|
|
| |||||||||||||||||||||||||||||||||
Background Bacillary angiomatosis and bacillary peliosis are vascular proliferative manifestations of infection with species of the genus bartonella that occur predominantly in patients infected with the human immunodeficiency virus. Two species, Bartonella henselae and B. quintana, have been associated with bacillary angiomatosis, but culture and speciation are difficult, and there has been little systematic evaluation of the species-specific disease characteristics. We studied 49 patients seen over eight years who were infected with bartonella species identified by molecular techniques and who had clinical lesions consistent with bacillary angiomatosis–peliosis.
Methods In this case–control study, a standardized questionnaire about exposures was administered to patients with bacillary angiomatosis–peliosis and to 96 matched controls. The infecting bartonella species were determined by molecular techniques.
Results Of the 49 patients with bacillary angiomatosis–peliosis, 26 (53 percent) were infected with B. henselae and 23 (47 percent) with B. quintana. Subcutaneous and lytic bone lesions were strongly associated with B. quintana, whereas peliosis hepatis was associated exclusively with B. henselae. Patients with B. henselae infection were identified throughout the study period and were epidemiologically linked to cat and flea exposure (P<0.004), whereas those with B. quintana were clustered and were characterized by low income (P = 0.003), homelessness (P = 0.004), and exposure to lice (P = 0.03). Prior treatment with macrolide antibiotics appeared to be protective against infection with either species.
Conclusions B. henselae and B. quintana, the organisms that cause bacillary angiomatosis–peliosis, are associated with different epidemiologic risk factors and with predilections for involvement of different organs.
The study of bacillary angiomatosis–peliosis in HIV-infected patients has led to the discovery of bartonella infections in the United States,5,6,8 the naming of a new species, Bartonella henselae,9,10 the cultivation of two bartonella species from cutaneous bacillary angiomatosis lesions,11 documentation that B. henselae causes cat scratch disease,12,13 identification of the domestic cat as the principal reservoir of B. henselae from which humans can acquire infection,14 and identification of the cat flea as the vector of B. henselae.15
An initial case–control study to identify risk factors for bacillary angiomatosis–peliosis found a statistically significant association between cat bites or scratches and disease.16 However, the pathogenic species were not identified, and 30 percent of the patients did not have exposure to cats, leading us to hypothesize that the disease might be caused by more than one species, each with different reservoirs and vectors. We subsequently found that, in addition to B. henselae, B. quintana can also cause bacillary angiomatosis in HIV-infected patients,11 yet the risk factors, reservoirs, and vectors for B. quintana infection in immunocompromised patients with bacillary angiomatosis–peliosis remain unidentified.
In most studies of bacillary angiomatosis–peliosis, the infecting bartonella species has not been identified. To identify the microbiologic characteristics and risk factors associated with infection with particular bartonella species, we systematically evaluated 49 patients over eight years who had clinical lesions of bacillary angiomatosis–peliosis and who were infected with bartonella species identified by molecular techniques. The goal was to improve the recognition, diagnosis, and prevention of infection with this emerging human pathogen.
Methods
Study Design
Case Definition
A suspected case was defined as the presence of clinical symptoms and signs of bacillary angiomatosis, parenchymal bacillary peliosis, or both in a patient.16 A case was confirmed when an infecting bartonella species was identified by molecular techniques.5,11
Eligible Study Participants
Persons with suspected bacillary angiomatosis–peliosis were identified by their health care providers between July 1987 and January 1995. Recovery of bartonella from cutaneous tissue was not accomplished until 199111; therefore, for patients interviewed between July 1987 and October 1991, only those with archived tissue specimens containing well-visualized bacillary organisms16 for molecular speciation were eligible for enrollment. Patients with suspected bacillary angiomatosis–peliosis who had biopsies or cultures undertaken between May 1991 and January 1995 were also eligible.
Health care providers identified two controls per patient, matched according to HIV serologic status and hospital (for inpatients) or clinic (for outpatients) with inpatient-ward and outpatient-schedule registries, respectively.16 Written consent was obtained from all participants, as approved by the Committee for Human Research of the University of California, San Francisco.
