Phyllis C. Tien SM, MD
Jeffery S. Loutit MB, ChB
Essentials of Diagnosis
There are currently 11 known species of Bartonella, four of which are considered to be pathogenic in humans, namely B bacilliformis, B quintana, B henselae, and Bartonella elizabethae. B henselae and B elizabethae have only recently been isolated and identified, but B quintana and B bacilliformis have long been known as the causes of trench fever (B quintana) and Oroya fever and verruga peruana (B bacilliformis). The bartonellae establish intimate relationships with animal hosts, often within the vascular compartment but without causing disease.
The relationship between B bacilliformis and the other three Bartonella species that are pathogenic in humans was established in the early 1990s. This followed the recognition of a new clinical syndrome, bacillary angiomatosis in HIV-infected individuals, in 1983. B henselae was first identified with molecular methods as a cause of this syndrome in 1990, and both B henselae and B quintana were cultured from tissue samples of patients with bacillary angiomatosis in 1992. B henselae was also associated with cat scratch disease in 1992.
B bacilliformis is closely related to B henselae, B quintana, and B elizabethae, as indicated by 16S ribosomal RNA gene sequence analysis. Therefore, these last three organisms, formerly designated as Rochalimaea species, have been reclassified as members of the genus Bartonella.
The spectrum of human disease associated with the bartonellae includes regional and disseminated granulomatous disease in immunocompetent hosts (cat scratch disease), persistent bacteremia or endocarditis, and vascular proliferative disease (verruga peruana or bacillary angiomatosis) in immunocompromised hosts (Box 69-1). B henselae and B quintana are the recognized causes of bacillary angiomatosis, while only B henselae has been implicated as a cause of cat scratch disease. Finally, these two species along with B elizabethae have all been identified in patients with endocarditis and bacteremia.
B henselae is also globally endemic. Cat contact appears to be the single most important factor in the development of cat scratch disease and B henselae-associated bacillary angiomatosis, which reflects the fact that cats are the most important reservoir for B henselae..
There are ~ 22,000 cases of cat scratch disease reported per year in the United States, making this the most common Bartonella-associated disease. Over 80% of cases occur in individuals < 21 years of age. The number of cases peaks in the fall and winter months. This seasonality may reflect differences in human behavior that place persons at risk of exposure to Bartonella species, or it may reflect breeding patterns of cats and fleas. Approximately 18% of household members with cats are seropositive for B henselae antibodies. Risk factors for the development of cat scratch disease include ownership of a kitten (especially one with fleas) and being scratched or bitten by a kitten.
Cats in the southeastern United States, coastal California, the Pacific Northwest, and the south central plains have the highest prevalence of B henselaeantibodies (36–81%). These areas of high prevalence are regions of greater warmth and humidity and closely correlate with the distribution in the United States of the cat flea, Ctenocephalides felis. The lack of B henselae antibody in cats is highly predictive of the absence of bacteremia. Approximately 40% of cats are bacteremic with B henselae; kittens are more likely to be bacteremic than adult cats, as are cats that spend most of their time outdoors. B henselae has also been isolated from fleas of bacteremic cats. The cat flea has been shown to be a competent vector in the transmission of B henselaeamong cats. In the absence of the cat flea, direct cat-to-cat transmission among kittens has not been demonstrated. The role of the flea in transmission of B henselae to humans has not been elucidated.
Recently, an association between another Bartonella species, Bartonella clarridgeiae, and cat scratch disease was reported after isolation of this organism from the blood of a flea-infested kitten that induced cat scratch disease in a human. However, the role of B clarridgeiae as a cause of human infection is unclear.
Although > 90% of patients with cat scratch disease report recent cat contact, only two-thirds of patients with bacillary angiomatosis report recent cat exposure, suggesting that there are other risk factors for developing bacillary angiomatosis. Both B henselae and B quintana are known causes of bacillary angiomatosis, and yet the risk factors and reservoirs for infection with these organisms are quite different. For B quintana-associated disease, risk factors include low income, homelessness, and exposure to lice. Risk factors for B henselae-associated bacillary angiomatosis are similar to those for cat scratch disease, including cat ownership and being scratched or bitten by a cat. B henselae has also been found in fleas from cats owned by patients with bacillary angiomatosis.
