|
|
| |||||||||||||||||||||||||||||||||
Background Bartonella (Rochalimaea) quintana is a fastidious gram-negative bacterium known to cause trench fever, cutaneous bacillary angiomatosis, and endocarditis. Between January and June 1993 in Seattle, we isolated B. quintana from 34 blood cultures obtained from 10 patients not known to be infected with the human immunodeficiency virus (HIV).
Methods After identifying the isolates as B. quintana by direct immunofluorescence and DNA-hybridization studies, we determined strain hybridization with studies of restriction-fragment–length polymorphisms (RFLPs) of the intergenic spacer (noncoding) region of ribosomal DNA amplified by the polymerase chain reaction (PCR). To characterize the epidemiologic and clinical features of bartonella infections in these patients, we performed a retrospective case–control study using as controls 20 patients from whom blood was obtained for culture at approximately the same time as from the index patients.
Results B. quintana isolates from the 10 patients were indistinguishable by PCR–RFLP typing. All 10 patients had chronic alcoholism, and 8 were homeless (P = 0.001 for both comparisons with controls). The six patients who underwent HIV testing were seronegative. At the time of their initial presentation, seven patients had temperatures of at least 38.5°C. Six patients had three or more blood cultures that were positive for B. quintana, and in four of these patients B. quintana was isolated from blood cultures obtained 10 or more days apart. Subacute endocarditis developed in two patients and required surgical removal of the infected aortic valve in one of them. Nine patients recovered; one died of sepsis from Streptococcus pneumoniae infection.
Conclusions B. quintana is a cause of fever, bacteremia, and endocarditis in HIV-seronegative, homeless, inner-city patients with chronic alcoholism.
Until recent years, bartonella species have not been isolated from humans in the United States. Several recent studies, all involving patients infected with the human immunodeficiency virus (HIV), have shown that B. quintana can cause cutaneous bacillary angiomatosis, bacteremia, endocarditis, and chronic lymphadenopathy.5,6,7,8,9 Thus far, however, B. quintana has not been reported in HIV-negative people in the United States. B. henselae, a closely related organism that has also recently been reported in HIV-infected patients, causes cutaneous bacillary angiomatosis, peliosis hepatis, and bacteremia.5,10,11,12,13,14,15 In addition, B. henselae has been shown to infect immunocompetent persons and cause cat scratch disease, cutaneous bacillary angiomatosis, bacillary splenitis, bacteremia, and endocarditis.16,17,18,19,20
Between January and June 1993, the microbiology laboratory at the Harborview Medical Center in Seattle isolated organisms presumptively identified as bartonella species in blood cultures from 10 patients.21 Subsequently, we conducted a retrospective case–control study to characterize the epidemiologic and clinical features of bartonella infections in these patients. In addition, we performed studies to identify the species and to determine the strain relatedness of these bartonella isolates.
Methods
Patients
All 10 patients were seen at Harborview Medical Center. Case patients were defined as those with bartonella species isolated from blood cultures at the hospital's microbiology laboratory from January through June 1993. For each case patient, we selected two control patients: one with blood specimens obtained for culture just before those obtained from the case patient and one with specimens obtained just after those from the case patient. Thus, there were 20 controls. Epidemiologic, clinical, and routine laboratory data were obtained from the medical records of the case and control patients.
Isolation of Bartonella in Blood Cultures
Routine blood samples from each patient were inoculated into one Bactec aerobic Plus 26 bottle (Becton Dickinson Diagnostic Instrument Systems, Sparks, Md.) and one Bactec anaerobic NR 7 bottle, as previously described.21 Before the blood cultures were discarded on day 8, a small aliquot of blood-culture broth was removed from the aerobic bottles for staining with acridine orange.21 This procedure had been routinely performed with all blood cultures since May 1988. When bacteria were stained with acridine orange, subcultures were performed as previously described.21 Isolates were identified at the genus level by biochemical testing and analysis of cellular fatty acids.21
Identification of Bartonella Species by Direct Immunofluorescence
The immunofluorescence assays were performed on one blood-culture isolate from each of the 10 patients, as described previously.22 In brief, murine polyclonal antibodies specific for B. henselae or B. quintana were used to distinguish the two species. The fluorescent antibody label was fluorescein-conjugated antimouse IgG (Organon Teknika, West Chester, Pa.).
DNA-Hybridization Studies
DNA relatedness was determined by comparing at least one blood-culture isolate from each of the 10 patients with type strains of all known bartonella species. The methods used to extract and purify DNA and the hydroxyapatite hybridization method for determining DNA relatedness were performed as described previously.23 All studies were performed twice. The change in the thermal-stability midpoint of the heterologous DNA reaction was compared with that of the homologous DNA reaction. Divergence (unpaired bases in a reassociated DNA sequence), which is approximately 1 percent for each degree of decreased thermal stability, was calculated to the nearest 0.5 percent.