Questionnaire
A standardized questionnaire was administered; because of the increasing body of knowledge about bartonella over the eight-year study period, additional standardized questions were added pertaining to arthropod exposures and history of diarrheal illness (in 1991) and alcohol consumption and bathing habits (in 1993). Participants were asked about demographic and behavioral characteristics, medical history, and environmental exposures during specific periods preceding the onset of the illness (patients) or enrollment (controls). Additional clinical information was extracted from medical records with a structured data-collection instrument.
Statistical Analysis
Patients and matched controls were evaluated for differences. Matched analyses were repeated with stratification according to infecting bartonella species. Several subanalyses were conducted on unmatched data to maintain statistical power with diminished matched sets (for questions added in 1991 and 1993).
The Mann–Whitney U test was used to compare distributions of continuous variables between cases and controls. For dichotomous variables, univariate matched odds ratios with 95 percent confidence intervals were calculated by the Mantel–Haenszel method with Epi Info.17 For univariate unmatched odds ratios, 95 percent confidence intervals were calculated by Yates' corrected method or a two-tailed Fisher's exact test (for analyses with small numbers of observations).
Molecular and Microbiologic Methods
Isolation of Bartonella Species from Humans and Pet Cats
Blood, tissue, or both were obtained for culture of bartonella species. Tissue was homogenized in inoculation medium11 and plated onto both chocolate agar and heart infusion agar supplemented with 5 percent fresh defibrinated rabbit blood. Biopsy tissue from seven patients also was cocultivated with an endothelial-cell monolayer.11 Human and cat blood was cultured in adult8,11 and pediatric14 lysis-centrifugation tubes (Wampole, Cranbury, N.J.), respectively. Control medium, agar plates, and endothelial-cell monolayers were processed in parallel.
Extraction of DNA from Archived Tissue and Speciation of Bartonella Isolates by Sequencing
The infecting bartonella species was determined by extraction of DNA from bartonella isolates (25 patients), frozen biopsy tissue (9 patients), or formalin-fixed, paraffin-embedded tissue (15 patients).11,14 DNA was amplified under standard conditions by the polymerase chain reaction (PCR) with alpha-proteobacterium range-restricted 16S ribosomal RNA gene primers p24E and p12B.5 Control tissues matched by type and simultaneously processed were included in each group. Negative and positive PCR controls (containing no DNA and B. henselae DNA, respectively) were included in each experiment.
The amplified fragment was cloned and sequenced.11 Both strands of a minimum of two different clones were sequenced for each patient; if a sequence differed from the previously published bartonella sequence, additional clones were sequenced to resolve the difference. Speciation of each isolate was corroborated by PCR analysis of restriction-fragment–length polymorphisms.9
Molecular Subtyping of B. henselae Isolates
After a subspeciation method for B. henselae had been developed,18 we studied the 13 cats belonging to the last three consecutively enrolled patients with confirmed B. henselae infection who had been exposed to cats (June to December 1994) to determine the relatedness between the human B. henselae isolate and the corresponding cat isolate or isolates. Repetitive-element PCR with oligonucleotide primers based on highly conserved repetitive extragenic palindromic (REP) and enterobacterial repetitive intergenic consensus (ERIC) sequences was performed with whole-cell preparations.18 DNA fingerprints of isolates were compared for similarity by visual inspection of band patterns, and the findings were confirmed by computer-assisted analysis (RFLPscan, Scanalytics, CSP, Billerica, Mass.).
Results
Case–Control Study
Forty-nine patients infected with a bartonella species were enrolled: 33 at the onset of illness and 16 previously reported patients who had histologically confirmed bacillary angiomatosis–peliosis infection without identification of a bartonella species.16 Forty-six patients (94 percent) had histologic evidence of bacillary angiomatosis–peliosis; two (4 percent) had histologic features (one in the spleen and one in a lymph node) of an acute, necrotizing, inflammatory process without angiomatosis; and one (2 percent) refused liver biopsy but had B. henselae bacteremia and clinical features of peliosis hepatis (a temperature of more than 39°C, hepatomegaly, and an elevated alkaline phosphatase level). Forty-five patients (92 percent) had HIV infection; four (8 percent) were HIV-negative (of whom three were immunocompetent19 and one of whom had received immunosuppressive drugs5).
