Bartonella

TAXONOMY AND DESCRIPTION OF THE GENUS Back to top

Bartonella (including some species formerly known as Rochalimaea) is a genus of short,

facultative intracellular pleomorphic gram-negative coccobacillary or bacillary rods that

measure 0.2 to 0.6 μm by 0.5 to 1.0 μm (Fig. 1A and B). Members of this genus can induce

acute and persistent intravascular infections in healthy people and

animals. Bartonella species are members of the alpha-2 subgroup of the

class Alphaproteobacteria, within the Rhizobiales order. There are now more than 22 species

or subspecies described, and DNA sequences from numerous other species or strains have

been deposited in GenBank. Because of the slow growth of these bacteria on agar plates

(dividing time, approximately 24 h), standard biochemical methods for identification have

not been useful. Bartonella organisms are oxidase and catalase negative and do not produce

acid from carbohydrates. On primary culture all members of the genus typically require 15 to

45 days in the presence of 5% CO2 to form visible colonies on enriched blood-containing

media, as these bacteria are highly hemin dependent. The temperatures to achieve optimal

growth for various Bartonella spp. vary from 25 to 30°C forBartonella bacilliformis to 35 to

37°C for B. henselae, B. koehlerae, and B. elizabethae. On primary isolation,

someBartonella species, such as B. henselae, B. clarridgeiae, B. vinsonii, and B.

elizabethae, form colonies with a white, rough, dry, raised appearance that pit the medium

(Fig. 1C). The smooth colony morphology results from phase variation that correlates with

the loss of trimeric autotransporter adhesin (TAA) expression. Colonies are hard to extract or

transfer to subsequent plates. Other Bartonella spp., such as B. quintana, have colonies that

are usually smaller, gray, translucent, and somewhat gummy or slightly mucoid. A few

members of the genus, including B. bacilliformis, B. clarridgeiae, B. capreoli, and B.

schoenbuchensis, are motile by means of unipolar flagella.

Bartonella bacilliformis, the type species of the genus, is the etiologic agent of Carrion’s

disease, which is characterized by an acute hemolytic, bacteremic infection called Oroya

fever or a chronic vasoproliferative form called verruga peruana, which is characterized by

cutaneous nodular vascular eruptions. Bartonella quintana,the agent of trench fever, has also

been found to be one of the agents of bacillary angiomatosis, a vascular proliferative lesion

observed in immunocompromised individuals, mainly with AIDS (76). B. quintana has also

been associated with endocarditis and bacteremia in homeless people (17). Although the

organism is transmitted by the human body louse, B. quintana DNA has been amplified from

cat fleas and has also been isolated from feral cats and from dogs with endocarditis

(8, 16, 44, 70). Bacillary angiomatosis can also be caused by B. henselae. Although B.

henselae has historically been associated with cat scratch disease (CSD), a frequent but selflimiting

infection in immunocompetent individuals, more recent findings suggest that this

organism may cause persistent bacteremia accompanied by fatigue, arthritis, and

neurological or neurocognitive abnormalities (14–16, 93). Similarly, B. henselae has been

associated with endocarditis in both humans and dogs. A fourth human pathogen, B.

elizabethae, was isolated in an immunocompetent individual suffering from endocarditis.

Since then, Bartonella vinsonii subsp. berkhoffii, B. vinsonii subsp. arupensis, B.

koehlerae, and B. alsatica have also been associated with human cases of endocarditis (7)

and “CandidatusBartonella washoensis” with a human case of myocarditis. Bartonella

grahamii has been associated with human cases of neuroretinitis and bilateral retinal artery

branch occlusions. B. vinsonii subsp. arupensis was also isolated from the blood of a rancher

with fever and mild neurological symptoms (147). B. clarridgeiae is suspected to be a minor

agent of CSD, but based on serologic evidence only. Three new proposed Bartonellaspecies

have been isolated or detected from human cases, including “B. rochalimae,” “B.

tamiae,” and “Candidatus Bartonella melophagi.”

EPIDEMIOLOGY AND TRANSMISSION Back to top

Bartonella species are transmitted by insect vectors such as fleas, sand flies, and body lice

and potentially by ticks, biting flies, and wingless flies. In addition, transmission also occurs

by animal scratches or possibly bites. Some species are very limited geographically, such

as B. bacilliformis, found only in the Andes mountain region of South America, in association

with the limited distribution of its sand fly vector, Lutzomyia verrucarum.Others, such as B.

quintana, B. henselae, B. elizabethae, and B. vinsonii subsp. berkhoffii, appear to have a

worldwide distribution. B. quintana outbreaks are associated with poor sanitation and

personal hygiene, particularly in homeless populations worldwide. Infestation by the human

body louse, Pediculus humanus,results in inoculation of B. quintana in arthropod excreta

through broken skin (39, 112, 114).

