Endocarditis caused by Bartonella

INTRODUCTION — Bartonella was first described as a cause of endocarditis in two separate reports in 1993 [1,2] and has subsequently become appreciated as a significant cause of "culture-negative" endocarditis [3-11]. In the United States, Bartonella is one of the most common, if not the most common, cause of culture-negative endocarditis.

At least eight Bartonella species have been reported to cause infective endocarditis in humans: B. quintana, B. henselae, B. elizabethae, B. vinsonii, B. koehlerae, B. clarridgeiae, B. washoensis, and B. alsatica [3,12-15]. However, more than 95 percent of the cases have involved either B. quintana or B. henselae.

This topic will review the epidemiology, clinical manifestations, diagnosis, and treatment of Bartonella endocarditis. Other Bartonella infections are discussed in separate topic reviews. (See "Microbiology, epidemiology, clinical manifestations, and diagnosis of cat scratch disease" and "Bartonella infections in people with HIV" and "Bartonella quintana infections: Clinical features, diagnosis, and treatment" and "South American bartonellosis: Oroya fever and verruga peruana".)

EPIDEMIOLOGY — The epidemiologic features of patients documented to have Bartonella endocarditis have varied considerably. Most reports have involved adults, although multiple cases have also been described in children [10,16]. Overall, approximately 70 percent of reported cases have involved men [11,17,18]. Case reports have also described Bartonella endocarditis in persons with HIV [1], and there are rare reports of Bartonella endocarditis in persons who have undergone solid organ transplantation [19].

Certain risk factors appear to be associated with Bartonella endocarditis. These include:

●Demographic factors and exposures – Available data suggest that lack of housing, alcohol use disorder, and infestation with body lice are associated with B. quintana endocarditis, whereas contact with cats serves as the major risk factor for B. henselae endocarditis [3,4,18].

●Valvular heart disease – A significant proportion (approximately 40 to 60 percent) of persons with Bartonella endocarditis have prior cardiac valvular disease, particularly when B. henselae is the cause of the endocarditis [3,11,20]. As an example, in a large series, which included 22 patients with Bartonella endocarditis, approximately 55 percent had evidence of pre-existing cardiac valvular disease [3]. (See "Bartonella infections in people with HIV".)

●Congenital heart disease – Multiple cases of Bartonella endocarditis have been reported that involved children, adolescents, and adults with congenital heart disease [21-24]. The timely recognition and diagnosis of Bartonella endocarditis in persons with congenital heart disease are particularly important, since delayed diagnosis can have severe consequences, especially if repair of a valve or graft is required.

CLINICAL MANIFESTATIONS — Patients with Bartonella endocarditis have clinical manifestations similar to other patients with subacute bacterial endocarditis as described by the Duke criteria [25]. (See "Clinical manifestations and evaluation of adults with suspected left-sided native valve endocarditis", section on '2023 Duke-ISCVID criteria'.)

The epidemiologic features of patients with documented Bartonella endocarditis are described above. (See 'Epidemiology' above.)

Affected patients typically present with subacute, nonspecific symptoms that include fever, fatigue, weakness, and weight loss. In addition, most patients have evidence of a murmur on cardiac auscultation. Based on reports from several case series, embolic events appear to be common with Bartonella endocarditis, but the true incidence is difficult to determine due to limited data [3,11,17]. In some patients, one or more typical features of endocarditis, such as fever, elevated white blood cell count, and elevated erythrocyte sedimentation rate (ESR), may be absent [11].

Although most cases involve native valves, case reports of prosthetic valve endocarditis have also been described. Such patients typically have an aggressive disease course characterized by valvular perforation or rapid progression of heart failure [21,26,27].

