Infective endocarditis (IE) is a disease of increasing medical importance in most developed countries with a growing proportion of elderly persons in the population39. Cases of IE associated with negative blood culture constitute 5% of all endocarditis cases, with a reported range of 2.5%-31%97,98. Five main factors have been reported in the literature as causes for blood culture-negative IE38,69: 1) fastidious slow-growing bacteria, especially Coxiella burnetii and Bartonella species; 2) nonbacterial organisms, namely fungi; 3) antibiotic administration preceding blood cultures; 4) right-sided endocarditis; 5) endocarditis in patients with a permanent pacemaker. These situations present a similar diagnostic dilemma and uncertainty for the clinician: a belief that the patient has IE, but the key confirmatory test, blood culture, is negative4. While the 1994 Duke criteria, with their later modifications, have provided a reference for establishing the diagnosis of the disease, determination of the score is only possible after all clinical, epidemiologic, laboratory, and echocardiographic features have been characterized21,27,36,37,54,100. Therefore, diagnosis of blood culture-negative IE is frequently delayed, increasing the morbidity and the mortality rate. Further, control of the disease via appropriate antimicrobial chemotherapy is difficult, because the causative agent has not been identified13,35.
Our laboratory, the French National Reference Center for Rickettsial Diseases, specializes in intracellular and other difficult-to-grow pathogens. In the current study, we report our experience with blood culture-negative endocarditis over 18 years. We have previously reported series of patients infected with specific pathogens10,28. We conducted the current study to determine the etiology of confirmed cases of blood culture-negative IE in order to compare the clinical, epidemiologic, laboratory, and echocardiographic features associated with each etiology, as well as the characteristics shared by unexplained cases.
PATIENTS AND METHODS
Our laboratory, located in Marseille, France, received samples from 3 major sources for patients suspected of having IE: 1) serology tests, blood culture, and, in case of surgical treatment, valvular analysis were systematically performed in the laboratory for patients hospitalized in public hospitals in the area of Marseille, when IE was suspected by the physician; 2) we received sera, blood, and valves of patients with IE from most hospitals in France when routine analysis remained negative; 3) samples from other countries in and outside of Europe were sent to the laboratory, often for diagnostic confirmation, when blood culture-negative IE due to fastidious bacteria was suspected. Therefore, the current study was more a compilation of cases from various origins than a systematic prospective study of blood culture-negative IE.
For each patient suspected of having IE, a physician of the laboratory completed a form to collect administrative, epidemiologic, clinical, laboratory, and echocardiographic data. Data included sex; age; permanent address, which determined a rural or urban life or homeless status; contact with animals; raw cheese consumption; and intravenous drug use. The presence of immunosuppression was also recorded, as well as a history of previous valvular disease, including IE. Factors considered responsible for immunosuppression were human immunodeficiency virus (HIV) infection, chronic alcoholism, cirrhosis, chronic hepatitis, renal insufficiency, diabetes mellitus, neoplasm, steroid treatment, and transplantation. The following clinical symptoms were recorded: the presence of fever, weight loss, myalgia, arthralgia; cardiac murmur and signs of heart failure, pericarditis, myocarditis, and pulmonary involvement; rash, petechiae, digital clubbing; Osler nodes, Janeway lesion, Roth spot; hepatomegaly, splenomegaly, and embolic lesions. The results of electrocardiography and echocardiography were recorded, as well as the presence of leukocytosis, leukopenia, anemia, thrombocytopenia, elevated liver enzymes, immunologic features, and elevated sedimentation rate over 50 mm the first hour. Finally, the antibiotic treatment and the surgical procedures were recorded, as well as the evolution. These data were collected from various sources, thus answers were not obtained for all items.
Patients were assessed by the modified Duke criteria. Only patients who presented between 1 June 1983 and 31 May 2001 with definite IE according to the Duke criteria, with at least 3 blood cultures that remained negative after a 7-day incubation, were included in the study.
Indirect immunofluorescent antibody tests were carried out as previously described6,59,79,94 to detect specific antibodies reactive with C. burnetii, Bartonella quintana, Bartonella henselae, and Legionella pneumophila. Titers of IgG, IgM, and IgA to both phase I and phase II C. burnetii antigens were determined. A serum sample with a C. burnetii phase I IgG titer of 1:800 or more (PPV = 98%) was considered a major criterion for IE, according to the modified Duke criteria27. A serum sample with an IgG antibody titer to Bartonella species of 1:800 or more has been demonstrated previously to be 95.5% predictive of Bartonella endocarditis. Specific antibodies to Brucella species were detected with an immunoenzymatic antibody test (titer ≥1:200); anti-Aspergillus species antibodies were detected using an indirect hemaglutination assay1,56,96. Serology tests for C. burnetii were performed during the whole study period. Serology tests for Bartonella, Legionella (indirect immunofluorescent antibody, titer ≥1:256), Aspergillus, and Brucella were performed systematically starting in 1994.
When available in sufficient amount, serum samples with Bartonella IgG antibody titers of 1:800 or more were analyzed through cross-adsorption and Western immunoblotting as previously described, to overcome cross-reactivity between Bartonella species42,60. Fifty-one samples were analyzed with this technique, of which 22 had not been tested by PCR and/or culture. Western blot analysis was performed with antigens from the 4 Bartonella species that have been involved in IE in humans: B. quintana, B. henselae, B. vinsonii subsp. berkhoffii, and B. elizabethae. Each serum was tested before and after cross-adsorption with B. quintana and B. henselae. Cross-adsorption results in the disappearance of homologous and heterologous antibodies when it is performed with the bacterium causing the disease, while only heterologous antibodies disappear when it is performed with the bacterium that did not cause the disease but is responsible for the cross-reaction61.
To isolate C. burnetii, specimens of heparinized blood and homogenized cardiac valves were inoculated onto human embryonic lung cells grown in shell vials, as reported elsewhere, and the bacteria were detected by immunofluorescence testing59,81. To isolate Bartonella species, blood and valvular tissues were inoculated both on blood agar at 37°C in a 5% carbon dioxide incubator and in human endothelial cells (ECV 304) cultured in shell vial, following the same procedure used for C. burnetii50. Detection of the bacteria was performed by Gimenez staining and specific immunofluorescence17. When serologic results suggested an etiology other than C. burnetii or Bartonella species or were not informative, specimens of heparinized blood and homogenized cardiac valves were inoculated in shell vial ECV 304 cultures following the same procedure used for C. burnetii46. Three weeks after inoculation, growth of the bacteria was detected by Gimenez and acridine orange staining, electron microscopy, and immunofluorescence using the patient's serum. Bacteria were then identified through amplification and sequencing of the 16S rRNA gene as previously described9,31.