Cats are the main reservoir for B. henselae, B. clarridgeiae, and B. koehlerae. CSD, caused

by B. henselaeinfection, is transmitted from cat to cat mainly by the cat

flea, Ctenocephalides felis (25); however, cat flea transmission as a cause of human

infection has not been confirmed. More likely, human infection results from the inoculation of

infective flea feces at the time of the scratch (7). Ticks could be a potential vector for some

human B. henselae infections (7). Stray cats, cats living outdoors, and young cats are more

likely to be bacteremic. The prevalence of B. henselae infection is usually highest in warm

and humid climates, where cat fleas are abundant. Bartonella henselae antibody prevalence

in domestic cat populations can range from 14 to 50% (4, 10). Genotype I (Houston) is more

common in cats in the Far East (Japan and the Philippines), whereas type II (Marseille) is

predominant in Western Europe, North America, and Australia (7). In addition, genotype I

seems to have two distinctive variants in the United States: B. henselae Houston I and B.

henselaeSan Antonio 2, according to sequence differences in the internal transcribed spacer

(ITS) region (15, 43, 44,56, 92, 94). Coinfection of cats with different Bartonella species or

genotypes has been reported (53), as well as coinfections in humans or dogs (15, 43, 44).

Additionally, B. quintana, B. koehlerae, and B. clarridgeiae DNA has been detected in cat

fleas, suggesting their possible role as vectors for these organisms (123).

Several Bartonella species and subspecies, including B. clarridgeiae, B. quintana, B.

elizabethae, B. henselae, B. vinsonii subsp. berkhoffii, “B. washoensis,” and “B.

rochalimae” (58), can infect dogs and humans (Table 1) and may elicit a wide spectrum of

disease manifestations, including polyarthritis, cutaneous vasculitis, endocarditis,

myocarditis, epistaxis, peliosis hepatis, and granulomatous inflammatory disease. B.

alsatica (European wild rabbit reservoir) and “Candidatus Bartonella melophagi” (sheep

reservoir) have been reported in association with endocarditis, lymphadenitis, pericardial

effusion and fatigue, joint pain, and tremors (93, 113). OtherBartonella species, such as B.

vinsonii subsp. vinsonii, B. doshiae, B. taylorii, B. peromysci, B. birtlesii, B. tribocorum, B.

talpae, B. bovis, B. chomelii, B. schoenbuchensis, and B. capreoli, have only been isolated

from the blood of animals, including wild rodents, squirrels, felids, canids, and ruminants

(cattle, deer, and elk). Domestic and wild canids represent the main reservoir of B.

vinsonii subsp. berkhoffii, with high antibody prevalences in dogs from tropical countries (7)

and a high prevalence of bacteremia in coyotes (Canis latrans)in California (21). Ticks may

be involved in the transmission of B. vinsonii subsp. berkhoffii to dogs (11).

Similarly, Bartonella DNA has been identified in questing adult Ixodes pacificus ticks from

California, including from several Bartonella species that are pathogenic for humans (20); B.

henselae DNA has also been detected in Ixodes ricinus ticks collected from humans in Italy

(131); and Bartonella DNA has been amplified from Ixodesticks from numerous sites

throughout the world (5, 35a).

CLINICAL SIGNIFICANCE Back to top

Human Pathogens

Oroya Fever and Verruga Peruana (Carrion’s Disease): B. bacilliformis

The disease caused by B. bacilliformis, especially its chronic form known as verruga peruana,

has been recognized since pre-Columbian times in populations of the Andes Mountains.

However, the suspected link between the acute form (Oroya fever) and the chronic form was

confirmed in 1885 when Daniel Carrion, a medical student, died of Oroya fever after

inoculating himself with material from a verruga. The acute form of the disease, usually seen

in people who are not natives of the zone of endemicity, is an acute, progressive, severe,

and febrile anemia with intravascular hemolysis associated with the presence of B.

bacilliformis in the erythrocytes. The mortality rate was reported to range from 40 to 90%

prior to the antibiotic era. Chronic bacteremia occurs in the general population in regions of

endemicity, and evidence for persistent infection has been reported from zones of

nonendemicity (19, 80). The second stage of infection occurs weeks to months following the

acute infection and is characterized by nodular angioproliferative cutaneous lesions named

verruga peruana; mucosal and internal lesions can occur (118, 119). The lesions can persist

for several months, but the prognosis for full recovery is good at this stage.

