Bartonella spp. and Coxiella burnetii Associated with Community-Acquired, Culture-Negative Endocarditis, Brazil

We evaluated culture-negative, community-acquired endocarditis by using indirect immunofluorescent assays and molecular analyses for Bartonella spp. and Coxiella burnetii and found a prevalence of 19.6% and 7.8%, respectively. Our findings reinforce the need to study these organisms in patients with culture-negative, community-acquired endocarditis, especially B. henselae in cat owners.

W orldwide, Bartonella spp. and Coxiella burnetii endocarditis have varied prevalences and clinical effects (1,2). Detection is difficult in routine blood cultures, so different diagnosis methods are needed. Our study investigated the frequency of and the risk factors for Bartonella spp. and C. burnetii infection in cases of culture-negative, community-acquired endocarditis.

The Study
During January 2004-January 2009, the Infection Control Team from the university hospital at São Paulo, Brazil (Instituto do Coração-Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo) used active surveillance to identify 369 patients with endocarditis. The study focused on community-acquired endocarditis caused by fastidious bacteria. Patients >18 years of age with confirmed endocarditis were included as a prospective inception cohort of patients (3). Excluded were patients with health care-associated endocarditis (i.e., patients with prosthetic valve endocarditis in the first postoperative year, hemodialysis patients, and nosocomial endocarditis patients) (4).
Immunohistochemical and molecular methods were applied to valve tissue specimens and serum samples of patients whose serum samples were positive for Bartonella spp. or C. burnetii. DNA from paraffin-embedded valve tissue specimens and serum samples were extracted. Samples positive for Bartonella by IFA were analyzed by using 5 different PCRs to 4 distinct regions. Tissue and serum DNA from patients positive for C. burnettii by IFA were tested by quantitative PCR (online Technical Appendix  For the 221 patients in the study, findings from 10 (4.5%; 95% CI 3.96%-5.09%) patients (Figure) showed Bartonella spp., and 4 (1.8%; 95% CI 1.58%-2.04%) showed C. burnetii endocarditis. For the 51 culture-negative endocarditis patients, Bartonella spp. was found in cultures from 10 (19.6%; 95% CI 9.8%-33.1%), and C. burnetii was found in 4 (7.8%; 95% CI 2.2%-18.9%). The Table shows the immunohistochemical and molecular biology analyses for patients with positive IFA results. Bartonella spp. DNA was detected with >1 PCR in all 6 patients whose paraffin-embedded valve tissue samples were found positive for Bartonella spp. For the other 4 patients with Bartonella spp., DNA was detected in 2 serum samples. Amplicons were sequenced, and their analyses showed that the cultures from 2 patients had 100% similarity with B. quintana (GenBank accession no. BX897700.1); cultures from 4 patients had 100% similarity with B. henselae infection (GenBank accession no. BX897699.1). Cultures from 2 patients were positive for Bartonella spp. by using IFA but negative by using PCR.
All patients used antimicrobial drugs for 7 days before sample collection. All endocarditis patients whose cultures were found to be positive for C. burnetii by using IFA were also positive by using quantitative PCR: 3 by serum samples and 2 by paraffin-embedded valve tissue specimens (online Technical Appendix Table 2).
Clinical and follow-up findings from Bartonella spp. and C. burnetii endocarditis patients are shown in online Technical Appendix Table 3. Bartonella spp. infection was associated with low levels of C-reactive protein on admission and chronic symptoms related to endocarditis (online Technical Appendix Table 4). Three (75%) of 4 patients with Bartonella henselae endocarditis were associated with a cat living in the patient's home, compared with 6 (12.8%) of 47 patients with culture-negative Bartonella henselae negative endocarditis (p = 0.015 by Student t-test). Hydroxychloroquine was unavailable in our facility; therefore, we used a second-line therapy for C. burnetii endocarditis. Hydroxychloroquine was replaced with ciprofloxacin, and treatment was extended for 72 months (8). Subsequently, symptoms resolved, and antibody titers reduced substantially, considered a favorable response (9) (online Technical Appendix Table 3).

Conclusions
In this study, the systematic use of IFA detected a 4.5% (10/221) prevalence of community-acquired endocarditis due to Bartonella spp. and a 1.8% (4/221) prevalence due to C. burnetii. For the 51 culture-negative endocarditis patients, IFA enabled recognition of the endocarditis etiology in 14 (27.5%) patients (Bartonella spp. in 10 [19.6%] and C. burnetii in 4 [7.8%]). Some of these patients have been recognized as having the first cases of endocarditis caused by these microorganisms in Brazil (10,11).
Prevalences of Bartonella spp. endocarditis vary worldwide by region studied (1). In a broad series of 759 culture-negative endocarditis patients in France, serum samples showed high sensitivity for detection of C. burnetii and Bartonella spp. infections, compared with other diagnostic tools, such as PCR, cell culture, and immunohistochemical analysis (2). In Brazil, studies of Bartonella spp. infection among culture-negative endocarditis patients have shown varied results. A retrospective case series of 51 surgically treated, culture-negative endocarditis patients found 2 cases of Bartonella spp. and 1 case of C. burnetii by using PCR on valvular tissue (12). Another series of 46 culture-negative endocarditis patients from the city of São Paulo used PCR to investigate Bartonella spp. in blood and found 13 (28%) patients with positive results (13).
We found an association between B. henselae endocarditis and the presence of a cat living at a patient´s home, a risk factor indicating that clinicians should consider this infection when assessing endocarditis patients. The relatively small sample of patients with endocarditis caused by Bartonella spp. and C. burnetii limited the statistical analyses of factors associated with these infections. Serologic investigations of infections by these agents were applied only to patients with negative cultures. Although rare (2,14), co-infection by these microorganisms in culture-positive endocarditis is possible, so frequency of Bartonella spp. and C. burnetii infections in these patients may be higher than shown. Our study indicates that systematic serologic research for Bartonella spp. and C. burnetii in community-acquired, culture-negative endocarditis may be clinically useful, particularly in screening for B. henselae in cat owners.
Dr. Siciliano is an infectious disease specialist working in infection control at the Heart Institute at the university hospital at São Paulo, Brazil, a tertiary care hospital dedicated to care of heart disease patients. His clinical research interest is endocarditis.