Despite the advent of serologic testing for Bartonella species, the incidence of bacillary angiomatosis remains unknown. Of all reported cases of bacillary angiomatosis, ~ 90% are men that are coinfected with HIV. Cases have been reported from all areas of the United States, with greater numbers in New York City, San Francisco, and some parts of Florida and Texas. Bacillary angiomatosis is uncommon in immunocompetent individuals; conversely, it can occur in patients who are immunosuppressed from causes other than HIV. Bartonella bacteremia with fever occurs in both immunosuppressed and immunocompetent individuals. Because special, nonroutine techniques are required for its isolation, bacteremia with Bartonella spp. is likely to be underreported.
B bacilliformis is geographically confined to intermediate altitudes (between 1000 and 3000 m) of the South American Andes. This coincides with the limited distribution of its sandfly vector, Lutzomyia vernicarum.
Bartonella species can be cultivated when plated on freshly prepared, enriched (blood-containing) solid media and placed in a humidified atmosphere containing 5% CO2. Optimum temperature for growth is 35–37°C, except for B bacilliformis, which grows best at 25–30°C. Colonies are detected in 5–21 days. The colonies of Bartonella species from primary isolation are either white, raised, rough, and autoadherent or small, tan, moist appearing, and imbedded in the agar. Both types of colonies are usually seen in the same culture, but B henselae is usually more heterogeneous with a larger proportion of the rough morphology than B quintana, which has predominantly smooth colonies. B henselae and B quintana are difficult to distinguish because they are closely related at both the phenotypic and genotypic levels.
Bacillary angiomatosis is a result of lobular proliferation of blood vessels. Small capillaries are arranged around ectatic ones with edema, mucin, or fibrotic stroma surrounding the lobules. The endothelial cells that make up the vascular lobules occur either singly, in small groups, or in areas of solid proliferation. Endothelial cell necrosis may also be seen along with neutrophils and sometimes macrophages. In most cases, clusters of neutrophils and neutrophilic debris are seen adjacent to the capillaries scattered throughout the lesion. It is believed that B henselae and B quintana directly stimulate the vascular proliferation seen in the lesions of bacillary angiomatosis. Angioproliferative lesions, however, have not been reported in patients with other forms of B henselae and B quintana infection such as trench fever and cat scratch disease.
The flagella of B bacilliformis allow invasion of human erythrocytes, and the resulting cell lysis leads to the severe hemolytic anemia seen in the early stages of South American bartonellosis (Oroya fever). The nodular skin lesions that develop in the late stages of this disease reveal a characteristic vascular proliferation of endothelial cells that resembles bacillary angiomatosis.
BOX 69-1 Bartonella Infection Syndromes
Table 69-1. Clinical Manifestations of Bartonella Infections
Subcutaneous nodules or lobules are the second most common presentation and may vary markedly in size and number. Underlying cortical bone erosion may be associated with these nodules. There is often minimal epidermal change or hyperpigmentation, but, infrequently, the overlying skin changes may resemble cellulitis. While cutaneous disease is associated with both Bartonella species, subcutaneous and bony disease is associated almost exclusively with B quintana. Regional lymph node enlargement may be present; these nodules may be tender but rarely ulcerate or bleed.
Less common are indurated hyperpigmented plaques, which are often oval in shape and several centimeters in diameter. These hyperkeratotic plaques are typically found on the extremities.
Figure 69-1. Typical papular and nodular lesions of bacillary angiomatosis in the same patient. At the macroscopic level, these lesions are easily confused with those of Kaposi's sarcoma. (Reprinted with permission from Ann Intern Med 109:451, Figure 2, 1988.)
Liver and spleen involvement by Bartonella spp. with or without cutaneous disease is frequently associated with B henselae in the immunocompromised host. The spectrum of liver and spleen pathology varies from a nonspecific necrotizing inflammatory reaction to bacillary peliosis. Bacillary peliosis is characterized by cystic blood-filled spaces throughout the tissue parenchyma. Since the advent of AIDS, peliosis, a previously rare entity associated with chronic debilitating illnesses (such as tuberculosis and malignancy) and anabolic steroid use, has increased. Bacillary peliosis is fatal in rare cases.
Bacillary angiomatosis may also involve the mucus membranes of the mouth, nose, larynx, bronchi, conjunctiva, and anus; the lung and pleura; bone; and the central nervous system. Bacillary angiomatosis should strongly be considered in any HIV-infected patient with skin and lytic bone lesions.