Trench Fever, Bacillary Angiomatosis, Endocarditis, and Prolonged

Fever or Bacteremia Caused by B. quintana

Trench or quintana fever is a recurrent fever with three to five or more febrile episodes

lasting 4 to 5 days each after 15 to 25 days of incubation. Severe headaches and shin pain

are common symptoms associated with malaise, anorexia, abdominal pain, restlessness, and

insomnia. Mild forms and asymptomatic carriage are also reported (17, 65). Several cases of

endocarditis have been associated with B. quintana infection (84). In human

immunodeficiency virus-infected persons, bacteremia caused by B. quintana develops

insidiously, involving recurrent fever, headaches, and hepatomegaly. B. quintana and B.

henselae are the two Bartonellaspecies involved in the etiology of bacillary angiomatosis.

Bacillary angiomatosis, also called epithelioid angiomatosis, is a vasoproliferative disease of

the skin characterized by multiple blood-filled, partially endothelial cell-lined cystic structures

(135). It is usually characterized by violaceous or colorless papular and nodular skin lesions

that clinically suggest Kaposi’s sarcoma but histologically resemble epithelioid hemangiomas.

When visceral parenchymal organs are involved, the condition is referred to as bacillary

peliosis hepatis, splenic peliosis, or systemic bacillary angiomatosis. Fever, weight loss,

malaise, and organomegaly can develop in people with disseminated bacillary angiomatosis.

Subcutaneous and lytic bone lesions are also associated with B. quintana infection (75).

Zoonotic Bartonellae

CSD

CSD is caused mainly by B. henselae (115, 153), whereas B. clarridgeiae has been suspected

in a few cases, based on serologic evidence (77, 97). There are now strong arguments

against Afipia felis being one of the etiological agents of CSD (84). In classical CSD, 1 to 3

weeks elapse between the scratch or bite of a cat and the appearance of clinical signs. In

50% of the cases, a small skin lesion, often resembling an insect bite, appears at the

inoculation site, usually the hand or forearm. The lesion evolves from a papule to a vesicle

and in some instances becomes a partially healed ulcer. These lesions generally resolve

within a few days to a few weeks. Lymphadenitis develops approximately 3 weeks after

exposure and is generally unilateral. Epitrochlear, axillary, or cervical lymph nodes are most

frequently involved. Lymph nodes are usually swollen and painful, and lymphadenopathy

persists for several weeks to several months. In 25% of CSD cases, suppuration occurs. The

large majority of the cases show signs of systemic infection: fever, chills, malaise, anorexia,

and headache. In general, the disease is benign and lymphadenopathy resolves

spontaneously without sequelae. Atypical manifestations of CSD occur in 5 to 10% of the

cases. The most common of these is Parinaud’s oculoglandular syndrome (periauricular

lymphadenopathy and palpebral conjunctivitis), but meningitis, encephalitis, osteolytic

lesions, and thrombocytopenic purpura may also occur. Encephalopathy is one of the most

serious complications of CSD, which usually occurs 2 to 6 weeks after the onset of

lymphadenopathy but generally resolves with complete recovery and few or no sequelae. It

is estimated that 22,000 human cases of CSD occur yearly in the United States. From 55 to

80% of CSD patients are <20 years old. There is a seasonal pattern, with most cases seen in

autumn and winter.

Other clinical presentations associated with B. henselae infection are reported to occur in

immunocompetent persons, including neuroretinitis or bacteremia accompanied by chronic

fatigue syndrome, memory loss, and arthralgia. Bartonella henselae is also a frequent cause

of prolonged fever and fever of unknown origin in children. Rheumatic manifestations

of Bartonella infection have been described to occur in children, including cases of myositis,

arthritis, and skin nodules. Arthritis is described in a very limited number of cases. Other

rheumatic manifestations include erythema nodosum, leukocytoclastic vasculitis, fever of

unknown origin with myalgia, osteolytic lesions, and arthralgia (7). Neurological dysfunction

and neurocognitive abnormalities are reported for people infected with B. henselae (15).

Unlike CSD, which appears to be a self-limiting infection in most patients, these cases

suggest that B. henselae can induce persistent intravascular infections resulting in varied and

complex chronic disease manifestations.