Paraffin-embedded heart valves were examined with hematoxylin-eosin for histopathologic features51. Nonspecific stains such as Giemsa, Gram (Brown-Brenn and Brown-Hopps), periodic-acid Schiff, Grocott-Gomori, Warthin-Starry, Gimenez, and Ziehl-Nielsen were systematically performed to detect microorganisms within tissues, as described elsewhere11,102. Finally, presence of C. burnetii, Bartonella species, Chlamydia species, or Tropheryma whipplei in valvular tissues was investigated by immunohistochemistry as previously reported, using an immunoenzymatic assay with polyclonal specific antibodies8 (Figure 1).
Bacterial DNA was detected directly in surgically excised valves by PCR amplification with species-specific and universal primers and sequencing of the amplified fragment, as previously described. C. burnetii DNA was detected by use of a specific pair of primers that amplified an htpAB-associated repetitive element29,101. As for Bartonella species, genus-specific primers that amplified the 16S-23S intergenic spacer regions were used84,85. When no evidence of any causative agent was available, amplification was performed using universal primers that amplified the 16S rRNA gene31,100.
Data were transferred to Epi-Info 6 for statistical analysis. The Pearson chi-square test or the Fisher exact test was used to compare qualitative data. Mean values were compared using the Student t test. A difference was considered significant when p < 0.05. To characterize the presentation of each specific etiology of blood culture-negative IE, clinical, epidemiologic, laboratory, and echocardiographic data were compared in each group of patients with a shared etiology, including the group that was not related to any explicative factor. Among the cases of Bartonella endocarditis, features of patients with B. quintana endocarditis were compared with those of patients with B. henselae endocarditis, to identify the specific features of each disease.
Between 1983 and 2001, 348 cases of definite IE with negative blood cultures were reported to our laboratory and included in the current study. The male:female sex ratio of the patients was 2.94, and the mean age was 54 ± 16 years (range, 13-88 yr.). Of the 348 cases, 325 patients (93%) were hospitalized in France and 23 (7%) in other countries, including 11 (3%) in the United Kingdom. Echocardiographic signs of endocarditis were found in 293 of 340 patients (86%). Serum was available for 346 (99%) of the 348 patients, heparinized blood for cell culture for 227 (65%), and valvular specimens for 151 (43%). Results of the specific laboratory tests (Table 1) allowed us to identify 167 of 348 cases (48%) as Q fever endocarditis, and 99 (28%) as Bartonella species endocarditis. Among these, 106 cases of C. burnetii and 98 of Bartonella endocarditis were detected by serology only. Five cases (1%) could be related to rare bacteria, including 2 (0.6%) to T. whipplei, 1 to Mycoplasma hominis, 1 to Abiotrophia elegans, and 1 to Legionella pneumophila. No case had more than 1 rare cause simultaneously. Four were Streptococci diagnosed by valve PCR amplification and sequencing of the 16S rRNA gene. In 1 patient, Escherichia coli was isolated from 3 blood cultures. Among the 73 without diagnosis, administration of antibiotics before blood cultures was reported in 58 (79%). Six patients had right-sided endocarditis and for 4 the disease was associated with a permanent pacemaker. Finally, there were only 5 cases that could not be associated with any currently recognized etiology or causative factor of blood culture-negative IE.
Q Fever Endocarditis
Of the 167 patients diagnosed as having C. burnetii endocarditis, 125 (75%) were male and their mean age was 56 ± 17 years (range, 13-88 yr.). A cause of immunodeficiency was found in 53 patients (32%). Ten (6%) had a history of previous IE. The preexisting status of heart valves was known for 153 patients and a valvular disease was found in 139 (91%), including 22 cases (14%) with congenital malformation, 27 (16%) with mechanical prosthetic valve, and 47 (28%) with bioprosthetic valve. Living in a rural area was noted for 71 of 139 patients (51%), contact with domestic animals for 90 of 128 (70%), and ingestion of cheese made from raw milk in 28 of 94 (30%). As for clinical symptoms during Q fever endocarditis, the most prevalent were fever (87%), weight loss (59%), arthro-myalgia (32%), hepatomegaly (33%), splenomegaly (42%), and heart failure (38%). Thirty-five patients (21%) suffered arterial emboli. As for laboratory symptoms, increased erythrocyte sedimentation rate (52%), immunologic phenomena such as rheumatoid factor (52%), thrombocytopenia (39%), and anemia (37%) were the most common features (Table 2). Of 166 patients who benefited from echocardiographic examination, 137 (83%) presented a valvular regurgitation and 67 (40%) a valvular vegetation. The involved valve was aortic in 65 patients (53%), mitral in 56 (46%), both in 35 (21%), and tricuspid in 1 case. Seven patients (7%) suffered from endocarditis extended to the 3 valves together and 13 (8%) had a pacemaker. The antibiotic treatment was recorded for 145 patients. All received doxycycline, 81 (56%) in combination with hydroxychloroquine, 25 (17%) with a quinolone, and 19 (13%) with the 2 regimens successively. Valvular replacement was performed in 98 of 166 (59%) patients. The outcome was obtained for 150 patients, of whom 4 (3%) died and 9 (6%) relapsed. The duration of antibiotherapy preceding the surgical excision was recorded for 29 of the 30 patients with valvular samples analyzed by culture, PCR, and immunohistochemistry. Although the valvular analysis results were not significantly different between the 10 patients who did not receive any previous treatment and those who were treated for less than 3 months, detection of C. burnetii by culture or PCR was significantly less frequent in the group of 9 patients who received antibiotic treatment for 1 year or more.
