Pericardial Effusion in a Homeless Man Due to Bartonella quintana

Abstract

Bartonella quintana may cause trench fever, endocarditis, bacillary angiomatosis, and chronic bacteremia, and a reemergence among homeless populations in cities has been noted. Pericarditis from Rickettsia conorii and Coxiella burnetii infection has been described, but there have been no reports of pericarditis due to Bartonella spp. We report a case of pericardial effusion due to Bartonella quintana in a homeless man, diagnosed on the basis of PCR detection of Bartonella quintana in a pericardial biopsy sample and a fourfold rise in antibody titers. The patient recovered within 2 weeks with antibiotics active against bartonellae.

Bartonella quintana is a gram-negative bacterium transmitted to humans, its only known reservoir, by the body louse. It has been recently found in cat fleas (18). Trench fever is characterized by a fever, headaches, and leg pain, followed by relapses every 5 days. It was the first clinical manifestation of B. quintana infection to be described during World War I (7, 17). Since that date, additional manifestations of B. quintana infection, including endocarditis (5, 16), bacillary angiomatosis (8), and chronic bacteremia (1), have been described. Recent reports have indicated a reemergence of B. quintana among urban homeless populations in both Europe and the United States (20, 21). The major predisposing factors include poverty, low hygiene, and chronic alcoholism (4, 20). In addition to causing endocarditis, Bartonella species may affect the myocardium, as demonstrated recently in a case of chronic lymphocytic myocarditis caused by Bartonella henselae, the etiologic agent of cat scratch disease (14). A recent review on Rickettsia, Ehrlichia, and Bartonella infections of the myocardium and pericardium (19) reported pericarditis due to Rickettsia conorii (2), an α-proteobacterium closely related to Bartonella species (3), and Coxiella burnetii infection (9). However B. quintana has not been reported, to date, to be an agent of pericarditis. Herein, we report a case of pericardial effusion due to B. quintana in a homeless patient.

Case report.

A 41-year-old homeless male with a history of chronic alcoholism was admitted to the hospital in Marseille because of the onset of sudden and severe rest dyspnea and chest pain improved by anteflexion. Transthoracic echocardiography revealed a large anterior and posterior pericardial effusion and a major aortic insufficiency (regurgitation fraction = 50%). No valvular vegetation was noted. A pericardial drainage including a pericardium biopsy was performed and dramatically improved clinical signs. The standard culture of pericardial fluid as well as blood culture remained sterile. Serological tests for Coxiella burnetii, Bartonella species, Chlamydia species, Legionella pneumophila, Brucella species, Mycoplasma pneumoniae, Borrelia burgdorferi, Toxoplasma gondii, cytomegalovirus, human immunodeficiency virus, hepatitis C virus, and enterovirus were performed. He was discharged from the hospital 5 days after the pericardial drainage in healthy condition, and cardiac surgery for aortic replacement was scheduled 1 month later. He was readmitted 3 weeks later, earlier than expected, with a history of a 1-week fever (39°C) and asthenia. Informed consent was obtained from the patient. His serum C-reactive protein level and erythrocyte sedimentation rate (first hour) were high, 154 mg/liter and 85 mm/h, respectively. Transesophageal echocardiography showed no abnormality indicative of endocarditis but showed the persistence of fluid in anterior and posterior pericardium spaces. Standard blood cultures were negative. Serological tests performed on the first serum were all negative except for tests for B. quintanaand B. henselae (immunoglobulin G [IgG] titer = 200). Those performed on a second serum sample showed a fourfold rise in titers of antibodies against Bartonella (IgG titer = 800). B. quintana was identified by Western immunoblotting following cross absorption (Fig. 1). A treatment with amoxicillin (6 g/day) and gentamicin (3 mg/kg of body weight/day) was initiated. Fever resolved completely within 2 weeks, and the volume of pericardial fluid decreased significantly.

FIG. 1.

Western immunoblotting of our patient before and after cross adsorption with B. quintana or B. henselae. Lanes 1, 5, and 9, B. quintana antigen; lanes 2, 6, and 10, B. henselae antigen; lanes 3, 7, and 11, Bartonella vinsonii subsp. berkhofii antigen; lanes 4, 8, and 12, B. elizabethae antigen. Lanes 1 to 4, untreated serum; lanes 5 to 8, serum adsorbed with B. henselae; lanes 9 to 12, serum adsorbed with B. quintana; lanes M, molecular mass standard.