Bacillary angiomatosis patients with B. henselae infection are epidemiologically linked to cat

and flea exposure (75). Fever, weight loss, malaise, and enlargement of affected organs may

develop in people with disseminated bacillary angiomatosis. Bartonella henselae and B.

quintana have been implicated in a few cases of human immunodeficiency virus-associated

brain lesions, meningoencephalitis and encephalopathy, dementia, and neuropsychological

decline (132, 136).

Zoonotic Bartonella Species Associated with Endocarditis, Myocarditis,

Ocular Lesions, Fever, and Neurological Symptoms

Several zoonotic Bartonella spp. have been recognized as causative agents of blood culturenegative

endocarditis or myocarditis in humans, including B. henselae, B. koehlerae, B.

elizabethae, B. vinsonii subsp.berkhoffii, B. vinsonii subsp. arupensis (48), and

“Candidatus Bartonella washoensis” (7, 30). Bartonella spp. account for 3 to 4% of all

human cases of endocarditis in France, a percentage similar to that of endocarditis cases

caused by Coxiella burnetii, the agent of Q fever (60). Some rodent-borne Bartonella species

are also associated with cases of neuroretinitis (B. elizabethae and B. grahamii) or fever with

bacteremia and neurological symptoms (B. vinsonii subsp. arupensis) (147).

Other Bartonella Species or Subspecies

The clinical impact on animals or humans of many Bartonella species is still unknown.

Although “CandidatusBartonella melophagi” was isolated from the blood of two women (93)

and B. bovis DNA was amplified from the heart valves of cows with endocarditis, no other

specific pathology has been associated with Bartonellaspecies infecting domestic and wild

ruminants or for many of the rodent-borne Bartonella spp.

Canine and Feline Bartonella Species

Several of the 22 species and subspecies known today have been detected in or isolated

from pet dogs and cats, thereby highlighting the zoonotic potential of these bacteria for

persons with extensive animal contact (26, 28–30, 42, 44, 71, 73, 127, 137, 138). B.

henselae, B. clarridgeiae, B. koehlerae, B. quintana, B. bovis, B. elizabethae, and B.

vinsonii subsp. berkhoffii have been recognized as pathogenic for dogs (13, 27), and B.

henselae, B. clarridgeiae, “Candidatus Bartonella washoensis,” B. quintana, B. rochalimae, B.

elizabethae, and B. vinsonii subsp. berkhoffii also cause feline infections

(8, 41, 57, 77, 82, 88, 98). A broad array of manifestations, including endocarditis (22),

myocarditis, epistaxis, and lethargy, have been commonly associated

with Bartonella infection in dogs (9, 12, 13, 78, 79). Bartonella-infected cats are more likely

to have kidney disease and urinary tract infections, stomatitis, and lymphadenopathy (28).

In experimentally infected cats, fever, lymphadenopathy, mild neurological signs, and

reproductive disorders have been reported to occur (28). Nevertheless, the clinical spectrum

of Bartonella infection in dogs and cats may be highly variable, including chronic subclinical

infections accompanied by lethargy and weight loss as the only reported abnormalities.

COLLECTION, TRANSPORT, AND STORAGE OF

SPECIMENS Back to top

Most specimens used for Bartonella isolation are either blood or tissue. Approaches typically

used for recovery of other pathogens from such sites are suitable. Bartonella spp. are

difficult to isolate from blood of immunocompetent individuals, as opposed to the relative

ease of isolation from blood of immunocompromised patients. Blood samples can be

collected either in Isolator blood lysis tubes (Wampole, Cranbury, NJ), sodium citrate tubes,

or plastic EDTA tubes. If storage of specimens prior to culture is necessary, samples should

be kept frozen (at least −20°C). It was shown that blood collected from B. henselae-infected

cats into both EDTA and Isolator blood lysis tubes yielded good recovery and no loss of

sensitivity for EDTA tubes kept at −65°C for 26 days (10). Specimens should be collected

prior to antimicrobial therapy.

Tissue from enlarged lymph nodes, cutaneous lesions, or various organs can be cultured

after homogenization or processed for DNA extraction and PCR. Fresh tissues are preferred

for PCR amplification, but paraffin-embedded tissue may be used (23). Fine-needle

aspiration has also been successful for detection of Bartonella and is less invasive than

biopsy (3). Bartonella spp. have been successfully cultured from aqueous humor (51, 70).