Of the 99 cases with Bartonella species endocarditis, 1 patient had negative serology. Diagnosis at the species level was possible through isolation of the bacterium from the blood in 3 patients and through culture and/or PCR analysis of the valve in 46. Of these 49 patients, 38 were infected with B. quintana, 10 with B. henselae, and 1 with B. vinsonii subsp. berkhoffii. In the 50 other patients, serology was the only evidence for Bartonella species involvement and did not allow species determination because of extensive cross-reactions. Of these 50, however, 22 were analyzed by cross-adsorption and Western immunoblotting. Fifteen exhibited a specific antigenic profile for B. quintana, 4 exhibited a B. quintana profile, and 3 profiles were not informative. Globally, B. quintana was shown to be responsible for 53 (75%) cases, B. henselae for 17 (24%), and B. vinsonii subsp. berkhoffii for 1.
Comparison of the 53 B. quintana cases and the 17 B. henselae cases showed no difference in the sex ratio and the average age of the patients. B. quintana infection, however, was significantly associated with immunodeficiency, chronic alcoholism, homelessness, and contact with body lice, while a preexisting valvular disease and contact with cats were statistically related to B. henselae infection, as reported28. As for clinical, biologic, and echocardiographic features (found in 88 of 95 patients), no significant difference was measured between the 2 groups. Right-sided endocarditis and endocarditis with pacemaker represented 17% and 3%, respectively, of Bartonella endocarditis. The outcome was also similar in the 2 groups, with a prevalence of surgical replacement of 73/97 (75%) and a mortality rate of 6/82 (7%) in the whole group of 99 Bartonella endocarditis cases. The antibiotic treatment was recorded for 79 of these 99 cases, 66 (84%) of which received a combination including an aminoglycoside76.
Endocarditis Due to Rare Pathogens
The diagnosis of T. whipplei endocarditis was established in 1 patient through detection of the bacterium in valvular specimens by PCR amplification, immunohistochemistry, and culture. In another case, molecular detection in valve tissue was the only evidence for the diagnosis. In the 2 cases, microscopic examination of the valves revealed intracellular periodic-acid Schiff-positive bacilli within a mononuclear infiltrate. The 2 patients, 40- and 42-year-old men, respectively, were immunocompetent. Clinically, they presented with fever and heart failure. Echocardiographic examination revealed that both mitral and aortic valves were involved and showed a valvular abscess in the 2 patients. These echocardiographic features were statistically associated with T. whipplei endocarditis, when compared to cases due to C. burnetii or Bartonella species (p < 0.05). The 2 patients benefited from adapted antibiotic treatment and surgical valves excision. One died.
A. elegans was detected by PCR in a 29-year-old man with a bicuspid aortic valve who presented with fever, valvular regurgitation, and vegetation. Similarly, PCR provided evidence for M. hominis in the aortic valve of an 80-year-old man with a previous valvular disease. Specific antibodies to M. hominis were also detected in this patient. A case of L. pneumophila endocarditis was diagnosed by IFI in a 72-year-old man with a mechanical prosthetic aortic valve and a pacemaker.
Endocarditis With Antibiotic Administration Preceding Blood Cultures
Comparing the 58 cases of blood culture-negative IE that were treated with antibiotics before blood cultures with the 3 largest other groups of patients, that is, those infected with C. burnetii, B. quintana, and B. henselae, did not reveal any specific profile for this etiologic group. The sex ratio was in favor of males but more balanced than in the C. burnetii and B. quintana groups. When compared with these 2 groups, patients who received antibiotics before blood cultures less frequently suffered weight loss and more frequently had leukocytosis. Prevalence of splenomegaly was significantly lower than in patients infected with C. burnetii or B. henselae. No difference was noted with regard to echocardiographic results or outcome (Table 3). Of the 58 patients, serologic analysis was performed in 56, blood culture in shell vial in 25, culture of valve sample in 27, and PCR amplification of the 16S rDNA from valve tissue in 27.
Right-Sided Endocarditis and Endocarditis With Pacemaker
Comparing the 6 cases of blood culture-negative right-sided IE with the groups of patients infected with C. burnetii, B. quintana, and B. henselae, respectively, showed that the average age was significantly lower in patients with right-sided IE (36 yr vs 56, 48, and 50, respectively; p < 0.05). As for epidemiologic factors, intravenous drug use was significantly more frequent than in the 3 other groups. Clinically, right-sided IE was specifically associated with pulmonary embolic phenomena. Echocardiographic features and outcome of the 6 patients were similar to those of the other groups (Table 4). Comparing the cases associated with a permanent pacemaker with patients infected with C. burnetii, B. quintana, and B. henselae, respectively, did not yield any significant result.
Five cases could not be associated with any reported etiologic factor, despite serologic testing in 5 of 5 cases, blood culture in shell vial for 4 of 5, and analysis of valve sample through culture and PCR for 2 of 5. The patients were all men, aged 35-82 years with an average of 56 ± 21 years. These characteristics were not significantly different from those of the other groups. None of the 5 patients was immunodeficient. Status of the valve before IE was recorded in 4 patients who all presented with preexisting valvular disease that was not congenital. Fever was reported in 4 of 5 cases, isolated in 2 cases or associated with splenomegaly (1 case) or heart failure (1 case). One patient presented with myalgia and weight loss but no fever. Laboratory results were not statistically different from those of other patients in the study. According to echocardiographic examination, the disease was located in all 5 patients in the aortic valve only, a feature found in 53% of C. burnetii cases (NS), 55% of B. quintana cases (NS), and 29% of B. henselae cases (p < 0.05). A myocardic abscess was found in 3 of 5 cases (60%), compared with only 20% of B. henselae endocarditis cases (NS), 12% of Q fever endocarditis (p < 0.05), and 11% of B. quintana endocarditis (p < 0.05). Antibiotic treatment was recorded in 3 patients and consisted of a combination of betalactam and aminoglycosides. Three of 5 patients benefited from aortic surgical replacement, a rate similar to that of other patients. No specificity was observed in the outcome of the 5 patients (none relapsed, none died).
