Bartonellosis: an emerging infectious disease of zoonotic importance to animals and human beings

Introduction

Bartonella species are fastidious gram-negative bacteria that are highly adapted to a mammalian reservoir host and within which the bacteria usually cause a long-lasting intraerythrocytic bacteremia.^1–3 These facts are of particular importance to veterinarians, as an increasing number of animal reservoir hosts have been identified for various Bartonella spp. Among numerous other examples, Bartonella henselae has coevolved with cats, Bartonella vinsonii subsp. berkhoffii has coevolved with wild canid, and Bartonella bovis has coevolved with cattle.^1,2 Importantly, the list of reservoir-adapted Bartonella spp. including a large number of rodent species that might be kept as pets, continues to grow exponentially, as new Bartonella spp. are discovered. Before 1990, there were only 2 named Bartonella spp., whereas there are now at least 22 named and numerous unnamed or Candidatus species, based upon deposited GenBank sequences or preliminary reports, respectively. For example, there are now different proposed Candidatus species for members of the genus Bartonella that have coevolved with ground squirrels (Candidatus Bartonella washoensis), gray squirrels (Candidatus Bartonella durdenii), flying squirrels (Candidatus Bartonella volans), and ground hogs (Candidatus Bartonella monaxi).

In the natural reservoir host, chronic bacteremia with a Bartonella spp. can frequently be detected by blood culture or polymerase chain reaction (PCR) in outwardly healthy individuals.^1–3 In contrast, the diagnostic detection of a Bartonella spp. in a nonreservoir adapted host can be extremely difficult.⁴ Most, although not all diseases caused by Bartonella spp. occur in accidental hosts and these organisms are being increasingly implicated as a cause of zoonotic infections. As discussed in the section on canine bartonellosis, endocarditis has been reported in dogs due to various Bartonella spp., especially B. vinsonii subsp. berkhoffii and less frequently in cats due to B. henselae, and in cows due to B. bovis. In addition to the large number of documented reservoir hosts, an increasing number of arthropod vectors, including biting flies, fleas, keds, lice, sandflies, and ticks have been confirmed or suspected to be associated with the transmission of Bartonella spp. among animal populations.⁵ Considering the diversity of Bartonella spp. and subspecies, the large number of reservoir hosts and the spectrum of arthropod vectors, the clinical and diagnostic challenges posed by Bartonella transmission in nature appear to be much more complex than is currently appreciated in either human or veterinary medicine.^1–3

Until recently, mechanisms that facilitate persistent Bartonella bacteremia in mammals were not well understood. Once an animal is infected by a bite, scratch, or arthropod, Bartonella spp. localize to erythrocytes and endothelial cells, which provides a potentially unique strategy for bacterial persistence within the blood vessels of reservoir or nonreservoir species.³ In vitro infection of human CD34+ progenitor cells with B. henselae suggests that these bacteria are capable of infecting bone marrow cells, which may contribute to ongoing erythrocytic infection.⁶ However, it now seems likely that initial research studies overemphasized the intraerythrocytic localization of Bartonella spp.⁷ Development of a rodent model by Christoph Dehio's research group provided the initial evidence to support the concept that these are primarily endotheliotropic bacteria that can colonize the vasculature and periodically induce a relapsing bacteremia.^3,8,9 Before development of a rodent model, a relapsing pattern of bacteremia was documented in cats experimentally infected with B. henselae by blood transfusion.¹⁰ These observations suggested periodic sequestration in 1 or more tissue sites followed by reentry into the vasculature and a relapsing pattern of bacteremia. Based upon sequential blood culture, some cats had a consistent and predictable pattern of bacteremia, whereas in other cats, bacteremia occurred in a random fashion. Based upon sequential blood culture data, a relapsing pattern of bacteremia may occur in dogs, humans, and other animals infected with B. henselae and potentially other Bartonella spp., but this has not been confirmed. As discussed in the section on diagnosis of bartonellosis, serology and direct PCR from blood samples both lack sensitivity in some patients, and the documentation of bactermia in a patient in which the organism is not consistently present within erythrocytes (and therefore the diagnostic blood sample) also remains problematic. From the bacteria's perspective, nonhemolytic intracellular colonization of erythrocytes and localization within endothelial cells preserve Bartonella organisms for efficient vector transmission, protect Bartonella from the host immune response, facilitate widespread vascular dispersion throughout the tissues of the body, and potentially contribute to decreased antimicrobial efficacy.^1–3 Other in vitro studies indicate that Bartonella spp. can infect professional macrophages, including dendritic cells, microglial cells, monocytes, and tissue macrophages.^3,8,11 Infection of monocyte-macrophage type cells would allow Bartonella spp. to localize to injured tissues as a component of the inflammatory process. In addition, experimental infection of dogs with B. vinsonii subsp. berkhoffii-induced immunosuppression, characterized by sustained suppression of peripheral blood CD8+ lymphocytes, accompanied by an altered cell surface phenotype and an increase in CD4+ lymphocytes in the peripheral lymph nodes.^12,13 Because of the diversity of lesions and disease manifestations recently attributed to these bacteria in dogs and humans, the clinical implications related to intracellular infection, endothelial cell localization, and periodic transport by erythrocytes and macrophages to new sites within the body appear to vary among individual patients; however, these factors remain poorly understood on an experimental basis and in most instances have not been critically investigated in animal or human patients. In addition, the extent to which infection with 1 or more Bartonella spp. contributes to immunosuppression, thereby predisposing patients to opportunistic viral, bacterial, or fungal infections has not been reported.