Patients with Bartonella endocarditis can also develop an immune-complex glomerulonephritis [28-32]. Cases of anti-proteinase 3 (PR3)-positive glomerulonephritis in persons with Bartonella endocarditis have increasingly been reported in the past 10 years [29,30,33,34]. Many of the cases present as a necrotizing antineutrophil cytoplasmic antibody (ANCA)-positive glomerulonephritis, associated with antibodies against PR3. In patients who initially present with glomerulonephritis and the cause is unknown, clinicians should consider evaluation for Bartonella endocarditis as a potential cause, especially if the patient has an epidemiologic risk for Bartonella, such as homelessness or cat exposure. (See "Clinical spectrum of antineutrophil cytoplasmic autoantibodies".)

FINDINGS ON ECHOCARDIOGRAPHY — Echocardiography shows valvular vegetations in approximately 90 percent of patients with Bartonella endocarditis [11]. However, these vegetations do not appear to have unique or distinguishing features. (See "Role of echocardiography in infective endocarditis" and "Clinical manifestations and evaluation of adults with suspected left-sided native valve endocarditis", section on 'Cardiac imaging'.)

Approximately 60 percent of patients with Bartonella endocarditis have involvement of the aortic valve [11]. Several cases have involved multiple valves, usually the aortic and mitral or the aortic and tricuspid valves. The reason for the apparent predilection to infect the aortic valve is unknown. Isolated right-sided endocarditis appears to occur infrequently [11,35].

EVALUATION AND DIAGNOSIS — This section will review the approach to diagnosis, as well as the different diagnostic tests that are used.

Approach to diagnosis — Bartonella endocarditis should be suspected in patients with clinical and echocardiographic findings that suggest endocarditis if traditional blood cultures are negative after 72 to 96 hours of incubation and the patient has epidemiologic risk factors for Bartonella infection. (See 'Epidemiology' above.)

The evaluation of patients with possible Bartonella endocarditis should include:

●Obtaining at least two sets of blood cultures for Bartonella. It is important to communicate with the microbiology laboratory to ensure special methods are used to optimize yield for growth of Bartonella species. (See 'Culture' below.)

●Serologic testing for B. henselae and B. quintana; the test is highly predictive of endocarditis if there is a high titer for immunoglobulin G (IgG) antibodies to either of these species (eg, ≥1:1024 with the assay commonly used in the United States and ≥1:800 for the assay used in France) [17]. (See 'Serology' below.)

●Serum or plasma polymerase chain reaction (PCR) testing for Bartonella. (See 'Polymerase chain reaction' below.)

For patients who undergo surgical removal of a cardiac valve, Bartonella PCR should be performed on the tissue sample. In addition, a Warthin-Starry stain should be performed if feasible, although this test is labor intensive and not routinely performed as a tissue histologic stain. (See 'Polymerase chain reaction' below and 'Histopathology' below.)

A positive result for any of these tests should be considered a major Duke criterion for the diagnosis of Bartonella endocarditis [17,36]. (See "Clinical manifestations and evaluation of adults with suspected left-sided native valve endocarditis", section on '2023 Duke-ISCVID criteria'.)

In a retrospective study of 348 patients with suspected blood culture-negative endocarditis, Bartonella spp were identified as the etiologic agent in 99 patients (28 percent) when serologic tests and direct PCR from valve specimens (when available) were utilized along with culture [37]. In a subsequent prospective study involving 819 patients with culture-negative endocarditis [38], Bartonella species accounted for the diagnosis in 86 (18 percent) of the 476 patients who had a specific pathogen identified when serologic and PCR testing was included in the diagnostic evaluation.

Considerations for those with glomerulonephritis — Persons with Bartonella endocarditis who have evidence of glomerulonephritis should undergo evaluation for ANCA and proteinase 3 (PR3) antibodies. The diagnosis of immune-mediated glomerulonephritis is particularly important with Bartonella endocarditis, since use of an aminoglycoside as part of treatment of endocarditis would enhance the risk of nephrotoxicity in this setting.

Diagnostic tests — The diagnosis of Bartonella endocarditis often requires modalities other than blood culture to identify the organism. Blood cultures require prolonged incubation before they turn positive, if they turn positive at all [3,37,38].