Studies of Restriction-Fragment–Length Polymorphisms
Restriction-fragment–length polymorphisms (RFLPs) of B. quintana DNA amplified by the polymerase chain reaction (PCR) were studied in one blood-culture isolate from each of the 10 patients, as previously described,24 except that DdeI and RsaI were used to restrict the PCR-amplified ribosomal intergenic spacer region.
Results
Isolation and Characterization of B. quintana Isolates
B. quintana was isolated from a total of 34 Bactec nonradiometric aerobic resin bottles containing blood samples from the 10 patients.21 All 34 bottles were monitored with the Bactec NR-660 detection system for the first five days of incubation, and none produced sufficient levels of carbon dioxide to indicate growth. For 9 of the 10 patients, acridine-orange staining of the Bactec medium resulted in the first evidence of bacterial growth.21 After the isolates had been subcultured on chocolate agar, visible growth appeared in three to eight days. Representative isolates were negative for catalase, oxidase, and urease. The following major fatty acids, expressed as the percentage of total fatty acids, were detected in isolates from each of the 10 patients: C18:1 (60 to 65 percent), C18:0 (13 to 18 percent), and C16:0 (16 to 19 percent).
The isolates from all 10 patients were identified as B. quintana on the basis of immunofluorescence assays with species-specific polyclonal antibodies and DNA-hybridization studies (Table 1). Repeat isolates from the four patients who had positive cultures of blood obtained 10 or more days apart were also identified as B. quintana (data not shown). The isolates from the 10 patients were indistinguishable from each other by PCR–RFLP typing (Figure 1).
|
|
All cases of bartonella bacteremia occurred from January through June 1993 (Figure 2). The group of 10 case patients and the group of 20 control patients did not differ significantly in terms of age, sex, or history of injection-drug use (Table 2). A univariate analysis showed that the patients with bartonella bacteremia were significantly more likely than the controls to be homeless (8 of 10 vs. 3 of 20, P = 0.001), to have a history of chronic alcohol abuse (10 of 10 vs. 7 of 20, P = 0.001), and to be nonwhite (9 of 10 vs. 7 of 20, P = 0.007). All 10 patients with bartonella bacteremia either resided or spent most of their time in downtown Seattle. Because 8 of these 10 patients were homeless, we could not determine precisely where in downtown Seattle they spent most of their time.
|
|
Clinical Presentation of Patients with B. quintana Bacteremia
At the time of their initial presentation, the 10 patients with B. quintana bacteremia had a median temperature of 38.5°C; 7 of the 10 had a temperature of 38.5°C or higher, and 1 had hypothermia (35.5°C). Three patients reported a history of recent weight loss that exceeded 9.1 kg (20 lb). Other than fever, no consistent symptoms or abnormalities were noted on physical examination. Two patients had splenomegaly. Three patients reported a recent cat scratch, five were presumptively diagnosed as having scabies at the time of their initial presentation, and lice were detected on one patient. The mean white-cell count was 9600 per cubic millimeter.
Six patients had three or more positive blood cultures for B. quintana, and four of these patients had positive blood cultures 10 or more days apart. One patient, who did not comply with antimicrobial therapy, had positive blood cultures on five separate occasions over a period of eight weeks. One patient with B. quintana bacteremia also had a positive blood culture for Streptococcus pneumoniae.
Five patients underwent echocardiographic studies, and the results were normal in four of the five. However, one patient had echocardiographic evidence of an aortic-valve vegetation.25 After 21 days of antimicrobial therapy, surgical removal of the infected aortic valve was required; B. quintana was detected in the valve tissue by PCR techniques.25
Clinical Outcome
Nine of the 10 patients with B. quintana bacteremia survived; 1 patient died, presumably from overwhelming sepsis associated with S. pneumoniae infection. Follow-up of the surviving patients was limited, mainly because they were homeless (Table 3).
|
Discussion
We studied 10 patients in Seattle with bacteremia caused by B. quintana, the agent of trench fever. None were known to be infected with HIV, and six had a documented seronegative status. It is highly unlikely that the B. quintana isolates in these patients represent contamination of blood cultures, since six of the patients had three or more positive blood cultures, some of which were performed 10 or more days apart. One patient had positive blood cultures on five separate occasions over an eight-week period.
Our patients with B. quintana typically had a symptomatic febrile illness resembling that reported by Slater et al. in five patients with B. henselae bacteremia.14 In addition, the B. quintana bacteremia in most of our patients was clinically similar to previous descriptions of trench fever — a disease also caused by B. quintana.2,4 Some of the clinical manifestations of trench fever that have occasionally been observed, such as rash, headache, and bone pain, were not observed in any of our patients.
