Case series of Bartonella quintana blood culture-negative endocarditis in Washington, DC

Introduction: Prior studies (predominantly from Europe) have demonstrated blood culture-negative endocarditis due to Bartonella. Our objective was to describe three cases of Bartonella quintana endocarditis identified within one year at a large hospital in Washington, DC, USA. Case presentation: We constructed a descriptive case series from a retrospective review of medical records from April to December 2013 at an 800-bed urban hospital. All three patients (ages: 52, 55 and 57 years) were undomiciled/homeless men with a history of alcoholism. Although they had negative blood cultures, echocardiography demonstrated aortic/mitral valve perforation and regurgitation in one patient, aortic/mitral valve vegetation with mitral regurgitation in the second patient, and aortic valve vegetation with regurgitation in the third patient. The patients had positive Bartonella quintana serum immunoglobulin G (IgG) with negative immunoglobulin M (IgM). PCR on DNA extracted from cardiac valves was positive for Bartonella, and DNA sequencing of PCR amplicons identified Bartonella quintana. Patients received treatment with doxycycline/rifampin or doxycycline/gentamicin. Conclusion: Clinicians should consider Bartonella endocarditis as a differential diagnosis in patients who fit elements of the Duke Criteria, as well as having a history of homelessness and alcoholism.


Introduction
Bartonella quintana is a small, fastidious Gram-negative rod that caused trench fever during World War I. The organism can lead to chronic bacteraemia and endocarditis among the homeless and in patients who chronically abuse alcohol (Foucalt et al., 2002;Fournier et al., 2001). The human body louse (Pediculus humanus) is a known vector for Bartonella (Ohl & Spach, 2000). Bartonella is considered an uncommon aetiology of infective endocarditis as it is reported in only 3 % of cases in the literature (Lamas & Eykyn, 2003). In 1993, two separate reports first implicated Bartonella in endocarditis (Rolain et al., 2004;Spach et al., 1993), and over the years, the bacteria became more appreciated as a significant cause of blood culture-negative endocarditis (Daly et al., 1993).
Haemophilus, Aggregatibacter, Cardiobacterium, Eikenella and Kingella (HACEK) organisms more commonly cause blood culture-negative endocarditis, and they are easily isolated when the bacteria are incubated for at least five days. Hence, 'true' blood culture-negative endocarditis is due to other fastidious organisms, notably Coxiella burnetii and Bartonella spp. Six species of Bartonella can cause infective endocarditis in humans: Bartonella quintana, B. henselae, B. elizabethae, B. vinsonii, B. koehlerae and B. alsatica (Rolain et al., 2004;Spach et al., 1993;Daly et al., 1993). Of note, B. quintana and B. henselae have been associated with over 95 % of culture-negative infective endocarditis. Due to difficulties in culturing Bartonella from blood, diagnosis typically requires serological and molecular methodologies. The two most commonly used serological methods for Bartonella infections are indirect immunofluorescence assay (IFA) and enzyme-linked immunosorbent assay (ELISA). Titres of 1 : 1600 (via a microimmunofluorescence test) have a positive predictive value of 0.88 (Daly et al., 1993).
PCR-based tests on blood and cardiac valvular tissue have an important role in the diagnosis of Bartonella endocarditis, even in the setting of prior antibiotic use. One study reported positive PCR testing for Bartonella on cardiac valve Abbreviation: IFA, Immunofluorescence assay. tissue in more than 95 % of patients with Bartonella endocarditis . Some of the most commonly cited molecular targets for diagnosis of Bartonella endocarditis by PCR include 16S rRNA gene, internal transcribed spacer (ITS), RNA polymerase beta-subunit-encoding gene (rpoB) and citrate synthase gene (gltA). The 16S rRNA gene has been the focus for most PCR methods because it is one of the most conserved genes (Weisburg et al., 1991). However, the 16S rRNA genes of species of the genus Bartonella share more than 97.8 % similarity (Birtles, 1995), and as a result, differences between them are not sufficient for needed discrimination between species. PCR amplification targeting the 16S-23S rRNA gene ITS region is considered highly sensitive, but amplification specificity depends on primer design, and because of published concerns about nonspecific PCR amplification of species of the genus Mesorhizobium associated with ITS (Maggi & Breitschwerdt, 2005), we elected to use the gltA method which is now commercially available at the Mayo laboratory. Citrate synthase is the first enzyme of the tricarboxylic acid cycle and thus is a key regulator of intracellular adenosine triphosphate production in both prokaryotic and eukaryotic cells (Wiegand & Remington, 1986) The use of gltA comparisons to estimate divergence among closely related species was first proposed by Regnery and his colleagues, who demonstrated genotypic relationships among species of the genus Rickettsia by comparing multiple restriction endonuclease digestion patterns of partial gltA PCR amplicons (Regnery et al., 1991).
In the past decade, we have dramatically increased our awareness and understanding of human disease caused by Bartonella spp. Investigators from France have generated the majority of reports on endocarditis due to Bartonella, including several multi-centre international studies with patients from France, England and Canada ; Fournier et al., 2001;Daly et al., 1993). Until now, only one report of B. quintana bacteraemia (with and without endocarditis) in homeless persons without human immunodeficiency virus (HIV) infection had been reported in the USA, specifically in Seattle, WA (Spach et al., 1993).
We report three cases of B. quintana endocarditis that occurred within one year at our hospital in Washington, DC. We also compare our data to the increasing number of blood culture-negative endocarditis cases reported in the literature.