Culture — Bartonella species are fastidious gram-negative bacteria that require specific laboratory conditions for optimal growth. In a multicenter international study, only 5 of the 22 patients diagnosed with Bartonella endocarditis (approximately 25 percent) had a positive blood culture at any time in their clinical course [3].

Ideally, blood culture samples should be collected in pediatric or adult isolator tubes (Wampole, Cranbury, New Jersey), or in blood tubes containing ethylenediamine tetra-acetic acid (EDTA). The blood culture sample should then be added onto an agar plate that contains both chocolate agar (to optimize isolation of B. quintana) and heart infusion agar supplemented with 5 percent rabbit blood (to optimize isolation of B. henselae). The addition of a source of heme, such as rabbit blood, is important since Bartonella species require heme for growth [39]. The microbiology laboratory should use fresh agar plates (within seven days after pouring) and incubate the samples in 5 percent CO₂ at 35 to 37°C for a minimum of 28 days. If growth of colonies is detected, then subcultures should be performed and growth of Bartonella colonies in the subcultures typically occurs within seven days [40].

Bartonella species have also occasionally been isolated from BACTEC bottles, and by blind subculture of BACTEC bottle cultures, followed by acridine orange staining and subculture on chocolate and heart infusion agar supplemented with rabbit blood [41].

Isolation of Bartonella from cardiac valvular tissue samples has a very low yield, but if attempted, tissue homogenates should be directly plated onto agar and cocultivated with a bovine endothelial cell monolayer [42]. When trying to culture Bartonella, it is important to proactively communicate with the microbiology laboratory so they can optimize culture techniques and can extend the incubation period for a minimum of 28 days.

Polymerase chain reaction — Qualitative polymerase chain reaction (PCR) testing is performed by amplifying a 296 base pair fragment of the 16S ribosomal RNA. Multiple commercial laboratories now offer PCR testing for Bartonella on tissue or blood samples, but only some differentiate B. quintana from B. henselae.

The use of PCR-based tests on cardiac valvular tissue has played an important role in the diagnosis of Bartonella endocarditis [20,43]. As an example, in one study, investigators reported positive PCR testing for Bartonella on cardiac valve tissue in greater than 95 percent of patients with Bartonella endocarditis [20]. The organism was detected even though more than 60 percent of these patients had their valves analyzed following the receipt of antibiotics. Although paraffin-embedded specimens can be used, the yield appears to be greater using fresh valvular tissue.

PCR testing can also be performed on whole blood, plasma, or serum samples. In one study, investigators performed Bartonella PCR testing of serum samples taken from patients with a diagnosis of Bartonella endocarditis (based upon serologic, culture, and/or molecular detection in valvular tissue) [44]. In such patients, serum PCR testing demonstrated a sensitivity and specificity of 58 and 100 percent, respectively. In a French review of 106 cases of Bartonella endocarditis, PCR testing had a sensitivity of 92 percent for valvular biopsy specimens, 36 percent for serum samples, and 33 percent for blood samples [17].

Serology — In patients with suspected Bartonella endocarditis, serologic testing can play a supportive role in the appropriate clinical setting. In the United States, Bartonella serology testing is commercially available from a number of laboratories.

Indirect immunofluorescence assay (IFA) is the preferred and the most commonly used serologic method for diagnosing Bartonella infections. For persons with suspected Bartonella endocarditis, we rely predominantly on IFA IgG titer and not IgM titers. Available data suggest that IgM antibody titers to Bartonella are often negative in persons with Bartonella endocarditis and therefore are not considered a reliable diagnostic test in this setting [36]. One study found that Western immunoblotting had a very high sensitivity and specificity for identifying the causative species in cases of Bartonella endocarditis, but this test is not readily available [45].