Zoonotic Implications of Bartonellosis

When a genus of bacteria is discovered or in the case of Bartonella, rediscovered; numerous clinical, microbiologic, and pathologic concepts related to disease causation and microbial pathogenesis are sequentially redefined. Subsequently, the medical relevance of the genus undergoes continued maturation, as knowledge of the organism, the host immune response, diagnostic test sensitivity and specificity, treatment efficacy, and epidemiology expand. Since the early 1990s, a paradigm of discovery, rediscovery, and ongoing biological and medical redefinition of Bartonella spp. as zoonotic human pathogens has clearly applied to this genus. Owing to extensive contact with a spectrum of animal species, veterinary professionals appear to have an occupational risk of infection due to frequent exposure to Bartonella spp.; therefore, these individuals should exercise increased precautions to avoid arthropod bites, arthropod feces (ie, fleas and lice), animal bites or scratches, and direct contact with bodily fluids from sick animals.^14,15 Physicians should be educated as to the large number of Bartonella spp. in nature, the extensive spectrum of animal reservoir hosts, the diversity of confirmed and potential arthropod vectors, current limitations associated with diagnosis and treatment efficacy, and the ecological and evolving medical complexity of these highly evolved intravascular, endotheliotropic bacteria.

In humans, at least 6 and possibly up to 11 species or subspecies of Bartonella are responsible for a wide range of symptoms that can include pathology in multiple organ systems induced by chronic infection with these bacteria.^1,2,16Bartonella bacilliformis, the type species of the genus, is the etiologic agent of Carrion's disease in South America, that is characterized by an acute hemolytic, bacteremic infection called Oroya fever and a chronic vasoproliferative disease called verruga peruana, that is characterized by cutaneous nodular vascular eruptions, grossly similar to bacillary angiomatosis.¹⁷ Humans are considered the reservoir host and infection in animals has not been reported. Bartonella quintana, the agent of trench fever, is one of the etiologic agents of bacillary angiomatosis, a vascular proliferative lesion observed in immunocompromised individuals, most often due to acquired immunodeficiency syndrome.¹⁷ Furthermore, B. quintana has been associated with endocarditis and chronic bacteremia in homeless humans.¹⁸ Humans are also considered the primary reservoir host for B. quintana, which is transmitted by the human body louse during periods of war, famine, and deprivation. Recent observations have expanded historical, epidemiologic, and microbiologic concepts related to this human-adapted Bartonella spp. B. quintana DNA has been amplified from cat fleas¹⁹ and ticks,²⁰ and the organism has been isolated from feral cats,²¹ from dogs with endocarditis,²² and from a cynomolgus monkey (Macaca fascicularis) in a research colony.²³ In addition, B. quintana DNA has been amplified and sequenced from lymph node aspirates obtained from healthy Golden Retrievers and Golden Retrievers with lymphoma.^24,25 It is also likely that a cat bite resulted in transmission of B. quintana to a woman, as the organism was subsequently isolated from the feral cat that induced the bite wound.²¹ Clearly, these observations support much greater ecological diversity for B. quintana than was envisioned a few short years ago when humans were considered the sole reservoir host and lice were considered the sole vector. Human bacillary angiomatosis is also caused by B. henselae, which also is the only Bartonella sp. known to cause peliosis hepatis in dogs²⁶ and humans.²⁷B. henselae is the primary, if not the sole cause of cat scratch disease (CSD).²⁸ Based upon recent advances in our knowledge of the zoonotic potential of members of the genus Bartonella, the designations CSD and cat scratch fever may be most appropriate when defining a specific subset of patients or when considering human disease manifestations from a historical perspective. Because CSD generally denotes a self-limiting illness characterized by fever and lymphadenopathy and because the recognized spectrum of human disease manifestations associated with Bartonella infections (which may not include fever or lymphadenopathy) has expanded considerably in recent years, it is becoming more obvious that the designation CSD for all human Bartonella-induced illnesses lacks clinical, microbiologic, and zoonotic utility.^1,2,16 Currently, most physicians consider bartonellosis and CSD as synonymous terms, which then inaccurately implies that all B. henselae infections are self-limiting. Although cats are a major reservoir for B. henselae and potentially Bartonella clarridgeiae, some patients deny the possibility of a cat scratch or bite wound, or indicate no contact with cats. Transmission from environmental sources, arthropod vectors, or other animal hosts is probable and the more inclusive term human bartonellosis may facilitate enhanced future understanding of the epidemiology and ecology of diseases caused by members of the genus Bartonella. In addition, recent findings indicate that B. henselae and B. vinsonii subsp. berkhoffii cause persistent intravascular infection accompanied by fatigue, arthritis, neurologic, or neurocognitive abnormalities in immunocompetent humans.^14,15 Although causation remains unconfirmed due to the self-selected, biased populations studied to date, which were mostly composed of veterinary professionals, extensive animal contact, and arthropod exposure may represent risk factors for B. henselae and B. vinsonii subsp. berkhoffii infections.¹⁴ As feline and canine species are the respective reservoir hosts for these bacteria, veterinary professionals are frequently exposed to these bacteria due to contact with blood and other body fluids, arthropod feces, and at times, arthropod vectors. These exposures are more substantial when dealing with neglected, feral, or rescued pet populations, and with wildlife populations. For example, B. henselae, B. quintana, B. koehlerae, and B. clarridgeiae have been detected by molecular methods in cat fleas, thereby suggesting their possible role as vectors for these organisms to reservoir hosts and the potential for inadvertent transmission to nonreservoir hosts.¹⁹