DIRECT EXAMINATION Back to top

Microscopy

With the exception of B. bacilliformis, Bartonella spp. are not visualized in erythrocytes on

stained smears of blood from animal or human patients (84). In rats experimentally infected

with B. birtlesii, there can be up to eight bacteria per erythrocyte (6a, 29a). Warthin-Starry

silver stain is recommended for microscopic detection of Bartonella organisms in fixed tissue

sections but is not highly specific and is insensitive, even with lymph node biopsy samples

from CSD patients. For patients with bacillary angiomatosis, there is usually a larger number

of bacilli that are identifiable by Warthin-Starry silver staining.

Antigen Detection

Immunocytochemical labeling is a specific technique but is not widely available (76). Direct

immunofluorescence of blood smears allowed rapid diagnosis of B. quintana in a patient with

acute trench fever (26) and in bacteremic homeless patients (84).

Nucleic Acid Detection

Although PCR and agar plate culture are useful tests to document infection

with Bartonella species in cats, these traditional techniques are not generally sensitive

enough to detect active infection with a Bartonellaspecies in human or dog blood samples.

Therefore, PCR amplification directly from blood or other diagnostic samples is of limited

value for those patients with very-low-level bacteremia. Enrichment culture of the diagnostic

sample (blood, cerebrospinal fluid, joint fluid, or effusion) prior to performing PCR (see

below) can be used to increase the number of bacteria above the threshold limit of the PCR

(10, 14–16, 18, 24, 31, 32,37, 43, 44, 56, 69, 91, 93, 94, 105, 152; R. Maggi, A. W.

Duncan, and E. B. Breitschwerdt, presented at the 20th ASR Meeting and 5th International

Conference on Bartonella as Emerging Pathogens, Pacific Grove, CA, 2006).

Advances in PCR methodologies and equipment have facilitated an impressive increase in the

sensitivity of molecular detection of Bartonella DNA in patient samples. The 16S rRNA

gene (rrs) was first used by Relman et al. in 1990 for identification of the microbial etiology

of bacillary angiomatosis (34, 116). However, this gene does not discriminate

among Bartonella species (86). The most widely targeted genes are those coding for citrate

synthase (gltA) (68, 83, 109, 111), a heat shock protein (groEL) (154), riboflavin

synthase (ribC) (2), a cell division protein (ftsZ) (155, 156), and a 17-kDa antigen (47).

Sequences of gltA and rpoB (RNA polymerase β-subunit) (117) are congruent with DNA-DNA

hybridization for Bartonella species identification (86). The ITS region located between the

16S and 23S rRNA genes is a useful diagnostic target for Bartonella detection, species

identification, and genotyping (14–16, 42, 44, 59, 67, 89, 90). Confirmatory tests (e.g., PCR

followed by sequencing or the use of a second target gene PCR assay) must be performed to

avoid potential nonspecific amplification and misdiagnosis (89).

ISOLATION PROCEDURES Back to top

Blood

Bartonellae can be recovered from the blood of bacteremic patients. In immunocompromised

patients, including transplant recipients, the level of bacteremia is often higher than in

immunocompetent people (76). Routine cultures of blood from patients with infective

endocarditis caused by Bartonella spp. are rarely positive. Historically, Bartonella species

were optimally isolated from blood with the use of the lysis centrifugation system (Isolator)

or with EDTA anticoagulant and plating onto fresh chocolate or heart infusion agar containing

5% fresh rabbit blood in the absence of antibiotics. Commercial sheep or horse blood agar

plates have also been used. Bartonella koehlerae does not grow well on heart infusion agar

and requires fresh chocolate agar plates (41). Plates inoculated with blood should be

incubated at 35°C for at least 4 weeks in 5% CO2 and at high humidity. Colonies usually

appear after 5 to 15 days (74). Due to prolonged incubation times, other slow-growing

pathogens, including Mycobacterium tuberculosis, can grow on the plates; therefore,

appropriate safety precautions should be taken with all positive cultures. Broth-based or

biphasic culture system vials used for blood culture, such as the BACTEC Peds Plus vials

(Becton Dickinson, Sparks, MD) can be used for Bartonella isolation (54, 99, 143). DNA

staining (e.g., acridine orange staining) and blind subculture from negative bottles before

discarding them at 7 days may increase the likelihood of identifying Bartonella (1). An assay

was also developed in which samples of heparinized blood are sedimented and the plasma is

collected for inoculation into shell vials (66). Culture is then performed by the centrifugationshell

vial technique using the T24 bladder carcinoma cell line (inaccurately designated

ECV304 human endothelial cells) (76).