Histological analysis of the pericardium biopsy sample showed no evidence of inflammation, and immunohistochemistry using a monoclonal antibody against B. quintana (11) was negative. PCR amplification targeting the 16S-23S ribosomal DNA intergenic spacer region gene (its) performed on the fixed pericardial tissue and on EDTA blood collected during the second hospitalization was negative. However, a “suicide” PCR assay (nested PCR targeting a DNA fragment never amplified previously in the laboratory with single-use primers) of both pericardial tissue and blood was positive. The sequence obtained from the PCR product was 100% similar to that of B. quintana in GenBank. The blood culture performed during the second hospitalization was positive on day 17 in an automatic blood culture test (Bactalert; BioMérieux, Marcy l'Etoile, France); the culture was reinoculated onto Columbia blood agar plates, incubated at 37°C in a 5% CO₂ atmosphere, and examined weekly for evidence of growth for 3 months. No growth was observed.

Serological tests for B. quintana and B. henselae were performed by immunofluorescence using an IgG cutoff value of 1:100 as previously described (13). To identify the infecting species, we performed serological cross absorption using B. quintana, B. henselae, or Bartonella elizabethae as the antigens, followed by Western immunoblotting as previously described (6).

For PCR, the formalin-treated pericardium biopsy sample was washed overnight in sterile distilled water, and DNA was extracted with the QIAamp tissue kit as proposed by the manufacturer (Qiagen, Hilden, Germany). The primers URBARTO.1 (5′-CTTCGTTTCTCTTTCTTCAA-3′) and URBARTO.2 (5′-CTTCTCTTCACAATTTCAAT-3′) were used to amplify the 16S-23S ribosomal DNA intergenic spacer, with a hybridization temperature of 48°C, as previously described (18a). A suicide PCR was designed to target the hemin-binding protein E-encoding gene. The external primer pair was hbpEF1 (5′-GAGAGTGCTTCACCTAAATAG-3′) and hbpER1 (5′-CCACCAATCTGTCCTCCAAA-3′), with a hybridization temperature of 55°C, whereas the internal primer pair was hbpEF2 (5′-GAGACGAGTATTAAAGTTTC-3′) and hbpER2 (5′-CTGAGGAACTATTACATCT-3′), with a hybridization temperature of 48°C (15). Sequencing was performed using d-rhodamine. Negative controls (DNA extracted from sterile cardiac valves) were added and remained negative. PCR products were sequenced as previously described (22) with an AB Prism 3100 automated sequencer (Perkin-Elmer Applied Biosciences).

We believe that the patient whose case is described here suffered from pericarditis caused by B. quintana. The causative role was demonstrated by amplification of this bacterium from the pericardial fluid and the blood and by a specific antibody response to B. quintana as observed by cross absorption followed by Western blotting. His age, homelessness, alcoholism, and probable contact with body lice are all epidemiological factors linked with B. quintana infection (5). Endocarditis was first clinically suspected. However, evaluation against the modified Duke's criteria (10) indicated that this patient met no major criteria but did meet three minor criteria (native aortic insufficiency, fever, positive Bartonella serology), but the diagnosis was rejected because the pericardial effusion represented a firm alternate diagnosis. The second argument linking this pericardial disease to B. quintana is the clinical efficacy of the antibiotic therapy including gentamicin within 2 weeks in our patient. Aminoglycosides are currently the only known antibiotic class to be bactericidal for Bartonella species (12). We have previously reported that the administration of aminoglycosides for a minimum of 14 days was the most effective therapy for Bartonella endocarditis (16). Based on echocardiography, we noted a recurrence of pericardial fluid effusion before the initiation of antibiotics and a decrease after. Finally, the diagnosis was established definitively by specific PCR detection on the surgically resected pericardial tissue.

Chronic asymptomatic bacteremias in homeless people have been described (4). We assume that such a bacteremia can be responsible for a secondary location such as the endocardium or pericardium. In our patient, a history of previous pericardial damage to explain this location was not known. The mechanism of infection in this case remains unclear; pericardial effusion could have been due to direct infection during the bacteremia or could have been secondary, due to perivascularitis, or it could have been immunologically mediated by cross-reaction of the inflammatory response against myocardial or pericardial tissues, as has been described for myocarditis due to B. henselae (14). The fact that the suicide PCR was positive on the surgically resected pericardial tissue is in favor of a direct infection.

B. quintana can be added to the list of rare bacterial agents responsible for pericardial effusion, and this is critical since B. quintana may be one of the rare causes of treatable pericardial effusion with antibiotics. In our experience of 204 pericardial effusions (unpublished data), only one case was due to B. quintana infection compared to 12 due to Q fever infections. Detecting Bartonella bacteria will then help to diminish the number of pericardial effusions classified as idiopathic, and we propose that the bacterium should be systematically searched for in cases of pericardial effusion, especially in patients at risk for louse-transmitted diseases.