There is increasing evidence to support an important role for B. vinsonii subsp. berkhoffii as a human pathogen. B. vinsonii (berkhoffii), originally isolated from a dog with endocarditis, was subsequently isolated from a human endocarditis patient, from immunocompetent veterinary professionals and from a wildlife biologist.^29,30 Recently, we have documented B. vinsonii subsp. berkhoffii infection in 2 family members in each of 2 different families, all of whom were experiencing neurologic problems, including headaches, insomnia, memory loss, or incoordination.^a As improved methods to detect these bacteria have only recently become available, prospective studies are needed to address causation, which can only be implicated by case reports or case series. In 2 previous studies in immunocompetent persons from the United States, there were 8 B. vinsonii subsp. berkhoffii patients with arthritis, fatigue, neurologic, or neurocognitive abnormalities, with a mean age of 47 years, in whom the organism was isolated or sequenced from blood cultures.^14,15 All of these individuals had extensive animal contact and arthropod exposure as potential risk factors. Four of these 8 infected individuals were coinfected with B. henselae and B. vinsoniisubsp. berkhoffii, of which genotype II was sequenced from all but 1 person, who was infected with a genotype I strain. B. vinsonii subsp. berkhoffii genotype II was also isolated using a specialized blood culture approach described in a subsequent section and was subsequently sequenced from the biopsy of an epitheloid hemangioendothelioma liver tumor obtained from a boy.³¹ Similarly, B. vinsonii subsp. berkhoffii genotype II was cultured from the blood and retrospectively amplified from paraffin-embedded neoplastic tissues obtained from a dog with a cutaneous hemangiopericytoma.³¹ To date, genotype II has been the most frequent of the 4 genotypes to be found in dogs and humans.³² The extent to which dogs can serve as a reservoir host for B. vinsonii subsp. berkhoffii or other Bartonella spp., such as B. henselae, B. clarridgeiae, or B. elizabethae is poorly characterized. Although dogs have been implicated in the direct transmission of B. henselae to humans by a scratch or bite,^33,34 this mode of transmission for B. henselae or B. vinsonii subsp. berkhoffii is as yet, poorly established. Recently, DNA from several Bartonella spp. was amplified and sequenced from the saliva of healthy and sick dogs.^25,35B. henselae and B. vinsonii subsp. berkhoffii have also been isolated or PCR amplified and sequenced from pleural, pericardial, and abdominal effusions as well as from joint and seroma fluids.^36,37 Historically most of these types of fluids, including transudates, modified transudates, and chylous effusions have been consider sterile. It is possible that Bartonella spp. are opportunistic invaders when abnormal fluid accumulation occurs due to any cause or alternatively the bacteria may be a cause or function as a cofactor in the development of various effusions. Future studies are needed to define the role of Bartonella spp. as a cause of effusive disease in dogs and other animals. Regardless, these findings until further clarified suggest that veterinary professionals should avoid needle sticks and cuts and should limit direct contact with saliva and other body fluids from cats, dogs, and other animals. Recently, we have investigated a veterinarian who most likely became infected due to a needle stick while aspirating a cutaneous neoplasm^a in a dog.