CAT SCRATCH DISEASE
B BACILLIFORMIS-ASSOCIATED DISEASE
About 3 weeks after the bite of an infected sandfly, acute B bacilliformis infection becomes manifest as a febrile bacteremic illness (Oroya fever). It is accompanied by an acute hemolytic anemia that is caused by bacterial attachment and invasion of erythrocytes (Table 69-1). Patients exhibit malaise, fever, chills, diaphoresis, headache, and mental status changes. Lymphadenopathy, hepatosplenomegaly, and thrombocytopenia are also seen. At the end of the acute bacteremic phase, there is a transient immunosuppression that renders the patient susceptible to a wide range of opportunistic bacterial, viral, and parasitic secondary infections. Mortality during this acute phase is higher among persons without previous exposure to this agent. Overall, mortality is estimated to be ~ 8%.
After the acute phase of infection, 15% of patients develop a chronic form of disease manifested by painless superficial or subcutaneous nodules (verruga peruana). These nodules are histologically similar to those of bacillary angiomatosis.
FEVER & BACTEREMIA/TRENCH FEVER/ENDOCARDITIS
The four Bartonella species that are pathogenic for humans are capable of causing sustained or relapsing bacteremia accompanied by only fever (Table 69-1). All except B bacilliformis also cause endocarditis. After B quintana enters the body through broken skin from the excreta of the infected human body louse (Pediculus humanus), there is an incubation period of between 5 and 20 days before the onset of trench fever. Patients complain of fever, myalgias, malaise, headache, bone pain—particularly of the legs, and a transient macular rash. Usually the illness continues for 4–6 weeks. Sustained or recurrent bacteremia is common, with or without symptoms.
The form of trench fever described in the United States and Europe in the 1990s (urban trench fever) has no distinguishing features other than fever, malaise, and weight loss. Endocarditis has been reported in conjunction with B quintana bacteremia.
In HIV-infected patients, B henselae bacteremia usually presents insidiously with malaise, fatigue, weight loss, and fevers. Endocarditis may occur in immunocompetent hosts with patients presenting with fever and malaise.
Difficulty in the identification and isolation of Bartonella species from blood and tissue has made a diagnosis of infection with these organisms a challenge. Direct examination of tissue (not blood) with special stains is required, and special conditions for cultivation from tissue and blood are necessary.
The detection of these small, curved gram-negative bacilli in pathology specimens is best made by using the Warthin-Starry silver stain, electron microscopy, immunofluorescence, or the polymerase chain reaction. Conventional stains such as acid-fast, Giemsa, periodic acid-Schiff, and the Brown-Brenn modification of Gram's stain fail to demonstrate the bacilli. The Warthin-Starry stain results are not pathognomonic since it also stains other bacteria, including spirochetes and species of Nocardia, Legionella, and Campylobacter. Polymerase chain reaction-based assays offer high sensitivity and specificity but are not widely available.
Isolation of B henselae and B quintana after homogenization of tissue such as liver, spleen, lymph node, and skin can be performed by direct plating or by cocultivation with an endothelial cell line. The cocultivation method, while useful in recovering organisms, may not be practical for most microbiology laboratories. Growth of strains on freshly prepared heart infusion agar containing 5% or 10% defibrinated rabbit or horse blood has been more successful than growth on other cell-free media such as chocolate or 5% sheep blood agar.
The lysis-centrifugation blood culture system is currently the best method to isolate Bartonella species from the blood. If bacteremia with a Bartonella species is suspected, the laboratory must be informed in order that lysis centrifugation-processed blood can be placed on the appropriate media and incubated for at least 7–14 days. Because of the fastidious nature of these organisms, the interval from collection to processing should be minimized. Other alternatives to the lysis centrifugation method include screening suspected blood culture bottle contents with an acridine orange staining procedure and subculturing positive bottles.
Differentiation of the Bartonella species can be accomplished with biochemical and molecular methods. Cellular fatty acid analysis by gas liquid chromatography can be helpful, but many laboratories do not have the necessary capabilities to perform this test. The Microscan rapid anaerobic panel has been used with some success to distinguish between B henselae and B quintana. Other methods have included polymerase chain reaction–restriction fragment length polymorphism analysis of the citrate synthase gene and 16S rDNA sequence analysis.
Serologic testing is easier and more rapid than cultivation of the organism and has become the mainstay for confirming the diagnosis of cat scratch disease. Both immunofluorescence assays and enzyme immunoassays are available to detect cerebrospinal fluid and serum immunoglobulin G (IgG) and IgM directed against Bartonella spp. A titer of > 1:64 or a fourfold rise in titer is considered positive. Because of the cross-reactivity of antibodies against B quintana and B henselae, definitive species identification is difficult with these methods.
Treatment of Bartonella-associated disease varies according to the clinical syndrome (Box 69-2). All recommendations are based on anecdotal evidence of treatment response since controlled trials have not been performed.