The 348 cases of blood culture-negative IE in the current study constitute to our mind the most important series to date, as the main published studies did not include more than 100 patients49,69. For these 348 cases, the first striking feature was the high number of infections due to intracellular bacteria (n = 270; 78%). Two earlier publications38,49 assessing the involvement of such microorganisms reported rates of approximately 25%. Further, only 5 cases (1%) remained finally unexplained, confirming the observation of Lamas and Eykyn49 that truly negative IE is far less frequent than initially expected. These results clearly underline the crucial advantage of new diagnostic tools such as specific serologic and histologic assays for intracellular bacteria, new culture techniques, and PCR-based molecular tools to determine the etiology of blood culture-negative IE38,70,82. Indeed, one of the interests of the present study was the systematic, large-scale application of a complete set of tests: of the 348 patients, 346 benefited from serologic testing including C. burnetii and Bartonella species since 1994, 65% from blood culture in shell vial, and 43% from valve analysis through culture and PCR. As an illustration of the efficiency of our diagnostic strategy, serology of C. burnetii and Bartonella species, a simple and rapid noninvasive technique, would have been sufficient to establish the diagnosis of 78% of the cases.
The prevalence of intracellular bacteria in the current series, however, also reflects recruitment bias, as remarked by other authors who have conducted similar studies37,38. Patients from Marseille were systematically included as far as blood culture-negative IE was demonstrated, while patients from other origins were preselected by physicians who might have prioritized those they suspected to be infected by intracellular bacteria, namely C. burnetii or Bartonella species. Another possible bias could be related to variations in the performance of Duke criteria5,48. As demonstrated by Habib et al33, these criteria perform poorly in patients with blood culture-negative IE, since blood culture positivity is a major criterion72. Further, transesophageal echocardiography has been demonstrated as superior to transthoracic echocardiography in patients with blood culture-negative IE68,103. Of 227 documented patients, however, transesophageal echocardiography was performed in only 59%, including all patients from Marseille.
Q Fever Endocarditis
Endocarditis is the major clinical presentation of chronic Q fever and accounts for 80% of all chronic Q fever and for 3%-5% of all endocarditis79,80. Its yearly incidence is estimated at 1/1 million inhabitants in France88. Two host factors, a previous cardiac valve defect and an immunocompromised status, have been shown to be determinant in the evolution from acute Q fever to chronic endocarditis10,80. Immunodeficiency is reported in the literature for 9%-20% of patients (32% in the current series) and more frequently in patients who do not have previous valvular disease71. Thus, a study in 1993 by Brouqui et al10 showed that 5 of 7 patients without valvular disease were immunocompromised. In the current larger study, this situation was found in 10 of 14 (71%) patients. According to many authors, previous valve lesions are present in 85%-95% of cases and mainly concern the mitral or the aortic valve, or both, as observed in our series80. Q fever prosthetic valve endocarditis has been increasingly reported recently, and the presence of a prosthetic valve has been suggested to be a predisposing factor for evolution to chronicity in a mean delay of 6 months23,26. Overall, 38% of acute Q fever cases that occur in patients with previous valvular defects might evolve to chronic disease. Preventive administration of antibiotic treatment for 12 months is therefore recommended in such cases, together with regular controls of C. burnetii serology on a 3-month basis. Association of doxycycline and hydroxychloroquine has been shown to be significantly more efficient than doxycycline alone in this indication23.
As most clinical manifestations of Q fever endocarditis are protean and nonspecific, the diagnosis is often delayed for several months74. Further, the small size of valvular vegetation in Q fever endocarditis increases the difficulty of detecting this major criterion. While vegetations were reported in 40% of the 168 patients with C. burnetii endocarditis, they were found in 84% of the 99 patients infected with Bartonella species (p < 0.000001); these data were consistent with those reported in the literature10,59. Thus, in our series, the mean diagnostic delay was 14 months for patients infected with C. burnetii, 10 months for B. henselae infections, and less than 7 months for B. quintana infections. In recent years, however, an increase in interest in Q fever has been observed in several countries, leading to heightened awareness of the disease among physicians44. Also, routine examination with transesophageal echocardiography has allowed earlier detection of valvular lesions. Some authors have hypothesized that these evolutions could result in earlier diagnosis of the disease and therefore in changes in its clinical presentation as well as in its outcome7,41. It should be emphasized that an unexplained illness in a patient with an existing heart valve lesion is a strong indicator of Q fever, and that serologic testing for C. burnetii and transesophageal echocardiography should be performed routinely in such patients so that the diagnosis can be confirmed by use of the modified Duke criteria41,59.
With a mortality rate of 3% in patients with Q fever endocarditis, our series is consistent with other studies, which have shown that the mortality of the disease has been reduced from 60% until the 1970s to a mean of 5% today10,22,80. Together with the earlier diagnosis, the use of antibiotic regimens based on a combination of doxycycline with either quinolone or hydroxychloroquine accounts for this improvement77. In our study, 86% of patients benefited from at least 1 of these combinations. Relapses of the disease, however, have been commonly observed even after a prolonged treatment, and concern 6% of our patients. Such a high rate of treatment failure makes it necessary to ensure follow-up using sensitive tools for detection of C. burnetii53. Serology is a useful tool for this purpose, as stable low titers of antiphase I antibodies were shown to provide a strong argument for interruption of the treatment8,65,77. But recent studies by Lepidi et al52 suggested that valve analysis by immunohistochemistry, PCR, and culture might constitute a more sensitive tool for follow-up, as detection of the bacterium was correlated with the duration of treatment and decreased significantly only after 1 year of antibiotic treatment. In the current series, while C. burnetii was detected in 80% of valves in patients who were treated for less than 3 months, PCR and culture were positive in only 33% and 22%, respectively, of patients treated for 1 year or more.
