 cases. In one case, the molecular approach contributed to the etiological diagnosis of endocarditis (S. bovis), the discordance resulting from blood culture contamination. In the second case, Enterococcus faecalis was identified in blood cultures whereas both 16S rRNA and RpoB PCR were positive for S. bovis. This might be explained by dual infection of the valve with two faecal bacterial species. PCR and sequencing of the valves and interpretation of its result using the algorithm shown in Figure 2 exhibited a specificity of 100% (118/118) and a sensitivity of 61% (64/105) (Table 3). Importantly, PCR was positive in 4

patients with possible endocarditis that was not confirmed by histological examination (Table 6). Three cases involved prosthetic valves, which might explain the total or nearly total absence of inflammation seen on pathological examination of the valve samples. PCR targeting 16S rRNA gene identified an organism in 5 (38%) of 13 cases with definite and in 6 (33%) of 18 cases with possible blood culture-negative endocarditis. The etiological agents in these cases of culture-negative endocarditis were Coxiella burnetii (n ⫽ 3), Bartonella quintana (n ⫽ 2), Streptococcus sp. (n ⫽ 2), Mycoplasma hominis (n ⫽ 1), Granulicatella sp. (n ⫽ 1) and Aspergillus sp. (n ⫽ 1). Among the 13 patients with definite culture-negative endocarditis, 12 (92%) were treated with antibiotics before blood cultures were obtained, which may potentially explain the negative blood cultures. PCR contributed to the diagnosis of 6 (23%) of 26 cases of endocarditis of unknown etiology prior to valve surgery (including 2 patients with congruent valve culture); PCR identified an organism in 4 (17.4%) of 23 cases of endocarditis of unknown etiology after valve histology and valve culture.

Discussion In the present study, we report the histology culture and PCR examinations of 127 patients with endocarditis and

234

The American Journal of Medicine, Vol 118, No 3, March 2005

Figure 2

Algorithm for the interpretation of the PCR results performed on valve samples.

118 valves excised from patients without endocarditis. This study, which was performed during an 8½ year period, provides data on the sensitivity and specificity of these diagnostic approaches (Table 3). This work underscores the excellent specificity of valvular histology. This study also provides an interpretation scheme for positive PCR amplification plus sequencing results that may help in recognizing false positive results due to PCR contamination. Histologic examination of excised cardiac valves is considered the gold standard for the diagnosis of endocarditis.8,11,13,14 However, no data on the specificity and positive predictive value supports this postulate, since valves taken from patients without endocarditis have not been systematically examined for histological criteria of infectious endocarditis. Here, we confirm that valvular histology should indeed be considered the gold standard of the diagnosis of endocarditis, with specificity and positive predictive values of 100%. The strict histological criteria that we used to classify cases of endocarditis as proven (presence of

“typical lesions”) were associated with an overall sensitivity of 63%. This contrasts with the sensitivity of 79% reported by Morris et al.,8 who used less stringent histologic criteria such as the presence of any valvular inflammation. Using such criteria, the positive predictive value of valvular histology decreases from 100 to 93.5%. Consequently, even though inflammation is generally present in valves taken from patients with endocarditis, the presence of inflammation without concomitant vegetations or bacteria is insufficient to histologically confirm a diagnosis of endocarditis. Our study underscores the low sensitivity and negative predictive value of histologic valve examination. The 25 cases with definite endocarditis but normal histology may be due to “sampling errors” (ie, cases in which the examined valve portion was not affected by the infectious process). Excised heart valves are nowadays routinely cut into several pieces for histology, culture, PCR, and freezing. The division of the valves may thus lead to sampling errors that affect the sensitivity of all diagnostic approaches. This may

Greub et al

Diagnosis of Infectious Endocarditis

235

Table 2 Characteristics of the 245 patients included in the study according to the presence or absence of infectious endocarditis, as determined by Duke criteria

Demographic data Male sex Median age years (IQR†) Type of valve - native - bioprothesis - mechanic prothesis Which valve‡ - aortic - mitral - tricuspid

Definite infectious endocarditis* n⫽105 (%)

Possible infectious endocarditis n⫽22 (%)

No infectious endocarditis n⫽118 (%)

70 (67) 61 (51–70)

16 (73) 61 (53–70)

74 (63) 68 (56–73)

70 (67) 24 (23) 14 (13)

13 (59) 7 (32) 5 (23)

102 (86) 14 (12) 6 (5)

67 (64) 62 (59) 10 (10)

14 (64) 15 (68) 1 (5)

85 (72) 61 (52) 19 (16)

*Including 7 patients with possible infectious endocarditis prior to surgery, which were reclassified as definite, based on histological examination; †IQR⫽interquartile range. ‡The total is above the number of valve replacement as aortic and mitral valves were involved 31 times; aortic & tricuspid and mitral 28 times; and mitral and tricuspid 2 times.