Preenrichment Culture

In broth, Bartonella spp. do not usually produce turbidity or convert enough oxidizable

substrate to CO2; therefore, CO2 detection-based blood culture systems often fail to indicate

growth. Initial efforts to develop an optimized culture media for the isolation

of Bartonella were aimed at the isolation of these bacteria from sick animals and human

patients (22, 51, 84). Novel approaches, such as using an optimized liquid insect cell growth

medium as a preenrichment step, have enhanced the detection of Bartonella spp. in patient

samples (43, 91, 121). One such medium, designated Bartonella Alphaproteobacteria growth

medium (BAPGM), has enhanced molecular detection and isolation of Bartonella species from

animal and patient samples and is commercially available from Galaxy Diagnostics

(http://www.galaxydx.com). BAPGM supports the growth of at least seven Bartonella species

and facilitates growth of cocultures combining different Bartonella species (43,91). When

used to enhance microbiological documentation of infection, preenrichment culture with

BAPGM facilitated the molecular detection and/or isolation of several Bartonella species (14–

16, 24, 37, 43, 44, 69, 81,90–94; Maggi et al., presented at the 20th ASR Meeting and 5th

International Conference on Bartonella as Emerging Pathogens, Pacific Grove, CA, 2006).

Recently, the use of BAPGM has facilitated the isolation of B. henselae and B.

vinsonii subsp. berkhoffii from pleural and pericardial effusions from dogs and the

documentation of coinfection with B. henselae and B. vinsonii subsp. berkhoffii in blood and

joint effusions obtained from a dog with “nonseptic” neutrophilic polyarthritis (23). An

optimized BAPGM diagnostic platform could include PCR following preenrichment culture in

BAPGM flasks for 7 and 14 days and on colonies if visualized on an agar plate

(14, 16, 24, 93). Diagnostic use of the BAPGM platform has also enhanced detection of

coinfection with more than one Bartonella sp. in dogs (35, 37, 42–44) and in

immunocompetent people (14, 16). In addition, BAPGM preenrichment culture has facilitated

the isolation of two novel Bartonellaspecies, “Candidatus Bartonella melophagi”

and “Bartonella tamiae,” from human blood (81, 93).

Tissue

Recovery of Bartonella spp. from cutaneous lesions, liver, spleen, or lymph node is possible

after homogenization in medium and plating directly onto solid agar. Cocultivation with an

endothelial cell line or use of the shell vial method is more laborious but can occasionally

yield bacterial growth (76), as with isolation of Bartonella from heart valve tissues (84).

Although various liquid media have been developed for primary isolation of B. henselae from

blood, serum, and other tissues (14–16, 22, 24, 35, 37, 43, 44, 56, 69, 81, 91–94,151;

Maggi et al., presented at the 20th ASR Meeting and 5th International Conference on

Bartonella as Emerging Pathogens, Pacific Grove, CA, 2006), current evidence best supports

the use of an insect cell culture-based medium.

IDENTIFICATION Back to top

Colonies of B. henselae can be of two morphological types which can be present

simultaneously: (i) irregular, raised, whitish, rough (cauliflower-like), and dry or (ii) small,

circular, tan, and moist, tending to pit and adhere to the agar (Fig. 1C). B. quintana colonies

are usually smooth, flat, and shiny and do not pit the agar (76). B. clarridgeiae produces

small white, raised, indurated, and cohesive colonies which can also appear to spread during

primary isolation. Most Bartonella spp. usually appear uniformly smooth after repeated

subcultures. The bacilli (Fig. 1) are small (2 by 0.5 μm) and stain best with Gimenez stain

(76). With Gram staining, Bartonellaspp. are weakly counterstained with safranin or basic

fuchsin. Bartonella spp. appear in Gram stains as small, gram-negative, slightly curved rods

resembling Campylobacter, Helicobacter, or Haemophilus (148). Colonial morphology in

conjunction with slow growth requiring more than 7 days of incubation, and negative

catalase and oxidase reactions, is often sufficient for a presumptive identification. DNA

sequencing is optimal for confirmation of the genus, species, and strain.

Bartonella bacilliformis, “Candidatus Bartonella melo phagi,” B. clarridgeiae, B.

capreoli, and B. schoenbuchensisare the only members of the genus that are motile by

means of unipolar flagella. Bartonella quintana and B. henselae as well as several

other Bartonella species have a twitching motility on wet mounts associated with the

expression of TAAs (e.g., BadA of B. henselae;shownin reference 120). These TAAs are

responsible for cytoadherence and may mediate specific interactions with extracellular matrix

components and endothelial cells (96).