There are several limitations to serologic testing. These include:

●Lack of uniform diagnostic criteria – Clearly defined, uniform cut-off titers for a positive Bartonella serology do not exist since there are subtle differences in the IFA assays used, and the predictive values of titers in relation to the different clinical manifestations and diseases caused by Bartonella infection can vary. As an example, although most laboratories in the United States consider an IgG titer of ≥1:64 or ≥1:128 to either B. henselae or B. quintana as a positive IgG test, persons with Bartonella endocarditis typically have a Bartonella IgG serologic titer of ≥1:1024. This titer of 1:1024 is roughly equivalent to an IgG titer of 1:800 for the assay used in French laboratories, where much of the endocarditis literature has been published [17,36].

However, these suggested cut-off values should not be considered rigid diagnostic criteria. Several studies have shown a substantial percentage of persons with Bartonella endocarditis will have IgG values below these suggested cut-offs [17].

●Cross-reactivity – Significant cross-reactivity occurs at the species level between B. henselae and B. quintana, as well as with some other Bartonella species [45].

Cross-reactions can also occur between Bartonella and Chlamydia spp. In one series, nine patients previously thought to have Chlamydia endocarditis were reclassified as Bartonella endocarditis based on cross-adsorption studies [3]. The process of cross-adsorption consists of incubating patient serum with the bacteria suspected of causing cross-reaction in the serologic test. The adsorption process removes antibodies to the bacterium that caused serologic cross-reaction, but the antibodies to the organism causing the disease remain.

Among patients with Bartonella endocarditis, there is frequent cross-reactivity with Coxiella antigens, which can produce a false-positive result for Coxiella (but it is usually Phase II antibodies and does not increase with time). Further information regarding Q fever can be found on the United States Centers for Disease Control and Prevention website [46,47].

Histopathology — Cardiac valvular tissue stained with hematoxylin and eosin characteristically shows mildly inflamed connective tissue with focal granulation tissue reaction [4]. Warthin-Starry silver staining of these aggregates, if positive, reveals masses of small, dark-staining bacteria. When electron microscopy is performed, pleomorphic bacilli are seen predominantly located intracellularly [3]. Gram staining of cardiac valvular tissue is not useful for diagnosing Bartonella endocarditis.

In one study of 15 patients with confirmed Bartonella endocarditis, the cardiac valve pathology showed more fibrotic and calcified changes, less vascularization, and less extensive vegetations; these features suggested chronic infection when compared with cases of infective endocarditis caused by other pathogens [48]. In addition, approximately 11 (85 percent) of 13 had a positive Warthin-Starry stain of the valve tissue and 10 (77 percent) of the 13 had a positive Bartonella immunohistologic test.

ANTIMICROBIAL THERAPY — The mainstay of treatment for Bartonella endocarditis is antimicrobial therapy. In some patients, surgical management may be required as well. The indications for surgery in patients with Bartonella endocarditis are the same as for other causes of infectious endocarditis. (See 'Surgery' below.)

There are insufficient data to guide definitive recommendations for treating Bartonella endocarditis since there are no randomized trials; the existing literature consists of case series and case reports [3,11,17]. In addition, most patients with Bartonella endocarditis have undergone cardiac valvular surgery with removal of the infected valve, making it difficult to assess the separate role of antibiotics.

Treatment guidelines for suspected and proven Bartonella endocarditis were put forth by an expert panel in 2004 [40] and were subsequently incorporated into the 2005 American Heart Association Infective Endocarditis guidelines [49]. The 2015 American Heart Association endocarditis guidelines have not provided specific treatment recommendations for Bartonella endocarditis [50]. The following recommendations differ from the 2004 expert panel recommendations based on more recent data regarding renal disease in persons with Bartonella endocarditis, particularly Bartonella-associated anti-proteinase 3 (PR3)-positive glomerulonephritis [29,30,33,34]. (See 'Clinical manifestations' above.)

Preferred regimen — For nonpregnant adults with confirmed Bartonella endocarditis, we suggest combination therapy with doxycycline and rifampin. Considerations for pregnant women and young children are discussed below. (See 'Considerations in pregnant persons and young children' below.)