Because the follow-up of our patients was limited, clear treatment recommendations cannot be given. Although the optimal type and duration of therapy for this and other bartonella infections are not clear, such infections appear to have a tendency to persist or recur,14,19,26,27 especially if antimicrobial therapy is given for less than 14 days. Indeed, the one patient in our study who did not comply with antimicrobial therapy had persistent (or relapsing) bacteremia until he finally completed a course of azithromycin. Careful follow-up to detect relapsing infection is essential.
Although the retrospective nature of our investigation did not permit us to obtain definitive information about how these 10 patients became infected with B. quintana, there are several possibilities. In classic cases of trench fever, B. quintana is transmitted by lice among persons in close contact with one another under poor sanitary conditions, most notably the crowded trenches in World Wars I and II.2,4 The B. quintana infections we describe occurred predominantly among homeless people with chronic alcoholism — conditions that may have resulted in close contact and poor hygiene. Although these conditions may have led to infestation with lice, only one patient was actually observed to have lice at the time of his presentation to the hospital.
Koehler and colleagues have recently shown that the domestic cat serves as a reservoir for B. henselae; in addition, they found B. henselae in fleas from an infected cat.28 Three of our patients reported a recent cat scratch, but none reported flea bites. Five of our patients had scabies infestation. The scabies mite, Sarcoptes scabiei, although it has never been shown to harbor either B. quintana or B. henselae, theoretically could have transmitted B. quintana to these patients. Further study is clearly warranted to determine whether one or more of the above-mentioned vectors transmit B. quintana.
There are several possible explanations for this sudden cluster of 10 cases of B. quintana bacteremia. Increased detection of infection could have occurred as a result of our laboratory's use of acridine-orange staining and rigorous subculturing of presumptively positive isolates. The microbiology laboratory, however, had routinely performed acridine-orange staining on all blood cultures for approximately five years before January 1993 without identifying other cases of bartonella bacteremia. In addition, during the 12-month period after these 10 cases of infection were diagnosed, we isolated B. quintana from only 1 patient, despite use of the same blood-culture techniques and vigilant surveillance. Thus, it seems unlikely that increased laboratory detection of B. quintana could be the sole explanation for the sudden occurrence of these 10 cases.
The finding that the isolates were indistinguishable from one another by PCR–RFLP analysis suggests that these cases are probably epidemiologically linked. Four PCR–RFLP patterns have previously been observed among 18 isolates of B. quintana from the San Francisco area (Koehler JE: personal communication). Our PCR–RFLP studies, however, do not clarify the mode of transmission of B. quintana in our patients. The outbreak could have occurred as a result of close contact among the patients, one of whom served as the initial source of infection. Because most of the patients were homeless, it is possible that they came into contact with each other on the street or at shelters. We do not, however, have epidemiologic evidence to validate this hypothesis. Whether B. quintana is endemic in our region or was imported is also not clear.
In summary, we have identified and characterized a cluster of 10 cases of symptomatic B. quintana bacteremia in Seattle. As compared with the controls, the infected patients were more likely to be homeless, alcoholic, and nonwhite. Because of the small sample size, a stratified or logistic-regression analysis of these characteristics was not possible. Further studies are needed to determine the prevalence, source, mode of transmission, and optimal management of B. quintana infections in populations with demographic and clinical characteristics similar to those of our 10 patients.
We are indebted to the microbiology laboratory at Harborview Medical Center for its diligence in detecting and isolating these fastidious bacteria, to Arnold G. Steigerwalt at the Centers for Disease Control and Prevention for carrying out the DNA-relatedness studies, to the University of Oklahoma Medical Center microbiology staff for assistance with the serologic studies, and to Marcia Goldoft for helping to locate patients for follow-up.
Source Information
From the Division of Infectious Diseases and the Department of Medicine (D.H.S., R.K.R., W.E.S.) and the Department of Laboratory Medicine (M.J.D., A.M.L., M.B.C.), Harborview Medical Center and the University of Washington Medical Center, Seattle; the Department of Medicine, University of Washington Medical Center, Seattle (A.S.K.); the Emerging Bacterial and Mycotic Diseases Branch (D.J.B.) and the Foodborne and Diarrheal Diseases Branch (B.S., G.M.M.), Division of Bacterial and Mycotic Diseases, National Center for Infectious Diseases, Atlanta; and the Department of Pediatrics and the Clinical Microbiology Laboratories, University of Oklahoma Health Sciences Center, Oklahoma City (D.F.W.). Presented at the 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, October 17–20, 1993.
Address reprint requests to Dr. Spach at Madison Clinic, ZA-09, 1001 Broadway, Suite 206, Seattle, WA 98122.
References
| |||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