especially be true when examining prosthetic valves.23 The fact that prosthetic valve endocarditis is generally associated with less prominent inflammation on histologic examination may also explain the lower sensitivity of histology for the diagnosis of endocarditis involving prosthetic valves. Valve culture is not recommended for patients without clinical suspicion of endocarditis, as false positive results frequently occur.24 Valve culture was positive in 11.9 to 17.7% of patients without endocarditis who underwent valve surgery, but results were considered as true positive in only two of 638 cases.25–27 Though in our work, contamination occurred in only 2 of 118 cultures of valves taken from patients without endocarditis, valve culture should not be performed systematically because its sensitivity (13%) as well as positive and negative predictive values are low (Table 3), consistent with previous reports.23,24 Valve cul-

ture is also of limited utility in making the etiological diagnosis of endocarditis. Thus, we isolated a potential causal microorganism from only 3 valves taken from patients with infectious endocarditis of unknown etiology, and in two of these 3 cases, the etiological diagnosis was also provided by PCR and sequencing. Moreover, the identification of coagulase-negative Staphylococci in culture from the third valve may have represented contamination, as coagulase-negative Staphylococci commonly contaminating valve culture.25 Contamination is considered the most important problem in the molecular diagnosis of infectious endocarditis,28 having been previously reported in 13% (3/23)29 to 20% (3/15)28 of instances. Our work confirmed the occurrence of false positive PCR results in 18.7% (46/245) of cases. We observed both PCR contaminations and reagent contamination (see footnote of Table 4). Reagent contamination also was reported by

Table 3 Sensitivity, specificity, positive and negative predictive values of histology, PCR and culture performed on valvular samples taken from patients with and without infectious endocarditis Diagnostic approach

Sensitivity

Specificity

Positive predictive value

Negative predictive value

Histology*,† PCR after interpretation‡,§ Valve culture§

63% (62/98) 61% (64/105)

100% (118/118) 100% (118/118)

100% (62/62) 100% (64/64)

115/143 (82.5%) 74% (118/159)

13% (14/105)

98% (116/118)

87% (14/16)

56% (116/207)

*Using stringent criteria (ie, the presence of vegetation, microorganisms, and/or valvular inflammation with a predominance of polymorphonuclear cells). †Assessed on valves taken from 98 patients with definite infectious endocarditis (modified Duke criteria14 calculated prior to valve surgery) and from 118 patients without infectious endocarditis. ‡Amplification, sequencing and interpretation of the results according to Figure 1. §Assessed on 223 valvular samples taken from 118 patients without infectious endocarditis and from 105 patients with definite infectious endocarditis. (Seven patients with possible infectious endocarditis were reclassified as definite infectious endocarditis based on histological examination of valvular samples).

236

The American Journal of Medicine, Vol 118, No 3, March 2005

Table 4

Interpretation of the PCR results

Negative results* Positive results confirmed by - valve culture only† - blood culture only - PCR targeting a second gene - serology & PCR targeting a second gene‡ - valve & blood culture - PCR targeting a second gene & valve culture† - PCR targeting a second gene & blood culture - valve & blood culture & second PCR

definite infectious endocarditis n⫽105

possible infectious endocarditis n⫽22

no infectious endocarditis n⫽118

total n⫽245

41 64 1 27 6 4 2 1 19 4

18 4 0 0 0 2 0 0 2 0

118 0 0 0 0 0 0 0 0 0

177 68 1 27 6 6 2 1 21 4

*Including non-reliable PCR results due to PCR contamination (i.e. Acinetobacter sp.) and reagents contamination (Caulobacter sp., Pseudomonas sp., Stenotrophomonas sp. or Sphingomonas sp.). †E. coli (n⫽1), Coagulase negative Staphylococci (n⫽1). ‡Coxiella burnetii (n⫽3), Bartonella spp. (n⫽2), Mycoplasma hominis (n⫽1).

Bosshard et al.23 The occurrence of PCR contamination highlights the need for appropriate controls and careful interpretation. The interpretation algorithm that we present in Figure 2 is associated with an excellent specificity (100%). For endocarditis of unknown etiology, PCR amplification and sequencing provided an etiology in 23% of cases. Overall, the sensitivity of PCR and sequencing was similar to that of histology and significantly higher than that of valve culture. The most likely explanations for the relatively low sensitivity of the molecular approach, which was not overcome by repeating PCR amplification

Table 5

when initially found to be negative in patients with clinically documented infectious endocarditis, include sampling errors; antibiotic treatment given prior valve surgery; and the presence of PCR inhibitors in bloody valve samples, such as haemoglobin and lactoferrin.30 The sensitivity of PCR might be further increased by using bovine serum albumin to decrease PCR inhibition,30 or by using systems such as Light-Cycler, that have been associated with increased sensitivity.31 In conclusion, this study shows an excellent specificity and a low negative predictive value of valvular histology for

