Epidemiologic and clinical characteristics of Bartonella quintana and Bartonella henselae endocarditis: a study of 48 patients.

Abstract

Introduction Bartonella species have been the subject of numerous studies over the past few years, in part because of their role as etiologic agents in bacillary angiomatosis in patients infected with the human immunodeficiency virus. Of the 7 recognized species implicated in human disease, only Bartonella quintana and Bartonella henselae have been associated frequently with an increasing spectrum of clinical syndromes. These include cat-scratch disease, chronic bacteremia, chronic lymphadenopathy, meningoencephalitis, myelitis, granulomatous hepatitis, endocarditis, osteomyelitis, bacillary angiomatosis, and bacillary peliosis hepatis (24,32). Four Bartonella species have been recognized as causative agents of blood culture-negative endocarditis in humans (17,47). Since the first definite case of Bartonella endocarditis was published (47), 46 cases have been reported in the international literature, mostly due to B. quintana or B. henselae, but in 1 case each due to B. elizabethae and B. vinsonii subsp berkhoffii(39), including 30 reported by our team (1,2,4,6,8,10,11,15,16,19–22,30,36,37,41,46,46–48). A suspected 47th case was described by Spach et al (49), but no microbiologic or serologic evidence of the causative role of a Bartonella species could be obtained at the time of diagnosis. Although risk factors for Bartonella-induced bacillary angiomatosis have been established (27), such features have not been determined for Bartonella endocarditis, due to the small number of reported cases. As a reference center for the diagnosis and study of rickettsioses, we have collected data on bloodculture negative endocarditis over the past 5 years in order to estimate the prevalence of Coxiella burnetii and Bartonella endocarditis. In view of the serologic cross-reactions between Chlamydia and Bartonella species (11), we also have cultivated and/or PCR tested frozen excised valves from patients originally diagnosed as having Chlamydia endocarditis on the basis of positive serology. This has enabled us to report 30 cases of Bartonella endocarditis (6,10,11,30,39,41), eleven of which had been initially misdiagnosed as Chlamydia endocarditis (14,31). Direct evidence of B. henselae or B. quintana infection was obtained in 5 and 8 patients respectively, by culture and/or PCR. This report compares the speciesspecific clinical, epidemiologic, and serologic features of 48 confirmed cases of B. quintana and B. henselae endocarditis (35 of which have not been previously reported) with 192 control patients with endocarditis caused by other pathogens matched by age, sex, and geographic origin. Patients and Methods Study design Case definition: Patients were considered to have definite endocarditis according to the modified Duke criteria for the diagnosis of infective endocarditis (IE) (13,29). Bartonella infection was diagnosed by culture of a Bartonella species or positive Bartonella PCR from valvular tissue, vegetation, or blood. Study sites: Study sites included the following: Unité des rickettsies, Faculté de Médecine, Marseille, France; Laboratoire de Microbiologie, Hôpital Edouard Herriot, Lyon, France; Department of Infectious Diseases, St Thomas’ Hospital, London, United Kingdom; Service de Microbiologie Clinique, Hôpital Broussais, Paris, France; Department of Medicine and Microbiology, Dalhousie University, Halifax, Nova Scotia, Canada; Clinique de Réanimation des Maladies Infectieuses and Laboratoire de Microbiologie, Hôpital Bichat, Paris, France; Laboratoire de Microbiologie, Hôpital de la Croix-Rousse, Lyon, France; Public Health Laboratory, William Harvey Hospital, Willesborough, Ashford, United Kingdom; Laboratoire de Bactériologie, Hôpital de Rangueil, Toulouse, France; Microbiology Department, St Bartholomew’s Hospital, London, United Kingdom; Service de Bactériologie, Hôpital Henri Mondor, Créteil, France. Patients: From January 1995 through December 2000, the Unité des Rickettsies received serum from 45,000 patients for detection of Rickettsia, Coxiella, and Bartonella antibodies. Five hundred and fifty-one of these patients had endocarditis, including 204 patients with blood culture-negative endocarditis. Heparinized blood and/or heart valve tissue from some patients were also sent to our laboratory for PCR-amplification and culture of valve and blood specimens. A diagnosis of B. henselae or B. quintana endocarditis was made in 48 patients based upon microbiologic and/or PCR findings, including 35 new patients and 13 previously reported (39). For each of the 38 patients with B. quintana and 10 patients with B. henselae endocarditis, the responsible