Forty-five of the 49 patients (92 percent) and 89 of their 96 matched controls (93 percent) were residents of the San Francisco Bay area. Two controls were enrolled for each patient, except that one control each was enrolled for two of the three HIV-negative, immunocompetent patients. Relatives or friends answered questions about five patients who had died and one who was a child under 12 years old.
Patient Characteristics
Matched Analyses
The patients were similar to the matched controls in median age, sex, race, ethnic group, and proportion with a preceding diagnosis of AIDS20 (all P>0.22). The patients were less likely than the controls to be homosexual or bisexual (matched odds ratio, 0.2; 95 percent confidence interval, 0.1 to 0.5; P = 0.002). Clinical comparisons between the patients with B. quintana infection and those with B. henselae infection suggest that B. quintana infections have a predilection for causing subcutaneous and deep soft-tissue disease and lytic bone lesions (Table 1). In contrast, B. henselae infections were strongly associated with lymph-node disease and parenchymal peliosis of the liver or liver and spleen.
|
Unmatched Analyses
Patients with B. henselae infection (odds ratio, 6.5; 95 percent confidence interval, 1.1 to 51.1; P = 0.04) but not B. quintana infection (odds ratio, 1.7; 95 percent confidence interval, 0.2 to 13.0; P = 0.68) were more likely than controls to have had diarrhea.
Environmental Exposures
Matched Analyses
Table 2 shows environmental exposures and clinical characteristics associated with bacillary angiomatosis–peliosis, with stratification according to the infecting bartonella species. Patients with B. henselae infection but not those with B. quintana infection were more likely than matched controls to have owned a cat, been bitten or scratched by a cat, owned a cat with fleas, or been bitten by cat fleas (all P<0.004). Patients with B. quintana infection were 8.5 times as likely as matched controls to be homeless and 9.8 times as likely to have a low annual income. Neither B. henselae nor B. quintana infection was associated with alcohol consumption, intravenous drug use, or a prior AIDS-associated opportunistic infection or cancer.
|
Unmatched Analyses
Because only a subgroup of our 49 patients and 96 controls were queried about exposure to arthropods, use of alcohol, and bathing habits, these variables were also evaluated in an unmatched analysis (Table 2). The unmatched results were similar to the matched results for all variables. In the unmatched analysis, patients with B. quintana infection were significantly more likely than controls to have had recent infestations with head or body lice (21 percent of patients and 0 percent of controls, P = 0.03), although the results of a matched analysis did not achieve significance (P = 0.11), possibly because of the small number of patients and controls queried. B. henselae infection was not associated with exposure to lice in either matched or unmatched analyses.
Isolation and Speciation of Bartonella from Tissue
Twenty-six patients (53 percent) had B. henselae infection and 23 (47 percent) had B. quintana infection documented from either cultured tissues (25 patients) or archived tissues (24 patients). Figure 1 shows 44 patients residing in the San Francisco Bay area according to bartonella species and date of onset of illness (1 patient with onset of illness due to B. quintana in February 1987 is not shown). Disease due to B. henselae occurred at regular intervals over a period of 75 months; however, 15 of 22 cases of B. quintana infection were clustered in the 17-month period from February 1992 to June 1993.
|
Thirty-six bartonella isolates were obtained from the blood, tissue, or both of 25 patients: B. henselae was obtained from 9 and B. quintana from 16 patients. Of 20 patients with focal bacillary angiomatosis–peliosis and simultaneous culture of blood and tissue, 10 (50 percent) had bacteremia, including 6 of 14 (43 percent) with B. quintana infection and 4 of 6 (67 percent) with B. henselae infection. No bartonella species other than B. quintana or B. henselae was isolated, and multiple isolates from the same patient were always the same species.
B. henselae was isolated from the tissue of two patients taking prophylactic oral trimethoprim–sulfamethoxazole. B. quintana was isolated from one patient who had received nafcillin and gentamicin intravenously for several days before biopsy and from another patient pretreated with a first-generation cephalosporin. No bartonella isolate was obtained from any patient who was treated with a macrolide or tetracycline.