Bacterial identification obtained by sequencing DNA amplified from 68 valve samples PCR positive with diagnosis initially established by

Bacterial identification

serology or blood culture

valve analysis

Staphylococcus aureus Coagulase-negative Staphylococci Streptococcus bovis Streptococcus sp. (other than S. bovis) Enteroccocus faecium Enterococcus faecalis Bartonella quintana Coxiella burnetii Mycoplasma hominis Granulicatella sp. Abiotrophia sp. Actinobacillus sp. Pasteurella sp. Escherichia coli Enterobacter sp. Candida albicans Aspergillus sp. Total

11 4 12 15 1 6 2† 3† 0 0 1 1 1 1 1 1 0 60

1 0 2* 2 0 0 0 0 1‡ 1 0 0 0 0 0 0 1 8

*Both patients presented a discordant positive blood culture. †Blood culture negative, serology positive. ‡Later confirmed by serology.

Greub et al

Diagnosis of Infectious Endocarditis

237

Table 6 Characteristics of the 4 cases of possible infectious endocarditis, for which PCR was positive, despite the absence or low level of inflammation observed at histological examination of the same valve sample. Note that all patients had a prior history of infectious endocarditis or valve surgery Sex

Age years

Clinical presentation

Histology

Duke prior to surgery

F

70

moderate inflammation

1 major, 2 minor

Streptococcus sanguis

M

46

no inflammation

1 major, 2 minor

Streptococcus anginosus

F

85

no inflammation

1 major, 1 minor

M

33

Prosthetic valve since 1994; infectious endocarditis in 1997 due to S. sanguis with relapse in 1998; blood cultures congruent with PCR Infectious endocarditis treated since March 2002 with amoxicillin-gentamicin, surgery in June 2002, relapse with vegetation seen at transoesophageal echography in July 2002. Prosthetic valve, serology positive for Coxiella burnetii congruent with PCR, no fever. Vegetation and dehiscence of mitral prosthethic valve seen at transoesophageal echography in July 2001, surgery performed after 15 days of amoxicillingentamicin

lymphocytic inflammation

1 major, 1 minor

Coxiella burnetii Mycoplasma hominis

the diagnosis of infectious endocarditis. It also shows that 16S rRNA PCR amplification and sequencing prove efficient in providing an etiological diagnosis particularly in cases of blood culture and serology negative endocarditis. This study also shows that a second gene amplification and sequencing reaction is useful to confirm 16S rRNA identification. Valve analysis should be restricted, in the interest of cost effectiveness, to cases of blood culture-negative endocarditis and to patients undergoing valve replacement for hemodynamic reasons who present with fever and an increased erythrocyte sedimentation rate or other clinical presentations that are not specific enough to allow the diagnosis of infectious endocarditis.

Acknowledgments The authors are indebted to Veronique Brice (Marseille, France) for technical assistance and to Babu Gundi (Marseille, France) and Philip Tarr (Lausanne, Switzerland) for reviewing the manuscript. We thank the Swiss National Science Foundation for funding the postdoctoral fellowship of Gilbert Greub, who was the recipient of an “advanced researcher fellowship.”

References 1. Belli J, Waisbren BA. The number of blood culture necessary to diagnose most cases of bacterial endocarditis. Am J Med Sci. 1956; 232:284 –288. 2. 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– 506. 3. Berbari EF, Cockerill FR, Steckelberg JM. Infective endocarditis due to unusual or fastidious microorganisms. Mayo Clin Proc. 1997;72: 532–542.

PCR

4. Barnes PD, Crook DWM. Culture negative endocarditis. J Infect. 1997;35:209 –213. 5. Moreillon P, Que YA. Infective endocarditis. Lancet. 2004;363:139 – 149. 6. Brouqui P, Raoult D. Endocarditis due to rare and fastidious bacteria. Clin Microbiol Rev. 2001;14:177–207. 7. Houpikian P, Raoult D. Diagnostic methods current best practices and guidelines for identification of difficult-to-culture pathogens in infective endocarditis. Infect Dis Clin North Am. 2002;16:377–92. 8. Morris AJ, Drinkovic D, Pottumarthy S, et al. Gram stain, culture, and histopathological examination findings for heart valves removed because of infective endocarditis. Clin Infect Dis. 2003;36:697–704. 9. Lamas CC, Eykyn SJ. Suggested modifications to the Duke criteria for the clinical diagnosis of native valve and prosthetic valve endocarditis: analysis of 118 pathologically proven cases. Clin Infect Dis. 1997;25: 713–719. 10. Fournier PE, Raoult D. Nonculture laboratory methods for the diagnosis of infectious endocarditis. Curr Infect Dis Reports. 1999;1:136 – 141. 11. Lepidi H, Durack DT, Raoult D. Diagnostic methods current best practices and guidelines for histologic evaluation in infective endocarditis. Infect Dis Clin North Am. 2002;16:339 –361. 12. von Reyn CF, Levy BS, Arbeit RD, et al. Infective endocarditis: an analysis based on strict case definitions. Ann Intern Med. 1981;94: 505–517. 13. Durack DT, Lukes AS, Bright DK. New criteria for diagnosis of infective endocarditis: utilization of specific echocardiographic findings. Amer J Med. 1994;96:200 –222. 14. 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– 638. 15. Goldenberger D, Kunzli A, Vogt P, et al. Molecular diagnosis of bacterial endocarditis by broad-range PCR amplification and direct sequencing. J Clin Microbiol. 1997;35:2733–2739. 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– 680. 17. Millar BC, Xu J, Moore JE. Risk assessment models and contamination management: implications for broad-range ribosomal DNA PCR as a diagnostic tool in medical bacteriology. J Clin Microbiol. 2002; 40:1575–1580. 18. La Scola B, Raoult D. Culture of Bartonella quintana and Bartonella henselae from human samples: a 5-year experience (1993 to 1998). J Clin Microbiol. 1999;37:1899 –1905.