Bacillary angiomatosis should always be treated. Spontaneous resolution of superficial skin papules has been reported, but relapse and bacterial dissemination are common. Nevertheless, therapeutic endpoints remain undefined.
Anecdotal evidence strongly favors the use of oral erythromycin for treatment of bacillary angiomatosis at a dose of 500 mg four times per day for 8–12 weeks. Intravenous erythromycin should be administered to patients with severe disease or patients who are unable to take oral medications. Longer courses may be necessary in patients with bacteremia/endocarditis or bacillary peliosis and those with disease recurrence. The newer macrolides, azithromycin and clarithromycin, may also be effective, but experience with these drugs is more limited. Oral doxycycline (100 mg twice daily) has been consistently successful in the treatment of bacillary angiomatosis. Jarisch-Herxheimer–like reactions after the first several doses of doxycycline or erythromycin can occur, however. Use of an antipyretic before treatment may attenuate this response.
Limited data support the use of ciprofloxacin in the treatment of bacillary angiomatosis. Inhibitors of cell wall synthesis, such as the penicillins and cephalosporins, uniformly fail to cure the disease, despite the fact that in vitro testing indicates that both B quintana and B henselae are sensitive to these antibiotics. Susceptibility testing is not standardized for B henselae or B quintana and is difficult to perform. In vitro data should not be used to guide therapy at this time.
BOX 69-2 Treatment of Bartonella Infections
Cat scratch disease usually resolves spontaneously over a 2- to 8-week period. Aspiration of a suppurative node may provide symptomatic relief. Antibiotic therapy is not consistently successful, nor is it routinely indicated, unless the disease occurs in an immunocompromised host. Despite the success of macrolides and tetracyclines in the treatment of bacillary angiomatosis, they have not been used as widely in the treatment of cat scratch disease. A prospective, randomized, double-blind trial has suggested that azithromycin treatment for 5 days reduces lymph node size more rapidly than placebo. Rifampin, ciprofloxacin, trimethoprim-sulfamethoxazole, and gentamicin have been proposed as effective agents, based only on a retrospective review of the literature. Mild or moderate disease should receive only symptomatic therapy, since antimicrobial agents do not change the clinical outcome of cat scratch disease in the majority of cases.
Urban trench fever has been suppressed and sometimes cured with erythromycin or azithromycin given for at least 4 weeks; however, as with B henselaebacteremia, response to antimicrobial agents is variable, and some patients experience a relapse of symptoms after cessation of antibiotics. Some cases of endocarditis require valve replacement and most require prolonged antibiotic therapy.
The response of Oroya fever and verruga peruana to therapy is variable. Chloramphenicol has been recommended in endemic regions due to the possibility of intercurrent Salmonella infection and the high case fatality rate.
Prevention & Control
Cat owners can, and immunocompromised individuals should, reduce their risk of cat scratch disease or bacillary angiomatosis (Box 69-3). Cat owners who are immunocompromised should (1) wash hands after animal contact and keep all wounds and abrasions as clean as possible; (2) avoid cats < 1 year of age or avoid rough play with them since they are more prone to “carry” the organism; and (3) keep cats as free from fleas as possible. Declawing of cats is not recommended. Antibiotic treatment of infected cats is not useful.
B quintana infection may be prevented or controlled by delousing procedures. These include regular changing or washing of clothes and bedding. Insecticides such as permethrin should also be used to control louse infestation. Oroya fever and verruga peruana can be prevented by avoiding the bite of the sandfly vector. These insects tend to be nocturnal feeders.
BOX 69-3 Control of Bartonella Infections
Anderson BE, Neuman MA: Bartonella spp. as emerging human pathogens. Clin Microbiol Rev 1997;10(2):203.
Loutit JS. Bartonella infections. Curr Clin Topics Infect Dis 1997;17:269.
Maurin M, Raoult D: Bartonella quintana infections. Clin Microbiol Rev 1996;9(3):273.
Koehler JE, et al. Molecular epidemiology of Bartonella infections in patients with bacillary angiomatosis-peliosis. N Engl J Med 1997;337(26):1876.
Koehler JE, et al: Rochlimaea henselae infection: a new zoonosis with the domestic cat as reservoir. JAMA 1994;271(7):531.
Koehler JE, et al: Isolation of Rochalimaea species from cutaneous and osseous lesions of bacillary angiomatosis. N Engl J Med 1992;327(23):1625.
Zangwill KM, et al: Cat scratch disease in Connecticut: epidemiology, risk factors and evaluation of a new diagnostic test. N Engl J Med 1993;329(1):8.