Most species are biochemically inert except for the production of peptidases. None of the

various commercially available identification systems contain Bartonella spp. in their

databases. However, the MicroScan Rapid Anaerobe Panel (Baxter Diagnostics, Deerfield,

IL), RapID ANA II, and Rapid ID 32 A have been used for identification. The MicroScan Rapid

Anaerobe Panel is reported to provide species identification (code 10077640 for B.

henselae, code 10073640 for B. quintana, and code 10077240 for B. bacilliformis) (148).

Overall, these identification kits are of limited use for accurate diagnosis

of Bartonella infections. Measurements of preformed enzymes and standard testing reveal

minor differences between species.

Identification of Bartonella isolates is largely based on nucleic acid techniques, some of which

allow for the species determination, strain determination within a given species, or even

genotyping. Methods include Southern blotting, gel and capillary electrophoresis, PCR, DNA

hybridization, restriction fragment length polymorphism, and gene sequence analysis (1). As

described above, PCR and sequencing of target genes such as gltA, rpoB, the ITS

region, rrs (16S rRNA), groEL, or ricC are the most widely used. Recently, it was

demonstrated that mass matrix-assisted laser desorption ionization–time-of-flight mass

spectrometry (MALDI-TOF MS) is an accurate and reproducible tool for rapid and inexpensive

identification of Bartonella species (49a).

TYPING SYSTEMS Back to top

Several molecular methods, such as pulsed-field gel electrophoresis, multilocus sequence

typing, multispacer typing, and multiple-locus variable-number tandem-repeat analysis

(MLVA) (104) can be used to genotypeBartonella henselae (and in some cases B. quintana)

isolates. Also, a genomic fingerprinting technique using infrequent-restriction-site PCR can be

used to identify pathogenic Bartonella species (55). Diagnostic utility for rapid species or

genotype identification is limited when bacterial isolation is not successful. In contrast, even

when isolation of the infecting species is not possible, PCR amplification of ITS DNA directly

from diagnostic samples and/or from enrichment cultures followed by nucleic acid

sequencing is an invaluable tool for primary identification at the species, subspecies, and

genotype levels (10, 14–16, 18, 31, 36, 37, 42–44, 89–91, 93,94; Maggi et al., presented at

the 21st Meeting of the American Society for Rickettsiology, Colorado Springs, CO,

2007). Bartonella vinsonii subsp. berkhoffii can be separated into four distinct genotypes

based upon 16S-23S ITS sequences (90).

SEROLOGIC TESTS Back to top

Due to difficulties with traditional culture for isolation, serologic testing

for Bartonella infection, including immunofluorescence antibody assay (IFA), enzyme-linked

immunosorbent assay, and Western blotting, has been the cornerstone for clinical diagnosis.

Enzyme-linked immunosorbent assay is the simplest to perform and can be easily

automated, but it has a low sensitivity (17 to 35%). IFA using commercial antigen slides

forB. henselae and B. quintana has become the most frequently used serologic test

worldwide. Human infection with B. henselae or B. quintana is evaluated by detecting the

presence of immunoglobulin M (IgM) and/or IgG antibodies directed against these bacteria.

The first serologic test for CSD was an IFA based on B. henselaebacilli that were cocultivated

with Vero cells to inhibit autoagglutination (115). This test was found to have good

sensitivity (84 to 95%) and specificity (94 to 98%) using sera from patients with CSD

(33, 115, 153). Titers ranging from 64 to 256 or higher are usually indicative of ongoing

infection depending on the differences among laboratory standards; lower titers may indicate

an early or late clinical phase of the disease or prior exposure to the bacteria. In cases of

endocarditis caused by either B. henselae or B. quintana, high IgG antibody titers (≥800) are

usually detected (50). An IgG antibody titer of ≥800 to either B. henselae or B. quintana has

a positive predictive value of 0.810 for the detection of Bartonella infection in the general

population and 0.955 for the detection of Bartonella infections among patients with

endocarditis (50).

