Wolfson Medical Center in Holon, Israel, provides cardiothoracic care to children from developing regions, such as Africa, Palestinian Authority, and South America, who are referred for care through the Save a Child’s Heart fund activity (13). Each year, doctors in the Center perform heart surgery on ≈200 children from developing countries. After the unexpected diagnosis of B. quintana endocarditis in this group of children, we reviewed the presurgical cases of infective endocarditis in children referred for heart surgery during 2006–2015.

Before surgery, we conducted a thorough work-up on all patients with endocarditis, including several blood cultures and serologic testing for Bartonella, Legionella, and Brucella spp. and for Q fever (Coxiella burnetii infection). In addition, for most patients, we performed microbiologic studies of specimens obtained during cardiac surgery. We performed and interpreted enzyme immunoassays that have been shown to be 98% specific for detection of B. henselae IgM and IgG (14,15). We also performed serial dilutions to determine final serum titers.

Similar to other serologic assays used for the diagnosis of Bartonella infections (16,17), the enzyme immunoassay used in this study demonstrated high cross-reactivity between B. quintana and B. henselae (M. Giladi, unpub. data). Thus, we presented serologic results as Bartonella sp. IgG and IgM, without species identification. We also performed a genus-specific amplification assay of a 588-bp fragment of the Bartonella riboflavin synthase gene as previously reported (18). For species identification, we sequenced the PCR product and submitted it to a BLAST search (https://blast.ncbi.nlm.nih.gov/Blast.cgi). We considered patients to have endocarditis only after they were determined to fulfill the Duke criteria (19).

From January 1, 2006, through December 31, 2015, a total of 574 children from Africa underwent cardiac surgery, catheterization, or both at Wolfson Medical Center. During this time, 7 (1.2%) of the 574 children were diagnosed with infective endocarditis before surgery. The most frequent cause of endocarditis was B. quintana infection, which was diagnosed in 4/7 children. The 3 other endocarditis cases were caused by undetermined Bartonella spp. infection in an 11-year-old boy from Ethiopia, Streptococcus viridans infection in a 2-year-old boy from Zanzibar, and Q fever in a 7-year-old girl from Kenya. All patients with Bartonella endocarditis were from Ethiopia, where they lived in poor, crowded conditions. The clinical details of the children with Bartonella endocarditis follow.

This case series provides detailed information regarding the clinical presentation, course, and outcome of Bartonella endocarditis caused by B. quintana infection in 4 children and by Bartonella of undetermined species in 1 child. Of interest, all 5 children were natives of Ethiopia. When admitted to our medical center, all of the children were afebrile and had nonspecific symptoms, except for heart failure, which was attributed to their previously known heart disease. In 4 of the 5 patients, endocarditis was not suspected on clinical grounds, but we pursued the diagnosis after echocardiographs revealed vegetations. The diagnosis of B. quintana endocarditis was confirmed in 4 patients (patients A–D) by identification of B. quintana DNA in excised vegetations or endocardial tissue. The diagnosis was further supported by the presence of Bartonella IgG in these 4 patients, 3 of whom had high titers (1:1,600–1:12,800). The fifth patient had Bartonella endocarditis caused by an undetermined species; the diagnosis was based only on serologic test results (IgG titer of 1:6,400) because cardiac tissue was not available for molecular diagnosis. Western blot with cross-absorption studies, a method described by Houpikian and Raoult (22), could have discriminated between B. quintana and B. henselae if it had been applied.

Bartonella spp. are small, gram-negative bacilli whose natural cycle includes a reservoir host, in which Bartonella causes chronic intraerythrocytic bacteremia. In 1993, Bartonella spp. were described as a cause of endocarditis in 2 separate reports and subsequently has become appreciated as a substantial cause of culture-negative endocarditis (3,23,24). Raoult and colleagues have generated several reports on endocarditis caused by Bartonella spp., including several multicenter international studies that involved patients from France, England, and Canada (17,25). Seven Bartonella spp. have been reported to cause infective endocarditis in humans; >95 percent of the cases involved B. quintana or B. henselae (17).

This case series of Bartonella endocarditis in children reveals several common characteristics. All cases occurred in preadolescent and adolescent patients; all patients were afebrile, and the main pathophysiologic dysfunction was congestive heart failure. All patients had markedly elevated ESRs but normal or only mildly elevated CRP levels. Echocardiography revealed large and even giant vegetations in most of the patients, and 2 (40%) of the patients had embolic phenomena. Medical treatment consisted of a prolonged course of doxycycline combined with gentamicin during the initial period, as was recommended for adults with Bartonella endocarditis (21). Even though prolonged administration of doxycycline is relatively contraindicated in children <8 years of age, we suggested a 5-week course for the 7-year-old patient (patient A), as recently recommended (20), because of the extent of her valvular disease.