Methods
We performed a retrospective review of medical records after receiving Institutional Review Board (IRB) approval. Patients were evaluated by the infectious diseases consultation service, and laboratory testing for Bartonella was ordered as part of the clinical evaluation. The diagnosis of Bartonella endocarditis was based on the modified Duke criteria (Li et al., 2000) and microbiological findings (including serological and PCR results). Bartonella serological and PCR assays were performed at Mayo Medical Laboratories in Rochester, MN. Serology for Bartonella was performed with an IFA technique using antigen substrate slides consisting of two separate cell cultures infected with Bartonella henselae or Bartonella quintana as described by Maurin & Raoult (1996). An immunoglobulin M (IgM) titre of >1 : 20 was suggestive of a current infection with either B. henselae or B. quintana, while an immunoglobulin G (IgG) titre of >1 : 128 suggested a current or previous infection. PCR testing was performed on DNA extracted from cardiac valve tissue. A portion of the citrate gltA of B. henselae served as a target sequence, and a base substitution in the donor probe sequence was used to achieve greater separation in the melting-curve analysis between B. henselae and B. quintana. (Cockerill & Uhl, 2002). This assay was capable of identifying the genus Bartonella, but could not reliably distinguish among the various species. DNA sequencing of PCR amplicons was performed using the Promega gel purification kit and two picomoles of forward and reverse primers. Sequencing was performed with the Big Dye sequencing method on an Applied Biosystems 3130 genetic analyzer, and the resulting sequence was queried via the Basic Local Alignment Search Tool (BLAST) for species identification (Avidor et al., 2004).

Results
Three cases of infective endocarditis caused by B. quintana were identified between April and December 2013 (Table 1). All three patients were homeless and had a history of alcoholism. Blood cultures were negative on multiple occasions for all patients, but echocardiography demonstrated aortic/ mitral valve perforation and severe regurgitation in one patient, aortic/mitral valve vegetation with severe mitral regurgitation in the second patient, and large aortic valve vegetation with severe regurgitation in the third patient.  Due to cardiac valve destruction or incompetence, valve replacement was required for all patients. DNA extracted from cardiac valve tissue was identified by PCR as part of the genus Bartonella, and subsequent DNA sequencing and BLAST alignment revealed 99 % sequence similarity with B. quintana for the three cases. All patients received doxycycline for six weeks. Two patients received concomitant rifampin for two weeks, while one patient received gentamicin for one week (but it was changed to rifampin for an additional week due to aminoglycoside-induced nephrotoxicity).

Discussion
Species of the genus Bartonella are fastidious Gram-negative bacteria that require specific laboratory conditions for laboratory growth and identification. In our case series, we observed a cross-reaction between Bartonella and Chlamydophilia in one patient. Identification of Bartonella species with serology was unreliable in our patients, as B. henselae IgG titres were higher than B. quintana titres in two patients, and equivalent in one patient. All of our patients had detectable Bartonella DNA from cardiac valve tissue, and even though PCR of valve tissue reportedly has 100 % specificity, the sensitivity of this test can be compromised with prior antibiotic use (Fournier et al., 2010). DNA sequencing of the amplified PCR products and comparison with BLAST submissions allowed us to establish B. quintana as the causative bacteria in our cohort.
Similar to other cases reported in the literature, our patients were homeless, abused alcohol, and ultimately required cardiac valve replacement. In a report of 101 patients with Bartonella endocarditis, Raoult et al. (2003) reported that 85 % were male, 38 % were homeless, 48 % were alcoholic, and 75 % underwent valvular surgery because of severe valve damage at the time of diagnosis (Raoult et al., 2003). Due to the limited number of cases, the optimum treatment for suspected or confirmed Bartonella endocarditis is unclear, but current recommendations suggest doxycycline for six weeks plus gentamicin sulfate for two weeks. In our cohort, all three patients eventually received treatment with doxycycline/rifampin.
In summary, although much of the existing literature regarding Bartonella endocarditis stems from Europe, our findings suggest that Bartonella endocarditis should also be considered in the USA (particularly in patients who are homeless and have a history of alcoholism). Papers by both Fournier et al and Beitkopt et al have developed an algorithm that suggests a diagnostic strategy for evaluating blood culture-negative endocarditis and recommends serological testing for Bartonella (Fournier et al., 2010;Breitkopf et al., 2005). Based on our case series, we concur with that recommendation, and suspect that Bartonella endocarditis may be an under-recognized cause of blood culture-negative endocarditis in the USA. Our improved knowledge of the spectrum of pathogens involved in blood culture-negative endocarditis should also advance our empiric antibiotic therapy selections.