Nevertheless, IFA is time-consuming, requires appropriate equipment and expertise, and is

subject to interobserver variation due to the difficulty of reading the test (1). For humans, it

has been postulated that the sensitivity of different IFAs may range from 14 to 100%

depending on the antigen source and cutoff values used by different laboratories

(4, 106, 130, 139–142, 144, 145). Antigenic variability among Bartonellatest strains can

result in false-negative serologic results for some patients (38, 45, 49, 52, 139–

141, 145,149). Cross-reactivity can occur among different Bartonella species. Antigen

adsorption can be used to reduce cross-reactivity and to determine to which antigen the

antibodies are truly directed (1). Cross-reactivity for different Bartonella species can be

present in up to 95% of samples (129). In addition, cross-reactive antibodies to other

pathogens, e.g., Chlamydia pneumoniae, Coxiella burnetii, and spotted-fever

groupRickettsia, are reported (14, 16, 61, 85, 87, 100, 103, 134, 139, 144, 145). Some

studies suggest that genome rearrangement among B. henselae strains may play an

important role in the establishment of persistent infection and can promote antigenic

variation to escape the host immune response (3, 49). In several studies, both B.

henselae DNA and B. vinsonii subsp. berkhoffii DNA were repeatedly detected in dogs that

were not seroreactive (35, 37, 42, 44). For humans, several reports describe detection

and/or isolation of Bartonellaspp. from seronegative patients

(14, 16, 17, 38, 40, 45, 95, 110, 128, 130). The discrepancy between little or no serologic

detection of Bartonella antibodies and detection of Bartonella DNA or bacterial isolation leads

one to the conclusion that seronegative infection could be more common in animals and

humans than currently recognized.

Dihydrolipoamide succinyltransferase, the TAAs (BadA and Vomps), and other proteins of B.

henselae and B. quintana are immunodominant target proteins potentially useful for

diagnostic immunoblotting (6, 46, 146). However, immunoblot-based serologic tests do not

show clear immunoreactive profiles, diminishing their utility for routine serologic diagnosis

of Bartonella infections.

ANTIMICROBIAL SUSCEPTIBILITIES Back to top

Antimicrobial susceptibility testing can be performed by agar dilution methods using either

blood or chocolate agar or by microdilution techniques using various media supplemented

with blood (126). The Etest (AB Biodisk, Solna, Sweden) can also be used to determine

antibiotic susceptibility (150). Results of susceptibility testing against Bartonella spp. are

summarized in Table 2. Evaluation of susceptibility to antibiotics has been performed either

with cell cultures or with axenic media. These methods can also be used for determining the

bacteriostatic activity of antibiotics. Determination of antibiotic susceptibility in axenic

medium has been carried out both on solid media enriched with 5 to 10% sheep or horse

blood and in liquid media (101, 133). It should be noted that the conditions required to

grow Bartonella during susceptibility testing do not meet standardized criteria established by

the Clinical and Laboratory Standards Institute (CLSI). Bacteria of the genus Bartonella are

susceptible to many antibiotics when grown axenically, including β-lactams, aminoglycosides,

chloramphenicol, tetracyclines, macrolide compounds including telithromycin, rifampin,

fluoroquinolones, and co-trimoxazole (101, 102, 108, 126).

Choice and Use of Antibiotics In Vivo

CSD typically does not respond to antibiotic therapy. Most investigators have observed no or

minimal benefit with antibiotic treatment, whereas anecdotal reports indicate that

ciprofloxacin, rifampin, and co-trimoxazole may be effective (122). Additional efforts are

warranted to establish comparative antimicrobial efficacy when treating immunocompetent

patients with chronic bacteremia.

EVALUATION, INTERPRETATION, AND REPORTING OF

RESULTS Back to top

Diagnosis of Bartonella infection in humans, especially for typical forms of CSD, is mainly

based on serologic data, which is the most cost-effective approach. However, the sensitivity

of IFA ranges from 14 to 100%, depending on the antigen, cutoff, and test procedures used

(4). For endocarditis, the best approach is serological testing and the performance of nucleic

acid amplification methods on cardiac valves. A recent procedure using a single-step

serological assay against Coxiella burnetii and Bartonella species found a sensitivity of 100%

and a positive predictive value of 98% for the diagnosis of Bartonella infection in blood

culture-negative endocarditis (124). New methods of culture in liquid medium are proving

diagnostically useful, as direct isolation from blood or tissues is often unsuccessful despite

detection of Bartonella DNA (10, 14–

16,18, 24, 31, 32, 37, 43, 44, 56, 69, 91, 93, 94, 105, 152; Maggi et al., presented at the

20th ASR Meeting and 5th International Conference on Bartonella as Emerging Pathogens,

Pacific Grove, CA, 2006).