Involved in 3%-4% of IE in France, the genus Bartonella is the second cause of blood culture-negative IE following C. burnetii75,89. With 29% of cases due to these bacteria, the current study confirms the etiologic importance of the genus. To date, 4 species, B. elizabethae, B. quintana, B. henselae, and B. vinsonii subsp. berkhoffii, have been involved in IE in humans18,60,87. Therefore, it is necessary to develop diagnostic tools to identify the agents at the species level in order to improve our knowledge of species-specific characteristics of the disease. Culture of these fastidious organisms, however, is difficult and requires lengthy incubation periods, especially from patients already being treated with antimicrobials28,50. Thus, in the current study, the sensitivities of blood culture and valvular tissue culture were 20% and 31%, respectively, for isolation of Bartonella species. These values are consistent with those reported in other studies. Molecular amplification of gene fragments from valvular tissue has been demonstrated to have a higher sensitivity than culture, between 72% and 98% in the literature (92% in the current study)28,75. This technique, however, requires specialized equipment as well as an invasive intervention for collecting the sample2,67. In the current study, we have also used a serologic, noninvasive technique to overcome these limitations. Analysis of serum samples by cross-adsorption and Western immunoblotting allowed determination of the causative species in 49 of 51 (96%) cases, including 19 of 22 cases that were not documented by PCR and/or culture. The study confirms therefore the usefulness of a rapid, noninvasive technique that provides not only a sensitive means for species identification in Bartonella endocarditis, but also a tool for identifying specific immunodominant antigens recognized by the human immune system during the disease42,60. Such antigens could be good candidates for species-specific diagnostic tests, as illustrated by the recent development of monoclonal antibodies against a B. quintana-specific epitope located on a 34 kDa protein19,55,57.
Among the cases of Bartonella endocarditis, 75% were associated with B. quintana and 25% with B. henselae, a profile consistent with data from the literature75. Other Bartonella species seemed to be exceptionally involved, as only 1 case of IE due to B. elizabethae and 1 case due to B. vinsonii subsp. berkhoffii have been reported to date18,87. Comparison of B. quintana and B. henselae endocarditis confirmed that they represent 2 distinct diseases, as suggested by Fournier et al28. B. quintana IE is associated with chronic alcoholism and contact with body lice, B. henselae endocarditis with exposure to cats. As B. quintana has been detected in body lice in many countries, its role in transmitting the bacterium to humans is probable86. Association of body lice exposure and bacillary angiomatosis also supports such a hypothesis47. As for B. henselae, cats are considered the animal reservoir of the bacterium, which could be transmitted by cat fleas3,15,34,40,45. As found in our study, a previous valvulopathy is significantly more frequent in B. henselae endocarditis than in B. quintana endocarditis. This could be related to differences in the pathogenic properties of the bacteria, as suggested by Spach et al90, who demonstrated that B. quintana bacteremia could evolve to IE in the absence of any valvular disease. But this might also reflect difficulties for acceding to medical services among homeless patients. Clinical presentation and outcome did not reveal any species-specific characteristic.
With a global mortality rate of 7% among the 99 patients with Bartonella endocarditis, the current series confirms the slow improvement in the prognosis of the disease noted in more recent published studies when compared with the 20%-30% reported previously9,21,75,76. As demonstrated by Raoult et al, prolonged treatment with aminoglycosides was statistically related to survival of the patients62,64,76.
Endocarditis Due to Other Recognized Causes
The current study underlines the crucial role of valve analysis, in particular PCR, for the diagnosis of IE due to rare microorganisms or when blood cultures have been sterilized by prior administration of antibiotics25,73. Thus, of the 5 cases associated with rare, fastidious bacteria, 4 were diagnosed through amplification and sequencing of the 16S rRNA gene directly from valve samples20. Among the 58 patients previously treated with antibiotic drugs, the causative agent was determined in 4 (7%) cases by the same technique and corresponded to Gram-positive cocci commonly involved in IE30,69,100. In terms of treatment strategy, however, it is important to notice that previous antibiotic treatment was also found in 12/115 (10%) C. burnetii and 13/51 (25%) Bartonella endocarditis cases. Previous studies have shown that blood culture-negative IE due to administration of antibiotics was an ideal indication for universal PCR, and that the improvement of species identification by sequencing contributed to a significant improvement in patient management24,63,70. Limitations of this tool, however, should not be underestimated. Clinical samples may contain factors that inhibit the amplification reaction, contamination is frequent, and finally the quality of sequences in gene databases is sometimes insufficient to allow clear identification of the amplified fragment46,83,93,100. Therefore, results of molecular detection should be interpreted carefully and compared with other evidence30. Of the 7 cases diagnosed by 16S rDNA amplification, 5 were supported by other evidence, such as microscopic visualization of microorganisms (for T. whipplei and S. mutans), serology (for M. hominis), and culture (for T. whipplei).
The 2 cases of T. whipplei endocarditis reported in the current study illustrate the increasing importance of this emerging disease14,32,66. A recent review of 35 published cases showed that males were most frequently concerned (89%), with an average age of 49 years, fever in 33%, heart failure in 22%, and vegetation in 75%25. Multiple valvular involvement, including the aortic valve, was more frequent (37%) than in other etiologies, and the disease occurred commonly on previously healthy valves. These characteristics are consistent with those seen in our cases, even if the number was too small to allow statistical comparisons. To date, no consensus has been reached about treatment of T. whipplei endocarditis, but combination of penicillin and aminoglycoside IV followed by oral cotrimoxazole is currently recommended25. The prognosis is severe: the outcome was fatal in 20 of 35 cases (57%) reviewed by Fenollar et al25. Characteristics of the other cases due to rare microorganisms were also consistent with established data, namely the occurrence of A. elegans IE in an aortic valve with preexisting lesions, and of L. pneumophila endocarditis in a bioprosthetic valve in a nosocomial context91,95. The value of the case due to L. pneumophila, however, might be debated, as it was based on serology only92. The case due to M. hominis harbored features consistent with those of the 5 cases reported to date in the literature16,24. Although 28 cases of Chlamydia endocarditis have been reported, this microorganism was not involved in the current series. This result is consistent with other recent studies suggesting that the role of Chlamydia species might have been significantly overestimated, mainly because of serologic cross-reactions with Bartonella species49,61.
One of the most noteworthy results of the study might be the identification of 5 cases that could not be associated with any recognized explanatory factor despite a complete analysis including serology of intracellular bacteria, blood culture in shell vial, and, in 2 cases, culture and PCR analysis of the excised valve. These 5 patients had similar profiles: all were male, with an average age of 56 years, immunocompetent, and presented with previous aortic valvular disease. Because a key step in the search for new infectious diseases has historically consisted of identifying homogeneous, specific clinical profiles within complex syndromes, it would be instructive to investigate these 5 cases further. As a hypothesis, viral IE could be envisaged: to our knowledge, such an etiology has been suggested by experimental works but never demonstrated in humans12.