Bartonella spp. have been implicated as a cause of endocarditis in cats, cows, dogs, human beings, and sea otters (Table 1). B. vinsonii subsp. berkhoffii and B. henselae have been associated with endocarditis in dogs^29,38–41 and humans,⁴² whereas Bartonella vinsonii subsp. arupensis,⁴³B. koehlerae,⁴⁴Bartonella alsatica,⁴⁵ and B. elizabethae,⁴⁶ which are mouse, cat, rabbit, and rat-adapted Bartonella spp., respectively, have only been associated with human endocarditis to date. B. clarridgeiae has been associated with endocarditis in dogs^47,48 and (Candidatus B. washoensis) (proposed species) for which the ground squirrel is the reservoir, has been associated with a human case of myocarditis⁴⁹ and was isolated from a dog with endocarditis.⁵⁰Bartonella grahamii, a rodent Bartonella spp., has been associated with human cases of neuroretinitis and bilateral retinal artery branch occlusions.^51,52B. vinsonii subsp. arupensis, which is found in white-footed mice (Peromyscus leucopus), the reservoir for Borrelia burgdorferi, was isolated from a bacteremic rancher with fever and mild neurologic symptoms.⁵³ Most recently, Bartonella tamiae (rodent reservoir suspected) was isolated from 3 patients in Thailand with fatigue, fever, and headache,⁵⁴Bartonella rochalimae (fox reservoir) was isolated from a woman with bactermia, fever, and splenomegaly.⁵⁵ In California, B. rochalimae has caused endocarditis in a dog and has been isolated from dogs and gray foxes.⁵⁶B. rochalimae has also been reported to infect dogs in Europe.⁵⁷ Endocarditis and granulomatous lymphadenitis, due to infection with B. alsatica, have been reported in humans in France after hunting or butchering wild rabbits.⁴⁵Bartonella melophagi (sheep reservoir) was isolated from a woman with historical pericarditis, fatigue, and muscle weakness and from another woman with chronic fatigue and myalgia, joint pain, headaches, and neurocognitive abnormalities.⁵⁸ In addition, we have recently identified a potentially novel species closely related with B. volans^b in both dog and human blood samples from the southeastern United States. Collectively, these reports and observations emphasize the concept that inadvertent transmission of known or currently uncharacterized Bartonella spp. from wild animals, domestic animals, and pets occurs in nature. The extent to which these infections contribute to self-limiting infections or to chronic arthritis, fatigue, myalgia, or endocarditis and a spectrum of other disease manifestations in dogs and humans remains unclear. However, it is increasingly obvious that the dog is a natural model for human bartonellosis and vice versa.⁴⁸