The dosing and duration of these agents are as follows:

●Doxycycline (100 mg orally or intravenous [IV] every 12 hours) should be administered for three months.

●Rifampin (300 mg orally or IV every 12 hours) should be administered for the first six weeks of therapy.

If a patient with culture-negative endocarditis has suspected, but not proven, Bartonella infection, an additional agent, ideally ceftriaxone (2 g IV once daily for six weeks), should be added to cover other likely causes of culture-negative endocarditis. Although ceftriaxone probably has some activity against Bartonella [51], its role in this setting is primarily to treat common culture-negative endocarditis pathogens that are not effectively treated by doxycycline. (See "Antimicrobial therapy of left-sided native valve endocarditis", section on 'Culture-negative endocarditis'.)

Doxycycline is the key component of therapy for Bartonella endocarditis, based on available reports describing treatment of Bartonella endocarditis and other types of Bartonella infections [52-54]. However, since doxycycline is bacteriostatic, we suggest that it be used in combination with a second antimicrobial and administered for a prolonged duration. The use of a prolonged three-month course of doxycycline for Bartonella endocarditis is also based on the lack of definitive data on treatment duration, the intracellular location of Bartonella, and the relatively low risk of developing serious adverse reactions with an extended course of oral doxycycline.

Rifampin should ideally be included as the second antimicrobial in combination with doxycycline. Although there are no studies that have examined use of rifampin in combination with a tetracycline for the treatment of Bartonella endocarditis, rifampin has excellent intracellular penetration, which is important when treating Bartonella infections, particularly Bartonella endocarditis. In addition, rifampin has been shown to be effective as a component of therapy with other serious Bartonella infections. Previously, gentamicin had been the preferred second agent, but we now consider gentamicin an alternative for those who cannot take rifampin (eg, due to drug interactions) given the significant risk of nephrotoxicity. (See 'Alternative regimens' below.)

Alternative regimens — The approach to choosing an alternative regimen for patients with confirmed Bartonella endocarditis depends upon why the patient cannot take the preferred regimen.

●If doxycycline cannot be used – For patients who are intolerant to doxycycline, azithromycin (500 mg IV/orally once daily) for three months is an alternative agent. However, the efficacy of macrolides is likely lower compared with tetracyclines for the treatment of Bartonella endocarditis [40]. Considerations for regimen selection in pregnant women and young children are discussed below. (See 'Considerations in pregnant persons and young children' below.)

Similar to doxycycline, azithromycin should be administered with a second agent (preferably rifampin). (See 'Preferred regimen' above.)

●If rifampin cannot be used – For patients who cannot use rifampin (eg, due to drug interactions), doxycycline plus gentamicin is a reasonable alternative in those with normal renal function and no evidence of Bartonella-associated glomerulonephritis. The preferred dose of gentamicin is 3 mg/kg/day given once daily for the initial 14 days of therapy; however, some experts dose gentamicin as 1 mg/kg IV every eight hours since this dosing regimen has been used in most of the Bartonella endocarditis case reports. The once-daily gentamicin dose should be adjusted, if needed, to achieve trough serum concentrations of <1 mcg/mL. If every-eight-hour dosing is used, doses should be adjusted to achieve peak serum concentrations of 3 to 4 mcg/mL and trough concentrations less than 1 mcg/ml; the peak level should be obtained one hour after administration and the trough level 24 hours after the first dose and then twice weekly thereafter. In addition, renal function should be monitored closely. The benefit of including aminoglycoside therapy was supported in a retrospective analysis of 101 patients with Bartonella endocarditis that found persons who received a regimen that included at least 14 days of an aminoglycoside had higher recovery and survival rates compared with those who were treated with a regimen without an aminoglycoside [11].

For those who cannot receive either gentamicin or rifampin as the second agent, a less preferable option is monotherapy with doxycycline alone. However, in this setting we extend the duration of treatment for Bartonella endocarditis to six months. If doxycycline cannot be used, then azithromycin should be given for six months.