physicians identified 4 control patients with definite endocarditis caused by other pathogens (13) matched by age (± 5 years), sex, and year and place of diagnosis in order to evaluate the clinical and epidemiologic characteristics of B. quintana and B. henselae infection in patients with endocarditis. Questionnaire: A standardized questionnaire was completed for Bartonella endocarditis patients and controls. Questions asked included those used in the diagnostic score of the Duke Endocarditis Service (13,29), previous antibiotic therapy, valvular surgery, the outcome within 3 months following the diagnosis of culture-negative endocarditis, and presence of environmental exposure factors such as homelessness, chronic alcoholism (>100 g alcohol per day for more than 1 year) and immunodeficiency, which are epidemiologic features associated with other Bartonella-induced diseases. Other questions, such as presence of body lice and contact with cats and cat fleas, were added to the questionnaire in 1997 following our preliminary observations that these may be important risk factors for Bartonella infection. If such data were not available in the medical records, the patients were contacted by telephone. Data were analyzed at the Unité des Rickettsies, Marseille, France. Statistical analysis Patients were compared to controls with stratification according to the infecting species. The Mantel-Haenszel method was used to compare proportions, and the t-test for matched samples was used to compare means using Epi Info software (version 5, Centers for Disease Control, Atlanta, GA, 1990) (9). In order to identify variables independently associated with case status, a multivariate logistic regression analysis was conducted using the GB-Stat software (version 6.5, Dynamics Microsystem, Inc., Silver Spring, MD). Microbiologic and molecular methods Blood and valve cultures: To isolate Bartonella species from whole blood samples and valve tissue or vegetation, specimens were inoculated onto both 5% blood agar (bioMerieux, Marcyl’Etoile, France) and into tissue cultures of the human cell line ECV 304 (26). Inoculated media were incubated at 37 °C in a 5% CO 2 atmosphere for up to 60 days. In order to isolate other pathogens, samples were also inoculated onto chocolate agar (bioMerieux) and incubated at 37 °C in a 5% CO 2 atmosphere for 10 days. Isolates obtained from cell culture were identified using species-specific mouse polyvalent antisera (34) or PCR (23,42). Molecular amplification and identification of isolates: The DNA used as a template in PCR amplifications was prepared from agar-grown single colonies of the isolate or from isolates obtained in endothelial cells. These isolates were digested with proteinase K and sodium dodecyl sulfate and their DNA extracted using phenolchloroform and ethanol precipitation. Crude DNA extracts were prepared from serum and whole blood samples using the QI Amp-blood kit (Qiagen, Inc., Chatsworth, CA) as described previously (50). The presence of the 16S-23S rDNA intergenic spacer region (its) (43) was determined by PCR. Sterile water, processed as above, was used as a negative control in all PCR reactions. The identity of the Bartonella from which PCR products were obtained was determined by base-sequence determination (42). Results From January 1995 through December 2000, direct evidence of Bartonella endocarditis was obtained in 48 patients: 38 patients with B. quintana and 10 with B. henselae infection. Microbiologic and molecular results Culture-DNA amplification: Culture material was available for 39 patients, 25 of whom had received antibiotics before collection of blood or valve samples. B. quintana was isolated from 7 and B. henselae from 2 of the 31 available blood culture specimens, 10 of which were also positive by its amplification. B. quintana was grown from valve biopsies in 17 patients and B. henselae in 2. Bartonella its amplicons were obtained from 44 of the 45 available valves, 28 of which were sampled following antibiotic therapy. DNA-amplification from valve specimens was the only direct evidence of Bartonella infections in 24 patients. Sequence analysis of the PCR products obtained from the isolated Bartonella, the valve biopsies, and blood specimen enabled the identification of B. quintana in 38 cases (79%) and B. henselae in 10 (21%). The etiologic agent was identified in 182 of the 192 control patients and included the following:Staphylococcus aureus (22 patients), coagulase-negative staphylococci (12 patients), Streptococcus viridans (77 patients), S. pneumoniae (3 patients), group B Streptococcus (1 patient), group G Streptococcus (1 patient), S. bovis (19 patients), Enterococcus faecalis (22 patients), Abiotrophia defectiva (1 