Feline Bartonellosis

Cats can be infected with or serve as reservoirs for B. henselae, B. clarridgeiae, B. koehlerae, B. quintana, and B. bovis.^1,2 Fleas are responsible for the transmission of B. henselae, B. clarridgeiae, and potentially B. koehlerae among cats.^59,60B. bovis has only been isolated from a few cats (was originally reported with the proposed name Bartonella weissii) and domestic ruminants are clearly the reservoir hosts for this species.^1,2 The extent to which members of the genus Bartonella are pathogenic for cats remains unclear. As is the case for dogs infected with B. vinsonii subsp. berkhoffiii, cats can develop endocarditis in association with B. henselae infection, despite the fact that cats are the primary reservoir hosts for this Bartonella spp.^61,62 Recently, we isolated B. vinsonii subsp. berkhoffii from a cat with osteomyelitis in a carpal joint that developed 18 months after amputation of another osteomyelitis lesion that was located in a digit in the rear leg. Following successful blood culture using BAPGM (Bartonella/alpha-Proteobacteria growth medium), B. vinsonii subsp. berkhoffii-specific DNA sequences were retrospectively amplified from the paraffin block containing the original osteomyletis lesion, 18 months after digital amputation.⁶³ Detection of the same subspecies and genotype by specialized blood culture when osteomyelitis developed in a carpal joint and in the amputated digit 18 months earlier, support long-term, minimally clinical, bacteremic infection with localization to bone and joints. Current evidence derived from case studies involving cats, dogs, and human patients supports the concept that Bartonella spp. induce long-term, nonclinical, or minimally symptomatic infections and thereby behave as stealth pathogens.^64,65 Also documentation of B. vinsonii subsp. berkhoffii in a cat suggests that cats, like dogs and human patients, might be more likely to develop clinical disease manifestations when infected with a non–reservoir-adapted Bartonella spp. Also, as discussed in the section on diagnosis of bartonellosis, diagnostic detection of infection in an accidental host is technically more challenging then detection of infection with the same Bartonella sp. in a reservoir-adapted host. For example, B. henselae bacteremia can be documented by conventional blood culture in 25–41% of healthy cats in different regions throughout the world (Table 2).^1,2 Owing to the high prevalence of nonclinical infection among various feline populations; it is obvious that cats, fleas, and B. henselae share a long and highly adapted evolutionary relationship.^1–3 In a recent study from Korea involving 151 cats, B. henselae was detected in feral cat blood (41.8%), saliva (44.1%), and nail (42.7%) samples by PCR amplification.³⁵ A similar prevalence of B. henselae infection was also detected in pet cat blood (33.3%), saliva (43.5%), and nail (29.5%) samples. Feral cats (n=51) in Alabama and Florida commonly had B. henselae or B. clarridgeiae DNA amplified from blood (56.9%), skin biopsies (31.4%), gingival swabs (17.6%), and claw bed (17.6%) swabs.⁶⁶ In the United States, DNA of B. henselae or B. clarridgeaie were amplified from the blood of 92 cats (47.8%) or the fleas taken from these cats (59.8%) and concurrent infection of cats and their fleas with both organisms was common.⁶⁷ Following experimental infection of cats with B. henselae by blood transfusion, self-limiting febrile illness of 48–72 hours duration, mild to moderate transient anemia and transient neurologic dysfunction were reported in a subset of cats.^68–71 In 1 experimental study, cats inoculated with B. henselae IV remained healthy, but in cats infected by common exposure to Ctenocephalides felis, fever developed in 3 of 6 cats and myocarditis in 1 cat, suggesting passage in culture decreased pathogenicity and infection via the natural vector may have affected disease pathogenesis.⁷² Following experimental infection with B. henselae, reproductive efficiency was decreased in both male and female cats.⁷³ Based upon clinical experiences, self-limiting fever can also occur in B. henselae bacteremic cats following minor surgical procedures. Although unproven, and potentially similar to other chronic intravascular vector-borne infections, it is likely that stress, such as surgery or trauma, can induce transient disease manifestations in cats, including self-limiting fever, mild anemia, and neurologic dysfunction. Owing to the high percentage of chronically bacteremic healthy cats in the United States, establishing a cause and effect relationship between disease manifestations and bacteremia in cats may require large epidemiologic studies involving Bartonella endemic and nonendemic regions. Based on a seroepidemiologic study from Switzerland, Bartonella-infected cats are more likely to have kidney disease and urinary tract infections, stomatitis, and lymphadenopathy.⁷⁴ Although speculative, this study suggests that cats may pay a biological price for long-term intravascular infection. However, these associations have not been consistently found in cats from flea-infested endemic regions of the United States. Seroepidemiologic studies have generated contrasting results as to whether fever,⁷⁵ lymphadenopathy,⁷⁴ stomatitis, and gingivitis⁷⁶ are caused by B. henselae. In febrile cats, antibody tests were not predictive as to whether fever was due to Bartonella spp. infection and antibody prevalence in this population did not correlate with PCR results; therefore, antibody detection by ELISA or western blot should not be used to determine the Bartonella spp. infection status.⁷⁵ In another study, detection of B. henselae antibodies by ELISA and western blot and PCR amplification of Bartonella spp. DNA from blood was not statistically correlated with gingivitis, stomatitis, or glossitis in colony of rescued cats from southern California. These cats had been allowed to comingle for 3–15 years (mean 8.2 y).⁷⁶ A limitation of this study was the relatively small population of cats, with 9 having oral disease and 36 not having oral disease. In a study involving 298 cats presenting to a tertiary referral hospital at the University of California at Davis, isolation of B. henselae or B. clarridgeiae, but not serologic status, was statistically associated with stomatitis.^c In contrast, a study of pair-matched samples from cats with and without stomatitis from around the United States failed to find a correlation between Bartonella spp. antibody test results and PCR assay results and the presence of stomatitis.⁷⁷ While the 70 cats with stomatitis commonly had Bartonella spp. antibodies in serum (53.6%) and Bartonella spp. DNA in blood (11.4%), Bartonella DNA was amplified from the gingival tissues of only 3 of 34 cats (8.8%), for which tissue biopsies submitted. In a naturally infected cat with anterior uveitis, B. henselae intraocular antibody production and response to doxycycline therapy was demonstrated after other known causes of uveitis were excluded.⁷⁸ In a subsequent study, cats with uveitis and experimentally infected cats without uveitis had B. henselae DNA amplified from the aqueous humor; however, only cats with uveitis had intraocular production of antibodies against B. henselae suggesting that this test may have diagnostic utility in cats with suspected B. henselae-induced uveitis. In another study involving cats with uveitis, the authors documented serum antibody production against Bartonella spp. using western blot immunoassays, but evidence of active infection (blood culture or PCR) was not examined and nonuveitis controls were not tested.⁷⁹ However in a more recent study that compared 113 cats with uveitis to 42 cats without ocular disease from low flea-risk states, 65 were from high flea risk-states, and 7 cats for which the state was unknown, there was no difference in seroprevalence rates or titer magnitude between cats with uveitis and cats with nonocular diseases.⁸⁰ The median B. henselae titer was 1:64 for all groups and healthy cats were more likely to have higher titers than cats with uveitis or cats with nonocular disease. These results indicate that serum antibody tests alone cannot be used to document that clinical uveitis is associated with Bartonella spp. infection.

The extent to which infection with a Bartonella spp. induces neurologic disease in a cat, dog, or human being is not understood. In an initial study involving cats with seizures (n=63) and cats with other neurologic diseases (n=82), there was no difference in B. henselae seroprevalence rates between cats with seizures and cats with other neurologic diseases when compared with cats with nonneurologic illness (n=163) or compared with a healthy cat population (n=97).⁸¹ Seroprevalence rates were very high in all 3 groups, including seroreactivity in 50% of the cats with neurologic disease, 64% of cats with nonneurologic illnesses, and 70% of the healthy cat population.⁸¹ In a subsequent retrospective study of 100 client-owned cats with neurologic disease, serum and CSF were assayed for Bartonella spp. IgG antibodies and CSF was assayed for Bartonella spp. DNA.⁸²B. henselae IgG antibodies were detected in serum from 36 cats, Bartonella sp. C values >1 (suggesting intrathecal antibody production) were detected in CSF of 11 cats, and B. henselae DNA was amplified from the CSF of 10 cats. The results of this study, in conjunction with the development of neurologic signs in some cats following experimental inoculation of B. henselae warrant additional prospective evaluation of the association of Bartonella spp. with feline CNS disease. Treatment trials may be necessary to implicate infection with Bartonella spp. as a cause or cofactor in cats with neurologic disease. In conjunction with additional epidemiologic studies, documentation of active infection by blood or CSF culture, accompanied by therapeutic elimination of bacteremia in conjunction with resolution or marked improvement in neurologic abnormalities would be supportive of a cause and effect role for Bartonella spp. in feline CNS disease.