We thank Vincent Klassen for reviewing the manuscript. We thank the physicians who provided the samples and clinical infections, and Hubert Lepidi for the figures.
1. Alexander TS, Gray LD, Kraft JA, Leland D, Nikaido MT, Willis DH. Performance of meridian immunocard mycoplasma test in a multicenter clinical trial. J Clin Microbiol
2. Anderson BE, Neuman MA. Bartonella spp. as emerging human pathogens. Clin Microbiol Rev
3. Baorto E, Payne RM, Slater LN, Lopez F, Relman DA, Min KW, St Geme JW. Culture negative endocarditis caused by Bartonella henselae. J Pediatr
4. Barnes PD, Crook DWM. Culture negative endocarditis. J Infect
5. Bayer AS, Ward JI, Ginzton LE, Shapiro SM. Evaluation of new clinical criteria for the diagnosis of infective endocarditis. Am J Med
6. Black CM, Johnson J, Farskey CE, Brown TM, Bordal BP. Antigenic variation among strains of Chlamydia pneumoniae. J Clin Microbiol
7. Boyle B, Hone R. Q fever endocarditis revisited. Ir J Med Sc
8. Brouqui P, Dumler JS, Raoult D. Immunohistologic demonstration of Coxiella burnetii in the valves of patients with Q fever endocarditis. Am J Med
9. Brouqui P, Raoult D. Endocarditis due to rare and fastidious bacteria. Clin Microbiol Rev
10. Brouqui P, Tissot-Dupont H, Drancourt M, Berland Y, Etienne J, Leport C, Goldstein F, Massip P, Micoud M, Bertrand A. Chronic Q fever: 92 cases from France, including 27 without endocarditis. Arch Intern Med
11. Bruneval P, Choucair J, Paraf F, Casalta JP, Raoult D, Scherchen F, Mainardi JL. Detection of fastidious bacteria in cardiac valves in cases of blood culture negative endocarditis. J Clin Pathol
12. Burch GE, Tsui CY. Evolution of Coxsackie viral valvular and mural endocarditis in mice. Br J Exp Pathol
13. Cannady PB, Sanford JP. Negative blood cultures in infective endocarditis: a review. South Med J
14. Celard M, de Gevigney G, Mosnier S, Buttard P, Benito Y, Etienne J, Vandenesch F. Polymerase chain reaction analysis for diagnosis of Tropheryma whipplei infective endocarditis in two patients with no previous evidence of Whipple's disease. Clin Infect Dis
15. Childs JE, Rooney JA, Cooper JL, Olson JG, Regnery RL. Epidemiologic observations on infection with Rochalimaea species in cats living in Baltimore. J Am Vet Assoc
16. Cohen JI, Sloss LJ, Kundsin R, Golightly L. Prosthetic valve endocarditis caused by Mycoplasma hominis. Am J Med
17. Curry A. Electron microscopy as a tool for identifying new pathogens. J Infect
18. Daly JS, Worthington MG, Brenner DJ, Moss CW, Hollis DG, Weyant RS, Steigerwalt AG, Weaver RE, Daneshvar MI, O'Connor SP. Rochalimaea elizabethae sp. nov. isolated from a patient with endocarditis. J Clin Microbiol
19. Dehio C. Pathogenesis of Bartonella (Rochalimaea) infections. Bull Inst Pasteur
20. Dobbins WO. The diagnosis of Whipple's disease. N Engl J Med
21. Durack DT, Lukes AS, Bright DK. New criteria for diagnosis of infective endocarditis: utilization of specific echocardiographic findings. Duke Endocarditis Service. Am J Med
22. Duroux-Vouilloz C, Praz G, Francioli P, Peter O. Fievre Q avec endocardite: presentation clinique et suivi serologique de 21 patients. Schweiz Med Wochenschr
23. Fenollar F, Fournier PE, Carrieri P, Habib G, Mesana T, Raoult D. Risks factors and prevention of Q fever endocarditis. Clin Infect Dis
24. Fenollar F, Gauduchon V, Casalta JP, Lepidi H, Vandenesch F, Raoult D. Mycoplasma endocarditis: two case reports and a review. Clin Infect Dis
25. Fenollar F, Lepidi H, Raoult D. Whipple's endocarditis: review of the literature and comparisons with Q fever, Bartonella infection, and blood-culture positive endocarditis. Clin Infect Dis
26. Fernandez Guerrero ML, Muelas JM, Aguado JM, Renedo G, Fraile J, Soriano F, De Villalobos E. Q fever endocarditis on porcine bioprosthetic valves. Clinicopathologic features and microbiologic findings in three patients treated with doxycycline, cotrimoxazole, and valve replacement. Ann Intern Med
27. 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
28. Fournier PE, Lelievre H, Eykyn SJ, Mainardi JL, Marrie TJ, Bruneel F, Roure C, Nash J, Clave D, James E, Benoit-Lemercier C, Deforges L, Tissot-Dupont H, Raoult D. Epidemiologic and clinical characteristics of Bartonella quintana and Bartonella henselae endocarditis: a study of 48 patients. Medicine (Baltimore)
29. Fournier PE, Marrie TJ, Raoult D. Diagnosis of Q fever. J Clin Microbiol
30. Gauduchon V, Benito Y, Celard M, Mouren C, Delorme V, Philippe-Bert J, Etienne J, Vandenesch F. Molecular diagnosis of recurrent Streptococcus mutans endocarditis by PCR amplification and sequencing. Clin Microbiol Infect
31. 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
32. Gubler JG, Kuster M, Dutly F, Bannwart F, Krause M, Vogelin HP. Whipple endocarditis without overt gastrointestinal disease: report of four cases. Ann Intern Med
33. Habib G, Derumeaux G, Avierinos JF, Casalta JP, Jamal F, Volot F, Garcia M, Lefevre J, Biou F, Maximovitch-Rodaminoff A, Fournier PE, Ambrosi P, Velut JG, Cribier A, Harle JR, Weiller PJ, Raoult D, Luccioni R. Value and limitations of the Duke criteria for the diagnosis of infective endocarditis. J Am Coll Cardiol