Microbiologic Characteristics Distinguishing Bartonella Species
Phenotypic differences were readily apparent between primary isolates of B. henselae and B. quintana from humans. Colonies of B. henselae were usually rough and deeply embedded in agar (Figure 2A); the primary isolate was rarely smooth. In contrast, B. quintana primary isolates were always smooth and flat and never pitted the agar (Figure 2B). All isolates could be passaged on agar indefinitely, although the rough B. henselae isolates became smooth with continued passage. There was also preferential growth of B. henselae on heart infusion agar with 5 percent rabbit blood and of B. quintana on chocolate agar. B. henselae isolates from cat blood were often detected on agar at three days, but neither B. henselae nor B. quintana isolates from human blood or tissue could be detected before eight days.
|
Isolation of Bartonella Species from Pet Cats
Ten patients still lived with the same 25 cats to which they had been exposed before bacillary angiomatosis–peliosis developed. Nine of these 10 patients were infected with B. henselae, and 1 with B. quintana. B. henselae was isolated from 18 of the 25 cats (72 percent) and from at least 1 cat in each household of a patient infected with B. henselae, with one exception. This patient's banked serum samples demonstrated the presence of bartonella antibodies several years before he acquired his two culture-negative cats (unpublished data). The cat linked by household to the sole B. quintana–infected patient was culture-negative for bartonella.
Molecular Subtyping of B. henselae Isolates and Comparison of Isolates from Patients and Their Cats
Subspeciation of the B. henselae isolates from the last three consecutively enrolled patients and their 13 cats was performed by PCR with REP and ERIC primers.18 A bartonella species was isolated from the only cat of one patient, 1 of the 3 cats of the second patient, and 8 of the 9 cats of the third patient (overall, 10 of 13 cats [77 percent]). By REP PCR (Figure 3, upper panel), similar patterns were obtained for all isolates but one (lane 15); by ERIC PCR (Figure 3, lower panel), three different patterns were observed (pattern B in lanes 2 to 4, pattern C in lanes 5 to 14, and pattern I in lane 15).18 Subspeciation based on the composite pattern of REP PCR and ERIC PCR18 revealed that the three human isolates belonged to two different subtypes of B. henselae: Patient 1 had subtype II, and Patients 2 and 3 had subtype III. The patient's subtype matched the subtype obtained from the cats to which the patient was exposed, except for one of the eight cats of Patient 3, which was infected with B. clarridgeiae (Figure 3, lane 15). The subtype III isolate pattern for Patient 2 and Cat 2 was consistently distinguishable from the other subtype III isolates by a distinct band at 910 base pairs on REP PCR (Figure 3, upper panel). Although this finding does not meet the criteria for a different genomic fingerprint,18 it enabled us to match the Cat 2–Patient 2 pair when all isolates were initially run as blinded samples.
|
We found that the risk factors and organ tropism associated with bacillary angiomatosis–peliosis differed according to the infecting species, B. henselae or B. quintana. In addition, the temporal pattern of occurrence of bacillary angiomatosis–peliosis in the San Francisco Bay area differed for these two species (Figure 1). Interestingly, clustered B. quintana infection occurred simultaneously in Seattle (bacteremia in predominantly Native American, non–HIV-infected men)21 and San Francisco (bacillary angiomatosis in predominantly white, HIV-infected persons) from January to June 1993. B. quintana was associated with homelessness, low socioeconomic status, and exposure to head or body lice in our study. Cases of B. henselae infection occurred sporadically and were epidemiologically linked with cat bites and scratches and owning a cat with fleas. Finally, treatment with macrolide antibiotics (e.g., erythromycin or clarithromycin) appeared to be protective.
Of nine bartonella species, only two, B. quintana and B. henselae, caused bacillary angiomatosis–peliosis in our 49 patients. Although B. henselae and B. quintana were equally likely to cause cutaneous bacillary angiomatosis (Table 1), only B. henselae was associated with hepatosplenic disease. Every patient in our study with peliosis hepatis was infected with B. henselae, and one fourth of our patients with B. henselae infection had peliosis. In addition, all six speciated cases of peliosis hepatis reported in the literature were associated with B. henselae infection, and none with B. quintana.5,10,22,23 There was also a strong predilection for B. henselae, but not B. quintana, to cause lymph-node disease, whereas B. quintana was associated with osseous and subcutaneous infections.