Based upon data generated following experimental infection of cats, both humoral- and cell-mediated immunity play a role in the elimination or suppression and control of B. henselae, B. clarridgeiae, and B. koehlerae infection in cats.^83,84 In one study, activation of cell-mediated immune responses, including interferon-γ and tumor necrosis factor-α production, played an important role in elimination of B. henselae from bacteremic cats.⁸³ In an infection and challenge study, cross-protection was demonstrated as evidenced by an absence of bacteremia in cats primarily infected with B. henselae type I and subsequently challenged with B. henselae type II, whereas no cross-protection was found for cats primarily infected with B. henselae type II and subsequently challenged with B. henselae type I.⁸⁵ In the same study, cats primarily infected with B. henselae type II and challenged with B. clarridgeiae were more likely to have relapses for both primary and secondary infections. Conclusions from this study indicate that cats infected with the same Bartonella spp. and strain are less likely to become bacteremic, whereas cats challenged with a different strain of the same species or different species readily become bacteremic, although the duration of bacteremia was generally shorter than during a primary challenge, when using needle inoculation as the route of infection. Following experimental needle inoculation, the duration of B. koehlerae bacteremia was shorter (mean 74 d) than for cats inoculated with B. clarridgeiae (mean 324 d) and a relapsing pattern of bacteremia was not detected.^85,86 Although instructive and useful in the theoretical design of vaccines, these experimental infection studies should be interpreted with caution, as naturally infected cats can remain bacteremic for years, suggesting that natural immunity may not eliminate infection in an as yet, undefined subset of cats. Immunosuppression, associated with feline leukemia virus or feline immunodeficiency virus, may increase the pathogenicity of B. henselae infection in cats, with coinfection contributing to an increased prevalence of stomatitis; however, this association has not been investigated experimentally in controlled infection studies.⁸⁷

In addition to the development of clinical abnormalities including fever, lymphadenopathy, mild neurologic signs, and reproductive disorders, cats experimentally infected with B. henselae by blood transfusion also developed histologic changes in various tissues.⁷⁰ These cats were infected with B. henselae and B. clarridgeiae by blood transfusion from donor cats that had induced CSD or neuroretinitis in their respective owners; therefore, cotransmission of other infectious agents cannot be discounted. Gross necropsy results are unremarkable; however, histopathologic lesions included peripheral lymph node hyperplasia, splenic follicular hyperplasia, lymphocytic cholangitis/pericholangitis, lymphocytic hepatitis, lymphoplasmacytic myocarditis, and interstitial lymphocytic nephritis. These findings also seem to support a predominantly cell-mediated immunologic response following B. henselae infection in cats. However this ineffective immunologic response appears to suppress, but not eliminate the infection in many naturally infected cats resulting in chronic infection accompanied by a relapsing pattern of bacteremia. In 2006, the Academy of Feline Medicine Advisory Panel in association with the American Association of Feline Practitioners generated a detailed report on diagnosis, treatment, and prevention of Bartonella spp. infections.⁸⁴ The panel did not recommend treatment of healthy B. henselae seroreactive cats or pursuing diagnostic testing to determine if a healthy cat was exposed (serology) or actively infected (blood culture or PCR). However, documentation of bacteremia (positive blood culture or PCR) in a cat with unexplained illness or granulomatous, lymphocytic or lymphoplasmacytic inflammation would warrant consideration of antimicrobial treatment.

A number of epidemiologic studies and 1 experimental study have investigated potential associations between Toxoplasma gondii and B. henselae exposure in cat populations from various geographic regions.^88–93 Often very similar seroprevalence rates are found for both organisms.^88–90 However, more recent studies have failed to find a correlation between T. gondii and B. henselae seroreactivity in cat populations from Pennsylvania and Grenada.^91,92 In the experimental study, cats were first inoculated with T. gondii followed by sequential inoculation with B. henselae and feline herpesvirus-1, but did not develop significant clinical illness.⁹³ Coinfection with Bartonella spp. and feline herpesvirus-1 was also documented in client-owned cats.⁷⁶ It is clear that the cat is the primary reservoir host for each of these organisms and that evolutionary adaptation has facilitated the ability of overtly healthy cats to maintain long-term, nonclinical infections with both organisms. However, retroviral infection, drug-induced immunosuppression, or immune senescence associated with advanced age could result in circumstances in which coinfection with B. henselae and T. gondii contributes to clinical abnormalities in cats, despite the coevolutionary adaptation to these organisms. Therefore, serologic evidence of prior exposure to both organisms in a sick cat, in conjunction with diagnostic evidence of active B. henselae infection by blood culture or PCR would warrant consideration of treatment, particularly in very young or geriatric cats with evidence of systemic illness.