patient), Gemella morbillorum (1 patient), HACEK group bacteria (13 patients), Escherichia coli (3 patients), Neisseria sicca (1 patient), Propionibacterium acnes (1 patient), and C. burnetii (5 patients). In the remaining 10 patients, all with negative Bartonella serology, no etiologic agent could be identified. Patient characteristics The results of the case-control study are summarized in Table 1.TABLE 1: Comparison of characteristics of 48 patients with Bartonella quintana or B. henselae endocarditis and 192 matched controls, with stratification according to the infecting speciesPatients with B. quintana endocarditis: The mean age of the 38 patients with proven B. quintana endocarditis (± standard deviation [SD]) was 44.9 ± 12.4 years (range, 16–74 yr). Thirty-two patients (84.2%) were male. Epidemiologic information was partially missing for 2 patients. Twenty of 38 patients (57.8%) were homeless, 23 of 38 (60.5%) were alcoholic, 13 of 36 (36.1%) had contact with body lice, but none reported cat bites or scratches or contact with cat fleas. One patient (2.6%) was immunocompromised. Clinical information was available for all patients; 15 (39.5%) had known valvular heart disease, 34 patients (89.5%) presented with fever (>38 °C), and 15 (39.5%) suffered from embolic phenomena. Two patients first presented with acute renal insufficiency and had mesangioproliferative glomerulonephritis on renal biopsy. On echocardiography, 37 patients (97.3%) had findings consistent with endocarditis; vegetations were present in 36, and a valvular leak in 1. The aortic valve was involved in 25 patients (65.8%), the mitral valve in 7 (18.4%), both the aortic and mitral valves in 5 (13.1%), and the pulmonary valve in 1 (2.6%). Twenty-seven patients received a combination of a beta-lactam (amoxicillin in 18, ceftriaxone in 5, penicillin G in 3, and oxacillin in 1) and an aminoglycoside (gentamicin in 25, netilmicin in 1, and amikacin in 1), including 5 who received 1 additional antibiotic (doxycycline in 1, ofloxacin in 2, and vancomycin in 2). One patient was treated with vancomycin, netilmicin, and ofloxacin; 1 with rifampin plus pefloxacin; 1 with teicoplanin plus netilmicin; 2 with vancomycin plus gentamicin; 3 with gentamicin plus doxycycline; and 1 with amikacin plus rifampin. Overall, the mean duration of antibiotic therapy (± SD) was 42.2 ± 14.3 days. Thirty-seven patients (97.4%) underwent valvular surgery because of severe valvular damage at the time of diagnosis. Thirty-four patients (89.5%), 19 of whom were homeless, recovered from B. quintana endocarditis; 1 patient relapsed despite surgery and recovered following treatment of doxycycline for 6 weeks and gentamicin for the first 2 weeks. Three patients (7.9%), all of whom were homeless, died despite antibiotic therapy and valvular surgery, 2 from acute heart failure and 1 from multiple organ failure. All 38 patients were classified as definite IE according to the Duke criteria (29). Patients with B. henselae endocarditis: The mean age of the 10 patients with proven B. henselae endocarditis (± SD) was 46.6 ± 21.9 years (range, 22–81 yr), not statistically different from that of the patients infected with B. quintana (p = 0.6). Six patients (60.0%) were male. Epidemiologic information was not available for every patient. Seven of 9 patients (77.8%) reported cat bites or scratches, and 5 of 9 (55.5%) reported contact with cat fleas, but none was immunocompromised, homeless, alcoholic, or had contact with body lice. Nine patients (90%) had known valvular heart disease; 7 (70%) presented with fever (>38 °C) and 2 (20%) suffered from embolic phenomena. On echocardiography, all 10 patients had findings consistent with endocarditis; vegetations were present in 9 and a valvular leak in 1. The aortic valve was involved in 6 patients (60%), the mitral valve in 2 (20%), both the aortic and mitral valves in 1 (10%), and the tricuspid valve in 1 (10%). Six patients received a combination of a betalactam (amoxicillin in 2, ceftriaxone in 2, and penicillin G in 2) and an aminoglycoside (gentamicin in 5, and netilmicin in 1), including 1 patient who also received doxycycline. Two patients were treated with doxycycline, 1 with amoxicillin, and 1 with ceftriaxone plus doxycycline. Overall, the mean duration of antibiotic therapy (± SD) was 47.3 ± 18.6 days. Nine patients (90%) underwent valvular surgery because of severe valvular damage at the time of diagnosis; 8 of 10 patients (80%) recovered from B. henselae endocarditis. One patient relapsed despite surgery and a 6-week course of doxycycline. She then underwent a second valvular replacement and recovered following a treatment of gentamicin for 2 weeks and doxycycline for 6 weeks. One patient (10.0%) died, despite