Considerations in pregnant persons and young children — When choosing a treatment regimen for Bartonella endocarditis in pregnant persons and children <8 years of age, the benefit of using a regimen that contains doxycycline (the preferred regimen for treatment of Bartonella endocarditis) must be weighed against potential antimicrobial toxicity. Pregnant persons should be managed in consultation with a maternal-fetal medicine specialist.

Since there are no data to guide treatment decisions in these patient groups, the following key points related to the risks and benefits of doxycycline should be considered when determining the treatment regimen.

●The tetracyclines are a key component of therapy for Bartonella endocarditis, which can be a life-threatening disease. Evidence supporting the use of doxycycline is discussed above. (See 'Preferred regimen' above.)

Azithromycin, which is considered safe for use in pregnancy and in young children, is the best option as an alternative to doxycycline to treat Bartonella endocarditis. However, there is less experience with macrolides for treatment of endocarditis, and in general, they are felt to be less effective than doxycycline. (See 'Alternative regimens' above.)

●The use of tetracyclines are generally contraindicated in pregnancy because of the risk of hepatotoxicity in the mother [55] and adverse effects on fetal bone and teeth [56,57]. Subsequent evidence has suggested the relative safety of doxycycline compared with older tetracyclines in both pregnant women and children when a short course is used [58,59]. These studies, however, are low quality and insufficient to conclude that there is no risk. In addition, it is unclear if these findings can be applied to the longer courses required for treatment of endocarditis. (See "Tetracyclines", section on 'Pregnant or breastfeeding women'.)

●Tetracycline antibiotics have been associated with permanent tooth discoloration in children less than eight years of age if used repeatedly or for prolonged courses. However, doxycycline binds less readily to calcium than other tetracyclines, and the risk of dental staining with doxycycline is minimal if a short course is administered [58,60-62]. The American Academy of Pediatrics permits use of doxycycline for ≤21 days in children of all ages [63], but there are limited safety data when doxycycline is used for >21 days in children less than eight years of age. (See "Tetracyclines", section on 'Young children'.)

SURGERY

Indications — The indications for surgery in patients with Bartonella endocarditis are the same as for other causes of infectious endocarditis. The one exception is the indication based upon persistently positive blood cultures (longer than five to seven days despite appropriate antimicrobial therapy) [50]. Persistent bacteremia is not a feasible indicator to use in patients with Bartonella endocarditis, due to the prolonged time required for growth of Bartonella species in blood cultures. Topic reviews that discuss the indications for surgery in patients with endocarditis are presented elsewhere. (See "Surgery for left-sided native valve infective endocarditis" and "Surgery for prosthetic valve endocarditis".)

In case reports of patients with Bartonella endocarditis, approximately 60 percent have undergone cardiac valvular surgery during initial hospitalization [3,4,20]. The reason for the high proportion of patients undergoing valvular surgery is not precisely known but is thought to result primarily from the higher incidence of pre-existing valvular abnormalities, the indolent course, and delayed diagnosis [11,17,40].

Implications for antimicrobial therapy — Some patients with Bartonella endocarditis undergo cardiac valvular surgery and the infected valve(s) is removed. This can impact the approach to antimicrobial therapy.

●If antimicrobial therapy was initiated prior to surgery – If antimicrobial therapy was started prior to surgery, the remaining course of doxycycline plus rifampin therapy should be completed post-operatively (ie, the patient receives a total of three months of doxycycline and six weeks of rifampin).

However, there are a few situations in which this approach may need to be modified:

•If the patient was receiving gentamicin and had not completed the two-week course of gentamicin prior to surgery, we do not continue gentamicin post-operatively. This is due to the enhanced risk of nephrotoxicity with an aminoglycoside during and following cardiac surgery and the likelihood that the burden of infection was reduced with the previous antibiotics and the removal of the valve.

•If there were <2 weeks of the treatment regimen remaining, or if the entire antimicrobial course had been completed prior to cardiac surgery, we give an additional two weeks of doxycycline post-operatively.