Although technically challenging, long-term sequential studies maybe necessary to determine if cats pay a biological price, such as a tendency to develop gingivostomatitis or renal disease, when chronically infected with B. henselae, B. clarridgeiae, or B. koehlerae. If these Bartonella spp. prove to be highly evolved stealth pathogens in cats, this would support future efforts to develop and utilize Bartonella vaccines and alter current recommendations to not treat bacteremic healthy cats in an effort to eliminate infection. The recommendation to not treat healthy cats is based upon the high prevalence of bacteremia in pet and feral cat populations in flea endemic regions, the questionable pathogenic role of these bacteria, and the risk of inducing bacterial resistance. Efforts to minimize exposure to fleas by not allowing pet cats to roam and the routine use of acaracides on cats to prevent flea infestations are recommended.⁸⁴ As discussed in the section on diagnostic testing, diagnostic use of specialized culture media may be necessary to detect infection with nonreservoir adapted Bartonella spp. that induce disease in cats, an approach that may further enhance our understanding of feline bartonellosis.

Canine Bartonellosis

Infections with Bartonella spp. are being recognized in dogs with increasing frequency as a result of more sophisticated diagnostic techniques.^36,37,94,95Bartonella infections in dogs, as in humans, exhibit a wide range of clinical signs, relating to involvement of many organ systems (Table 3). This is perhaps due to the endotheliotropic nature of these bacteria and the fact that intracellular organisms located within the vasculature can become redistributed throughout the body by transport in erythrocytes and macrophages. As in humans, lesions may occur in the CNS,^96,97 eye,⁹⁸ nasal cavity,⁹⁹ endocardium,¹⁰⁰ myocardium,¹⁰¹ liver,^26,102 lymph nodes,^24,99 joints,^37,97,103 skin and subcutaneous tissues.^96,97,99 The clinical spectrum of Bartonella infection in dogs appears to be highly variable, ranging from chronic subclinical infections to protracted illnesses accompanied by lethargy and weight loss, as the only reported abnormalities. To date, B. vinsonii subsp. berkhoffii, B. henselae, B. clarridgeiae (Candidatus B. washoensis), B. quintana, B. rochalimae, and B. elizabethae have been recognized as pathogenic for dogs.^1,2B. vinsonii subsp. berkhoffii was the first Bartonella sp. to be isolated from a dog with endocarditis in our laboratory in 1993.^100,103,104 Retrospectively, long-term administration of immunosuppressive doses of corticosteroids for a presumptive diagnosis of systemic lupus erythematosus (SLE) (antinuclear antibody+) may have facilitated the isolation of the original type strain of B. vinsonii subsp. berkhoffii from this dog and may have contributed to the ultimate development of endocarditis. Subsequently, a statistical association was found between antinuclear antibody reactivity and seroreactivity to B. vinsonii subsp. berkhoffii, Ehrlichia canis, and Rickettsia rickettsii antigens in dog sera.¹⁰⁵ In 2 recent human case reports, administration of corticosteroids for a diagnosis of Wegners granulomatosis preceded the development of B. henselae endocarditis, as it occurred with B. vinsonii subsp. berkhoffii in the dog described above.^106,107 Collectively, these findings indicate several clinically important factors of potential relevance for an individual patient. First, immunosuppression of a Bartonella-infected dog or human might predispose the individual to the development of endocarditis and other potentially serious sequelae. Second, there can be substantial overlap between the clinical and hematologic manifestations of endocarditis, a localized infection with systemic showering of bacteria, and the autoimmune disease, SLE. Polyarthritis, thrombocytopenia, anemia, glomerulonephritis, and antinuclear antibodies can be documented in dogs with endocarditis, which are hallmark features of SLE. Finally, it is possible that exposure to or chronic infection with multiple vector-borne organisms, such as B. vinsonii subsp. berkhoffii, E. canis, and R. rickettsii predispose to the development of antinuclear antibodies in some dogs.¹⁰⁵ This latter suggestion is not unprecedented, as unmethylated CpG (cytosine linked to a guanine by a phosphate bond) motifs in bacterial DNA are very similar to CpG motifs in mammalian DNA (dogs and human beings); therefore, antibodies directed against bacterial CpG motifs can cross react with mammalian nuclear antigens, when performing ANA testing.¹⁰⁸ After the initial blood culture isolation and microbiologic characterization, B. vinsonii subsp. berkhoffii, this bacteria has now been increasingly identified as an important cause of culture-negative endocarditis and has also been associated with cardiac arrhythmias and myocarditis in dogs.^{38–41,109,110} Endocarditis, associated with B. vinsonii subsp. berkhoffii occurs in large breed dogs with a predisposition for involvement of the aortic valve. However, 20% of cases involve the mitral valve or multiple valves. In some dogs, intermittent lameness, bone pain, epistaxis, or fever of unknown origin can precede the diagnosis of endocarditis for several months, whereas other dogs will present with an acute history and rapid cardiopulmonary decompensation. B. vinsonii subsp. berkhoffii infection was diagnosed in 2 dogs that developed acute respiratory distress requiring ventilatory support.¹⁰⁹ Radiographs were not classical for cardiogenic pulmonary edema and additional diagnostic testing ultimately confirmed valvular endocarditis. It appears that acute severe pulmonary edema, requiring ventilatory support and necessitating aggressive antimicrobial therapy, can develop in some dogs infected with B. henselae or B. vinsonii subsp. berkhoffii. In addition to pulmonary edema, cardiac arrhythmias secondary to myocarditis can be detected in dogs without echocardiographic evidence of endocarditis.¹⁰¹