antibiotic therapy, from acute heart failure. All 10 patients were classified as definite IE according to the Duke criteria (29). Discussion In a previous study (39), we have shown that Bartonella endocarditis represents 3% of all cases of endocarditis. However, although species-specific risk factors and clinical symptoms have been described for other B. quintana or B. henselae-related infections, such features have not been adequately evaluated for Bartonella endocarditis. Our study allowed us to describe different risk factors according to the infecting species, B. quintana or B. henselae. As they were matched by age, the mean age of the patients and controls was not statistically different. However, the mean age of the 48 patients with direct evidence of Bartonella endocarditis (45 years) was younger than that usually observed in patients with endocarditis from other causes (50–70 years) (3), but similar to that observed in blood culture-negative endocarditis series (3,17,18,45). The male:female sex ratio in B. henselae-infected patients was similar to that in endocarditis caused by other pathogens (3). Patients with B. henselae endocarditis were as likely as the control subjects to have preexisting valvular disease, a common risk factor for infective endocarditis. B. henselae endocarditis was associated with cat scratches or bites and contact with cat fleas. However, as demonstrated by the multivariate logistic regression analysis, contact with cats was significantly and independently associated with B. henselae infection, whereas contact with cat fleas was significantly but not independently associated with B. henselae endocarditis. Such risk factors have also been described in 3 of the 4 cases of B. henselae endocarditis reported by others (1,2,4,15,16,19–21,36,37,46). Animal reservoirs have been demonstrated for B. henselae. Ownership of a cat has been identified as a predisposing factor in cat-scratch disease and bacillary angiomatosis (51,53). Moreover, cat fleas are potential vectors of B. henselae(7,25,40,52). Clinically, patients with B. henselae endocarditis and controls presented with similar symptoms, but B. henselae endocarditis required valvular surgery more often. Our data demonstrate that patients with B. henselae endocarditis are not different from patients with other causes of endocarditis except for their younger age and their contact with cats and/or cat fleas. The male:female sex ratio was higher in B. quintana-infected patients than in patients with endocarditis caused by other pathogens (3). Although this may be explained by the high rate of homeless patients, mostly men, among patients with B. quintana endocarditis, it appears to be an important epidemiologic characteristic of this disease. Patients with B. quintana endocarditis had a significantly lower rate of previously known valvular disease than their matched controls. Spach et al (49) described B. quintana bacteremia in 10 homeless patients, 2 of whom developed endocarditis. Persistent bacteremia may result in subacute endocarditis, even in the absence of previously known valve disease. However, since many of the 38 patients with B. quintana endocarditis were homeless, the lack of a history of predisposing valvular heart disease may reflect a lack of previous medical care. In this study, there was no association between immunodeficiency and Bartonella endocarditis despite the fact that the first reported patient with B. quintana endocarditis was human immunodeficiency virus (HIV)-positive (27). The multivariate logistic regression analysis demonstrated that homelessness, alcoholism, and contact with body lice were significantly and independently associated with B. quintana infection. Such risk factors have also been described in some of the 13 cases of B. quintana endocarditis reported by others (1,2,4,15,16,19–21,36,37,46,48). B. quintana bacteremia and endocarditis in homeless people (49) have been associated with body lice. Currently, the role of body lice in most B. quintana infections should be suspected. B. quintana has been found in body lice from various parts of the world, including France, Burundi, Zimbabwe, Peru, and Russia (44). However, no study has been conducted in the United Kingdom and in the United States to confirm the relationship between B. quintana infections and body lice, despite the occurrence of clinical cases. In our study, the epidemiologic findings in patients with B. quintana endocarditis were similar to those reported by Koehler et al (27) in patients with bacillary angiomatosis, but we describe a significant and independent association with chronic alcoholism and homelessness. Clinically, patients with B. quintana endocarditis and controls presented with