A mammalian reservoir has been identified for most bartonella species: the domestic cat for B. henselae14 and B. clarridgeiae,24 the vole for B. vinsonii,25 B. grahamii, and B. doshiae,26 and the field mouse for B. taylorii.26 Arthropod vectors have been identified for several bartonella species: the sand fly for B. bacilliformis, the body louse for B. quintana, and, most recently, the cat flea for B. henselae.15 Transmission of B. henselae by the cat flea occurs readily among cats,15 but epidemiologic data implicate the domestic cat as a major vector (by a scratch or bite) of B. henselae from cat to human.16,27,28,29 The role of the flea in infection of humans by B. henselae may either be indirect, because of the higher prevalence of B. henselae infection in cats exposed to infected fleas, or direct, by transmission to humans from fleas, although this has not been documented. Strong evidence for direct transmission of B. henselae from infected cats to owners is provided by the match between the DNA-fingerprint patterns of the isolates from cats and their respective owners.
B. quintana infection occurred in thousands of troops in Europe during World War I, when it caused a relapsing febrile illness known as trench fever.30 The principal vector during this epidemic of B. quintana was the body louse, Pediculus humanus, from a human reservoir. In our study, unmatched subanalysis for lice infestation revealed that only B. quintana infection was associated with lice infestation (P = 0.03), providing an epidemiologic association between lice and contemporary B. quintana infection. In addition, patients with B. quintana infection but not those with B. henselae infection were more often homeless and were of lower socioeconomic status than their matched controls and were thus at risk for exposure to lice.
We observed distinct differences between B. henselae and B. quintana with regard to optimal isolation medium and colony morphology. Inoculation of both agar types was necessary to optimize the recovery of bartonella from patient tissues and blood. Half the patients with bacillary angiomatosis–peliosis had bacteremia, which substantiates the systemic nature of this disease. Because culture of bacillary angiomatosis–peliosis lesions is extremely difficult, blood culture may be the most accessible technique for isolating bartonella species from patients with bacillary angiomatosis–peliosis.
We found that many of the in vitro susceptibilities reported in the literature31,32 do not correlate with the in vivo response to treatment. In our case–control study, trimethoprim–sulfamethoxazole, ciprofloxacin, penicillins, and cephalosporins were not protective, and we were able to isolate bartonella from patients taking most of these antibiotics. In contrast, macrolide therapy was protective, and we were never able to isolate bartonella from a patient taking doxycycline, tetracycline, rifampin, or a macrolide (even a single dose33). Prophylaxis or treatment regimens for Mycobacterium avium complex that include a macrolide may provide simultaneous prophylaxis or treatment of bartonella infection.
Supported by grants from the California Universitywide AIDS Research Program, the National Institutes of Health (R29 AI36075), and the Pierre Richard Dick-Virbac Foundation (to Dr. Koehler), the Fogarty International Center (D43-TW00003, to Ms. Sanchez), and the Florance Bequest of the Australian College of Dermatologists (to Dr. Whitfeld). Dr. Koehler is a Pew Scholar in the Biomedical Sciences.
We are indebted to the University of California, San Francisco, Dermatology faculty (Toby Maurer, M.D.), the Dermatology fellows and residents (Jorge Toro, M.D., Ricardo Ruiz-Rodriguez, M.D., and Anjali Dahr, M.D.), Jane E. Koehler, D.V.M., Carol A. Glaser, D.V.M., M.D., Clay J. Cockerell, M.D., and Arthur L. Reingold, M.D., for their assistance with this project; to the more than 80 clinicians who referred patients and controls for this study, especially from Ward 86 at San Francisco General Hospital; and to the Adult Immunodeficiency Clinic and Veterans Affairs Medical Center at the University of California, San Francisco.
Source Information
From the Departments of Medicine (J.E.K., C.S.G.), Dermatology (M.J.W., T.G.B., P.E.L., J.W.T.), and Pathology (P.E.L.), University of California, San Francisco; Epidemiology Program, School of Public Health, University of California, Berkeley (M.A.S., F.M.C.); the Section of Infectious Diseases, Medical Service, Veterans Affairs Medical Center, and Department of Medicine, Baylor College of Medicine, Houston (M.C.R.-B.); and the National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta (J.W.T.). Presented in part at the Joint Session on Emerging Pathogens, Infectious Diseases Society of America National Meeting and the 36th Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, September 15–20, 1996.
Address reprint requests to Dr. Koehler at Box 0654, 521 Parnassus Ave., Rm. C-443, University of California, San Francisco, CA 94143-0654.
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