34. Hadfield TL, Warren R, Kass M, Brun E, Levy C. Endocarditis caused by Rochalimaea henselae. Hum Pathol
35. Hampton JR, Harrison MJG. Sterile blood cultures in bacterial endocarditis. Q J Med
36. Hoen B, Beguinot I, Rabaud C, Jaussaud R, Selton-Suty C, May T, Canton P. The Duke criteria for diagnosing infective endocarditis are specific: analysis of 100 patients with acute fever or fever of unknown origin. Clin Infect Dis
37. Hoen B, Selton-Suty C, Danchin N, Weber M, Villemot JP, Mathieu P, Floquet J, Canton P. Evaluation of the Duke criteria versus the Beth Israel criteria for the diagnosis of infective endocarditis. Clin Infect Dis
38. Hoen B, Selton-Suty C, Lacassin F, Etienne J, Briancon S, Leport C, Canton P. Infective endocarditis in patients with negative blood culture: analysis of 88 cases from a one-year nationwide survey in France. Clin Infect Dis
39. Hogevik H, Olaison L, Andersson R, Lindberg J, Alestig K. Epidemiologic aspects of infective endocarditis in an urban population. A 5-year prospective study. Medicine (Baltimore)
40. Holmes AH, Greenough TC, Balady GJ, Regnery RL, Anderson BE, O'Keane JC, Fonger JD, McCrone EL. Bartonella henselae endocarditis in an immunocompetent adult. Clin Infect Dis
41. Houpikian P, Habib G, Mesana T, Raoult D. Changing clinical presentation of Q fever endocarditis. Clin Infect Dis
42. Houpikian P, Raoult D. Western immunoblotting for Bartonella endocarditis. Clin Diagn Lab Immunol
43. Jacobs F, Abramowicz D, Vereerstraeten P, Le Clerc JL, Zech F, Thys JP. Brucella endocarditis: the role of combined medical and surgical treatment. Rev Infect Dis
44. Jortner R, Demopoulos LA, Bernstein NE, Tunick PA, Shapira Y, Shaked Y, Kronzon I. Transoesophageal echocardiography in the diagnosis of Q-fever endocarditis. Am Heart J
45. Koehler JE, Glaser CA, Tappero JW. Rochalimaea henselae infection: a new zoonosis with the domestic cat as reservoir. JAMA
46. Koehler JE, Quinn FD, Berger TG, Leboit PE, Tappero JW. Isolation of Rochalimaea species from cutaneous and osseous lesions of bacillary angiomatosis. N Engl J Med
47. Koehler JE, Sanchez MA, Garrido CS, Whitfeld MJ, Chen FM, Berger TG, Rodriguez-Barradas MC, LeBoit PE, Tappero JW. Molecular epidemiology of Bartonella infections in patients with bacillary angiomatosis-peliosis. N Engl J Med
48. 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
49. Lamas CC, Eykyn SJ. Blood culture negative endocarditis: analysis of 63 cases presenting over 25 years. Heart
50. 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
51. Lepidi H, Fournier PE, Raoult D. Quantitative analysis of valvular lesions during Bartonella endocarditis. A case control study. Am J Clin Pathol
52. Lepidi H, Houpikian P, Liang Z, Raoult D. Cardiac valves in patients with Q fever endocarditis: microbiological, molecular, and histologic studies. J Infect Dis
53. Levy PY, Drancourt M, Etienne J, Auvergnat JC, Beytout J, Sainty JM, Goldstein F, Raoult D. Comparison of different antibiotic regimens for therapy of 32 cases of Q fever endocarditis. Antimicrob Agents Chemother
54. Li JS, Sexton DJ, Mick N, Nettles R, Fowler VG Jr, Ryan T, Bashore T, Corey GR. Proposed modifications to the Duke criteria for the diagnosis of infective endocarditis. Clin Infect Dis
55. Liang Z, Raoult D. Species-specific monoclonal antibodies for rapid identification of Bartonella quintana. Clin Diagn Lab Immunol
56. Lucero NE, Foglia L, Ayola SM, Gall D, Nielsen K. Competitive enzyme immunoassay for diagnosis of human brucellosis. J Clin Microbiol
57. Maiwald M, Schuhmacher F, Ditton HJ, von Herbay A. Environmental occurrence of the Whipple's disease bacterium (Tropheryma whippelii). Appl Environ Microbiol
58. Mallqui V, Speelmon EC, Verastegui M, Maguina-Vargas C, Pinell-Salles P, Lavarello R, Delgado J, Kosek M, Romero S, Arana Y, Gilman RH. Sonicated diagnostic immunoblot for Bartonellosis. Clin Diagn Lab Immunol
59. Maurin M, Raoult D. Q fever. Clin Microbiol Rev
60. Maurin M, Birtles R, Raoult D. Current knowledge of Bartonella species. Eur J Clin Microbiol Infect Dis
61. Maurin M, Eb F, Etienne J, Raoult D. Serological cross-reactions between Bartonella and Chlamydia species: implications for diagnosis. J Clin Microbiol
62. Maurin M, Raoult D. Antimicrobial susceptibility of Rochalimaea quintana, Rochalimaea vinsonii, and the newly recognized Rochalimaea henselae. J Antimicrob Chemother
63. Millar B, Moore J, Mallon P, Xu J, Crowe M, Mcclurg R, Raoult D, Earle J, Hone R, Murphy P. Molecular diagnosis of infective endocarditis-a new Duke's criterion. Scand J Infect Dis
64. Musso D, Drancourt M, Raoult D. Lack of bactericidal effect of antibiotics except aminoglycosides on Bartonella (Rochalimaea) henselae. J Antimicrob Chemother
65. Musso D, Raoult D. Coxiella burnetii blood cultures from acute and chronic Q fever patients. J Clin Microbiol
66. Naegeli B, Bannwart F, Bertel O. An uncommon cause of recurrent strokes. Tropheryma whippelii endocarditis. Stroke
67. Patel R, Newell JO, Procop GW, Persing DH. Use of PCR for citrate synthase gene to diagnose Bartonella quintana endocarditis. Am J Clin Pathol