B. quintana is historically known to cause trench fever, a recurrent febrile disease with acute onset characterized by fever and headache. Trench fever was epidemic among troops during World War I, causing millions of casualties. However, after the introduction of louse control measures, the disease was no longer considered a threat. Recently, however, trench fever has reemerged, causing bacteremia in homeless persons and persons affected with alcoholism in Europe and North America, where it has now been designated urban trench fever (26). B. quintana is mostly associated with human body lice but has also been found in fleas (27,28). The predisposing factors for B. quintana endocarditis are homelessness, alcoholism, and exposure to body lice (29). None of these risk factors for B. quintana infection were known to exist in the patients in this study.

Patients in this study denied having had lice infestation in the past, and we did not identify body lice, pruritus, or excoriations during the initial physical examinations. However, we believe that detailed and accurate histories regarding lice infestation were lacking, particularly because patients with B. quintana endocarditis have probably been infected with B. quintana for months or years before hospital admission for endocarditis. We speculate that residence in a developing country with presumably poor hygiene and low socioeconomic status might have exposed the patients in this study to ectoparasite infestations, including body lice, which could have served as a transmitting vector for B. quintana.

In contrast to our report of afebrile patients with sizable vegetations, previous reports of Bartonella endocarditis have described that fever is usually present (83% of cases) and that valve destruction is characterized by large calcifications but small vegetations (25). Description of B. quintana endocarditis in children is currently confined to a case in a 13-year-old girl from Senegal with underlying rheumatic heart disease, an insidious afebrile clinical course, and prominent vegetations of the left side of the heart (4). Another 2 children with endocarditis and B. quintana–positive serologic test results were included in a series from India, but no clinical or laboratory details were provided (30).

Of the 5 children in our study with Bartonella endocarditis, 4 had involvement of the aortic valve. The predilection for Bartonella spp. to infect the aortic valve has been described (25), but the reason is unknown. Before surgery, 4 of the 5 children fulfilled the Duke criteria for definite infective endocarditis, and the fifth child (patient E) fulfilled criteria for possible endocarditis (Table).

The Duke criteria do not address Bartonella endocarditis specifically, and a definitive diagnosis of Bartonella infection requires positive, high-titer serologic test results; PCR identification of Bartonella sp. DNA in affected tissue or blood; or, on rare occasions, isolation of Bartonella sp. from blood or tissue culture. Recent studies have shown that direct immunofluorescence antibody assays can reliably detect Bartonella IgG, and an IgG titer of >1:800 has a high positive predictive value (95.5%) for Bartonella infection among patients with endocarditis (31,32). However, in 2015, Edouard et al. (32) reported that an IgG titer of <800 does not exclude the diagnosis of Bartonella endocarditis in patients with valvulopathy and that a serologic diagnosis can be confirmed by a positive Western blot result, which they showed exhibited a sensitivity of 100%. Similarly, we showed that high Bartonella IgG titers can be detected by enzyme immunoassay; only 1 patient in our series had IgG titers <1:800. Thus, the enzyme immunoassay has a meaningful role in the diagnosis of Bartonella sp. endocarditis.

Epidemiologic data suggest a north–south gradient distribution in the prevalence of Bartonella endocarditis, from 0% in Sweden to 3% in France and Germany and reaching 15.6% in Algeria and 9.8% in Tunisia (32). Lice are a well-recognized reservoir of B. quintana. Using reverse transcription PCR testing of lice from residents of 9 African countries, Sangaré et al. (33) showed B. quintana DNA was present in 54% of body and 2% of head lice, and they found a clear correlation between the presence of B. quintana in head and body lice and the degree of country poverty, as determined by the gross domestic product. Bartonella spp. were found among 6 (9.2%) of 65 head lice pools and 1 (3.0%) of 33 clothing lice pools from Jimma (34). These data indicate that B. quintana may be quite abundant in East Africa. However, due to the lack of serologic surveys for Bartonella species in this region, its extent is unknown.

In our series, B. quintana was the most frequent causative organism of native valve endocarditis among children from Africa referred to our center for heart surgery. These cases by no means represent the whole spectrum of infective endocarditis in children in Africa, or even Ethiopia, because a selection bias might exist toward cases of nonacute, indolent, infective endocarditis in patients referred for complicated surgeries. However, the predominance of B. quintana infection, even in this specific, small subgroup of patients, is quite impressive and may imply a broader role of this microorganism in infective endocarditis cases in children in Ethiopia or Africa as a whole.

In conclusion, B. quintana is a substantial cause of endocarditis in children in Ethiopia with heart disease. Diagnosis may easily be missed because of the afebrile, insidious nature of this disease and the apparent lack of traditional risk factors for Bartonella infections.