similar symptoms, but valvular surgery was more often necessary in patients infected with B. quintana. Based on our data, a typical patient with B. quintana endocarditis is a middle-aged man, homeless and/or a chronic alcoholic, with no previously known valvular disease, in contact with body lice. In 8 of the 48 patients with Bartonella endocarditis, no risk factor could be found. Moreover, 3 patients with B. quintana endocarditis had contact with cats and/or cat fleas but were not homeless or alcoholic and did not have body lice. We previously reported (12,38) 2 patients with chronic lymphadenitis due to B. quintana who had had contact with cat fleas but not lice. There must, therefore, be as yet undescribed risk factors for B. quintana infections. The isolation of Bartonella is difficult, especially from immunocompetent patients, but it has been improved with a combination of prolonged culture on blood agar (5) and the use of cell-culture systems (11,39). Since Bartonella species are susceptible to most antibiotics, even a few doses may prevent isolation (28,33). Twenty-five of the 39 patients for whom culture material was available had received antibiotics before blood cultures and valvular surgery, which may explain why culture was positive in only 64% of patients. PCR was particularly useful in diagnosis, and although positive in only 10 patients on blood, DNA amplification has been shown to be valuable when performed on fresh valve tissues (21,36). In our study, a PCR amplicon was obtained in 97.8% of the available specimens despite the fact that 62.2% were sampled following antibiotic therapy. PCR is also useful for amplifying Bartonella DNA from paraffin-embedded tissues (35), thus allowing for retrospective diagnosis. This study has established that B. henselae and B. quintana are the main Bartonella species responsible for endocarditis in patients younger than those with endocarditis from other etiologies, and cause severe but distinct diseases with different risk factors. Apart from its association with cat scratches and bites and contact with cat fleas, which are specific exposures, B. henselae endocarditis occurs, as do most cases of infective endocarditis, in patients with preexisting damage to heart valves, whereas B. quintana endocarditis mainly occurs in homeless and/or chronic alcoholic men with body lice but no known preexisting heart valve abnormality. Summary Among the 4 Bartonella species currently recognized as agents of endocarditis, Bartonella quintana and Bartonella henselae are the most commonly encountered. However, reported cases of Bartonella endocarditis are mostly sporadic, and the species-specific characteristics of this infection have not been adequately evaluated. In order to characterize the species-specific epidemiologic and clinical features of Bartonella endocarditis, we conducted a case-control study of patients with culture-or PCR-confirmed B. quintana or B. henselae endocarditis from January 1995 through December 2000 and controls who suffered endocarditis caused by other pathogens. Epidemiologic and clinical data were obtained by means of a standardized questionnaire. A total of 48 patients were diagnosed as having Bartonella endocarditis on the basis of isolation or PCR-amplification of Bartonella species from blood or valvular biopsies. Of these, 38 patients with B. quintana endocarditis (79%) and 10 with B. henselae endocarditis (21%) were compared with 152 and 40 control patients, respectively, matched by sex, age, and geographic origin, selected among patients with infective endocarditis of other etiology. Patients with B. henselae endocarditis frequently had a previous valvulopathy (90%) and were epidemiologically linked with cat bites or scratches (p < 0.001; odds ratio [OR] = 44.3) and cat flea exposure (p = 0.001), whereas those infected with B. quintana were mostly male (84.2%), less likely than controls to have known valvular disease (p < 0.001; OR = 0.2) but more likely to be homeless (p < 0.001; OR = 55.2), alcoholic (p < 0.001; OR = 19.6), and have body lice (p < 0.001; OR = 78.5). Valvular surgery was needed in 95.8% of patients. In conclusion, we report here the largest series of Bartonella endocarditis to date, to our knowledge. B. quintana should be highly suspected in homeless and alcoholic patients with no previous cardiac history, whereas B. henselae should be suspected in patients with known valvular heart disease and contact with cats and/or cat fleas. Acknowledgment: The authors thank Drs. X. Belenfant, P. Jourdain, M. Archambaud, O. Launay, and P. Collet for providing clinical material, Dr. Patrizia Carrieri for helping in the statistical analysis, and Dr. P. J. Kelly and Dr. L. Bell-Sakyi for reviewing the manuscript.