68. Pedersen WR, Walker M, Olson JD, Gobel F, Lange HW, Daniel JA, Rogers J, Longe T, Kane M, Mooney MR. Value of transoesophgeal echocardiography as an adjunct to transthoracic echocardiography in evaluation of native and prosthetic valve endocarditis. Chest
69. Pesanti EL, Smith IM. Infective endocarditis with negative blood cultures: an analysis of 52 cases. Am J Med
70. Podglajen I, Bellery F, Poyart C, Coudol P, Buu-Hoi A, Bruneval P, Mainardi JL. Comparative molecular and microbiologic diagnosis of bacterial endocarditis. Emerg Infect Dis
71. Raoult D. Q fever: still a query after all these years. J Med Microbiol
72. Raoult D. Afebrile blood-culture negative endocarditis. Ann Intern Med
73. Raoult D, Birg ML, La Scola B, Fournier PE, Enea M, Lepidi H, Roux V, Piette JC, Vandenesch F, Vital-Durand D, Marrie TJ. Cultivation of the bacillus of Whipple's disease. N Engl J Med
74. Raoult D, Etienne J, Massip P, Iaocono S, Prince MA, Beaurain P, Benichou S, Auvergnat JC, Mathieu P, Bachet P. Q fever endocarditis in the south of France. J Infect Dis
75. Raoult D, Fournier PE, Drancourt M, Marrie TJ, Etienne J, Cosserat J, Cacoub P, Poinsignon Y, Leclercq P, Sefton AM. Diagnosis of 22 new cases of Bartonella endocarditis. Ann Intern Med
76. Raoult D, Fournier PE, Vandenesch F, Mainardi JL, Eykyn SJ, Nash J, James E, Benoit-Lemercier C, Marrie TJ. Outcome and treatment of Bartonella endocarditis. Arch Intern Med
77. Raoult D, Houpikian P, Tissot-Dupont H, Riss JM, Arditi-Djiane J, Brouqui P. Treatment of Q fever endocarditis. Comparison of 2 regimens containing doxycycline and ofloxacin or hydroxychloroquine. Arch Intern Med
78. Raoult D, Levy PY, Harle JR, Etienne J, Massip P, Goldstein F, Micoud M, Beytout J, Gallais H, Remy G. Chronic Q fever: diagnosis and follow up. Ann NY Acad Sci
79. Raoult D, Tissot-Dupont H, Enea-Mutillod M. Positive predictive value of Rochalimaea henselae antibodies in the diagnosis of cat scratch disease. Clin Infect Dis
80. Raoult D, Tissot-Dupont H, Foucault C, Gouvernet J, Fournier PE, Bernit E, Stein A, Nesri M, Harle JR, Weiller PJ. Q fever 1985-1998. Clinical and epidemiologic features of 1,383 infections. Medicine (Baltimore)
81. Raoult D, Vestris G, Enea M. Isolation of 16 strains of Coxiella burnetii from patients by using sensitive centrifugation cell culture system and establishment of strains in HEL cells. J Clin Microbiol
82. Relman DA. The identification of uncultured microbial pathogens. J Infect Dis
83. Relman DA. The search for unrecognized pathogens. Science
84. Roux V, Raoult D. The 16S-23S rRNA intergenic spacer region of Bartonella (Rochalimaea) species is longer than usually described in other bacteria. Gene
85. Roux V, Raoult D. Inter- and intraspecies identification of Bartonella (Rochalimaea) species. J Clin Microbiol
86. Roux V, Raoult D. Body lice as tools for diagnosis and surveillance of re-emerging diseases. J Clin Microbiol
87. Roux V, Eykyn SJ, Wyllie S, Raoult D. Bartonella vinsonii subsp. bekhoffii as an agent of afebrile blood culture-negative endocarditis in a human. J Clin Microbiol
88. Siegman-Igra Y, Kaufman O, Keysary A, Rzotkiewicz S, Shalit I. Q fever endocarditis in Israel and a worldwide review. Scand J Infect Dis
89. Spach DH, Callis KP, Paauw DS, Houze YB, Schoenknecht FD, Welch DF, Rosen H, Brenner DJ. Endocarditis caused by Rochalimaea quintana in a patient infected with human immunodeficiency virus. J Clin Microbiol
90. Spach DH, Kanter AS, Dougherty MJ, Larson AM, Coyle MB, Brenner DJ, Swaminathan B, Matar GM, Welch DF, Root RK. Bartonella (Rochalimaea) quintana bacteremia in inner-city patients with chronic alcoholism. N Engl J Med
91. Stein DS, Nelson KE. Endocarditis due to nutritionally deficient streptococci: therapeutic dilemma. Rev Infect Dis
92. Stout JE, Yu VL. Legionellosis. N Engl J Med
93. Street S, Donoghue HD, Neild GH. Tropheryma whippellii DNA in saliva of healthy people. Lancet
94. Tissot-Dupont H, Thirion X, Raoult D. Q fever serology: cutoff determination for microimmunofluorescence. Clin Diagn Lab Immunol
95. Tompkins LS, Roessler BJ, Redd SC, Markowitz LE, Cohen ML. Legionella prosthetic-valve endocarditis. N Engl J Med
96. Tonder O. Indirect haemagglutination for demonstration of antibodies to Aspergillus fumigatus. Acta Pathol Microbiol Scand Microbiol Immunol
97. Tunkel AR, Kaye D. Endocarditis with negative blood cultures. N Engl J Med
98. Van Scoy RE. Culture-negative endocarditis. Mayo Clin Proc
99. Von Reyn CF, Levy BS, Arbeit RD, Friedland G, Crumptacker CS. Infective endocarditis: an analysis based on strict case definitions. Ann Intern Med
100. Wilck MB, Wu Y, Howe JG, Crouch JY, Edberg SC. Endocarditis caused by culture-negative organisms visible by Brown and Brenn staining: utility of PCR and DNA sequencing for diagnosis. J Clin Microbiol
101. Willems H, Thiele D, Frolich-Ritter R, Krauss H. Detection of Coxiella burnetii in cow's milk using the polymerase chain reaction. J Vet Med
102. Woods GL, Walker DH. Detection of infection or infectious agents by use of cytologic and histologic stains. Clin Microbiol Rev
103. Yvorchuk KJ, Chan KL. Application of transthoracic and transesophageal echocardiography in the diagnosis and management of infective endocarditis. J Am Soc Echocardiogr