●If the diagnosis is made post-operatively – On occasion, a diagnosis of Bartonella endocarditis is made post-operatively and the patient has not received pre-operative antimicrobial therapy. In this situation, a six-week course of doxycycline plus rifampin may be sufficient, assuming all infected valve tissue has been removed. If rifampin cannot be used, then gentamicin should be used for the initial 14 days of therapy as an alternative but with very careful monitoring.

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Treatment and prevention of infective endocarditis".)

SUMMARY AND RECOMMENDATIONS

●Microbiology/epidemiology – Six species of Bartonella have been reported to cause infective endocarditis in humans, although the vast majority of cases have involved either B. quintana or B. henselae.

Homelessness, alcoholism, and infestation with body lice are associated with B. quintana endocarditis, whereas contact with cats and previous valvular disease are the major risk factors for B. henselae endocarditis. (See 'Epidemiology' above.)

●Clinical manifestations – Patients with Bartonella endocarditis present with fever, fatigue, weakness, and weight loss, similar to those with other forms of subacute bacterial endocarditis. Most also have evidence of a murmur on cardiac auscultation. Patients with Bartonella endocarditis can also develop an immune-complex glomerulonephritis. (See 'Clinical manifestations' above.)

●Evaluation and diagnosis – Bartonella endocarditis should be suspected in patients with clinical and echocardiographic findings that suggest endocarditis if traditional blood cultures are negative after 72 to 96 hours and the patient has epidemiologic risk factors for disease. The initial evaluation of patients with possible Bartonella endocarditis should include the following (see 'Evaluation and Diagnosis' above):

•Blood cultures using special methods to optimize growth. (See 'Culture' above.)

•Serologic testing using either an indirect immunofluorescence assay (IFA) or an enzyme-linked immunosorbent assay (ELISA). A high titer for IgG antibodies (eg, ≥1:1024 with the assay commonly used in the United States) to either B. henselae or B. quintana strongly suggests a diagnosis of Bartonella endocarditis. (See 'Serology' above.)

•Polymerase chain reaction (PCR)-based testing for Bartonella on serum or plasma. (See 'Polymerase chain reaction' above.)

For patients with culture-negative endocarditis who undergo cardiac valve replacement, PCR-based testing for Bartonella should be performed on the cardiac valvular tissue. In addition, the tissue should be stained with Warthin-Starry silver staining, which may demonstrate masses of small, dark-staining bacteria. (See 'Polymerase chain reaction' above and 'Histopathology' above.)

●Antimicrobial therapy – The mainstay of treatment for Bartonella endocarditis is antimicrobial therapy. In some patients, surgical management may be required as well. Treatment recommendations are based upon case series and case reports.

•For nonpregnant adults with proven Bartonella endocarditis, we suggest treatment with doxycycline (100 mg orally or intravenous [IV] every 12 hours for three months) plus rifampin (300 mg orally or IV every 12 hours for the first six weeks of therapy) (Grade 2C). (See 'Preferred regimen' above.)

•For those who cannot take rifampin, gentamicin (eg, 3 mg/kg IV every 24 hours for 14 days) is a reasonable alternative for those with normal kidney function. A less preferred option is doxycycline alone for six months if the patient cannot take rifampin or gentamicin. (See 'Alternative regimens' above.)

For patients who are intolerant of doxycycline, azithromycin can be used; however, macrolides should not be viewed as equivalent to tetracyclines for the treatment of Bartonella endocarditis. For pregnant women and young children, the decision to use a prolonged course of doxycycline versus azithromycin must be determined on a case-by-case basis. (See 'Alternative regimens' above and 'Considerations in pregnant persons and young children' above.)

●Surgery – The indications for surgery in patients with Bartonella endocarditis are generally the same as for other causes of infectious endocarditis. However, the timing of surgery may impact the choice and duration of antimicrobial therapy. (See 'Surgery' above.)