238 19. Raoult D, Vestris G, Enea M. Isolation of 16 strains of Coxiella burnetii from patients by using a sensitive centrifugation cell culture system and establishment of strains in HEL cells. J Clin Microbiol. 1990;28:2482–2484. 20. Drancourt M, Bollet C, Carlioz A, et al. 16S ribosomal DNA sequence analysis of a large collection of environmental and clinical unidentifiable bacterial isolates. J Clin Microbiol. 2000;38:3623–3630. 21. Greub G, Raoult D. “Actinobaculum massiliae,” a new species causing chronic urinary tract infection. J Clin Microbiol. 2002;40:3938 –3941. 22. Benson D, Lipman DJ, Ostell J. GenBank. Nucl Acids Res. 19935;21: 2963-2965. 23. Bosshard PP, Kronenberg A, Zbinden R, et al. Etiologic diagnosis of infective endocarditis by broad-range polymerase chain reaction: a 3-year experience. Clin Infect Dis. 2003;37:167–172. 24. Watkin RW, Lang S, Lambert PA, et al. The microbial diagnosis of infective endocarditis. J Infect. 2003;47:1–11. 25. Giladi M, Szold O, Elami A, et al. Microbiological cultures of heart valves and valve tags are not valuable for patients without infective endocarditis who are undergoing valve replacement. Clin Infect Dis. 1997;24:884 – 888. 26. Chuard C, Antley CM, Reller LB. Clinical utility of cardiac valve gram stain and culture in patients undergoing native valve replacement. Arch Pathol Lab Med. 1998;122:412– 415. 27. Campbell WN, Tsai W, Mispireta LA. Evaluation of the practice of routine culturing of native valves during valve replacement surgery. Ann Thorac Surg. 2000;69:548 –550.

The American Journal of Medicine, Vol 118, No 3, March 2005 28. Grijalva M, Horvath R, Dendis M, et al. Molecular diagnosis of culture negative infective endocarditis: clinical validation in a group of surgically treated patients. Heart. 2003;89:263–268. 29. Gauduchon V, Chalabreysse L, Etienne J, et al. Molecular diagnosis of infective endocarditis by PCR amplification and direct sequencing of DNA from valve tissue. J Clin Microbiol. 2003;41:763–766. 30. Abu Al-Soud W, Radstrom P. Effects of amplification facilitators on diagnostic PCR in the presence of blood, feces, and meat. J Clin Microbiol. 2000;38:4463– 4470. 31. Zeaiter Z, Fournier PE, Greub G, et al. Diagnosis of Bartonella endocarditis by a real-time nested PCR assay using serum. J Clin Microbiol. 2003;41:919 –925. 32. Weisburg WG, Barns SM, Pelletier DA, et al. 16S ribosomal DNA amplification for phylogenetic study. J Bacteriol. 1991;173:697–703. 33. Poyart C, Quesnes G, Trieu-Cuot P. Sequencing the gene encoding manganese-dependent superoxide dismutase for rapid species identification of enterococci. J Clin Microbiol. 2000;38:415– 418. 34. Mollet C, Drancourt M, Raoult D. rpoB sequence analysis as a novel basis for bacterial identification. Mol Microbiol. 1997;26:1005–1011. 35. Reid SD, Selander RK, Whittam TS. Sequence diversity of flagellin (fliC) alleles in pathogenic Escherichia coli. J Bacteriol. 1999;181: 153–160. 36. Raad I, Hanna H, Sumoza D, et al. Polymerase chain reaction on blood for the diagnosis of invasive pulmonary aspergillosis in cancer patients. Cancer. 2002;94:1032–1036.