B. vinsonii subsp. berkhoffii and B. henselae can induce granulomatous inflammation in a single tissue or in multiple organ systems.^{99,102,111,112}Bartonella DNA has been amplified from paraffin blocks obtained from dogs with granulomatous paniculitis, accompanied by polyarthritis and meningitis.^96,97Bartonella-induced granulomatous lymphadenitis, involving the left submandibular lymph node, was diagnosed in a dog on the basis of seroreactivity to B. vinsonii subsp. berkhoffii antigens, visualization of Warthin-Starry silver staining bacteria within the lymph node and PCR amplification followed by Southern blot hybridization.⁹⁹ Seven days before enlargement of the lymph node, the owners removed an engorged tick from the dog's left ear. This dog's case report provides the best current clinical evidence to date that ticks can transmit Bartonella spp. to dogs and potentially to humans. Although controversial and unproven, evidence supporting the possibility of tick transmission of Bartonella spp. was recently reviewed.⁵ The granulomatous lymphadenitis that developed in this dog following tick attachment might be analogous to acute bartonellosis (CSD) in humans, where a cat scratch or bite injects the inocula (usually B. henselae), as compared with inoculation by the bite of a tick. More recently, B. vinsonii subsp. berkhoffii was detected in a dog with fatal, disseminated granulomatous disease involving the spleen, heart, lymph nodes, omentum, liver, kidney, lung, mediastinum, and salivary glands.¹¹¹B. henselae has been amplified and sequenced from the lymph nodes of dogs with generalized granulomatous lymphadentis,¹¹² and from the liver tissues from a dog with granulomatous hepatitis.¹⁰² These case reports predated knowledge of the possibility of Bartonella DNA carryover during necropsy and tissue processing.¹¹³ As lymph node and liver biopsy samples were obtained aseptically during surgery in these cases, the only potential source of DNA carryover for these tissues was during processing or while cutting sections on a microtome for DNA extraction, before PCR, which although perhaps unlikely due to the documented association of Bartonella spp. with granulomatous disease in dogs and humans, the possibility of DNA carryover can not be ruled out. As chronic bacteremia within various reservoir host populations (cats, cows, and rodents) can be extremely high (50–100%), Bartonella spp. DNA can be inadvertently transferred among patient samples during processing in the necropsy area or in histopathology laboratories, particularly if precautions are not taken.¹¹³

Recently, B. vinsonii subsp. berkhoffii was isolated by BAPGM blood culture from a dog treated with immunosuppressive doses of prednisone and azithiaprine for pancytopenia, which apparently predisposed to the development of bacillary angiomatosis.¹¹⁴ Efforts to document concurrent infection with B. henselae or B. quintana in this dog, which are known causes of bacillary angiomatosis in human patients, were not successful. Similar to bacillary angiomatosis in human immunodeficiency virus-infected humans,¹⁷ organisms were visualized in the skin lesions using a silver stain, B. vinsonii subsp. berkhoffii DNA was amplified directly from the bacillary angiomatosis lesions and as mentioned above, the organism was obtained by blood culture at a different diagnostic time point. Following antimicrobial treatment, the bacillary angiomatosis lesions resolved, as did the pancytopenia, which was most likely secondary to hypersplenism. This case provides another example of occult Bartonella infection in which the dog developed a prototypical lesion (bacillary angiomatosis) following immunosuppressive drug therapy. Although perhaps an infrequent occurrence, clinicians must be cautious when using immunosuppressive drugs in patients with occult bartonellosis.

Based on serologic evidence, B. vinsonii subsp. berkhoffii and B. henselae may contribute to the development of dermatologic lesions indicative of a cutaneous vasculitis, anterior uveitis, polyarthritis, meningoencephalitis, immune-mediated hemolytic anemia, and immune-mediated thrombocytopenia.^{103,115–117} Lameness and neutrophilic polyarthritis are among the more prevalent abnormalities in dogs diagnosed with Bartonella infection.^{103,115–117} Splenomegaly was also statistically associated with Bartonella seroreactivity in 1 of these studies.¹⁰³ Coinfection with B. henselae and B. vinsonii subsp. berkhoffii was documented on multiple occasions by blood and synovial fluid culture using BAPGM in a dog that was initially examined due to lameness, found to be thrombocytopenic and ultimately shown to have neutrophilic polyarthritis.³⁷ This dog failed repeated and prolonged courses of azithromycin and marbofloxacin during which time the joint disease progressed from a nonerosive to erosive polyarthritis, which was characterized by severe joint laxity and instability resulting in deformity of the distal limbs.³⁷ A clinical or cytologic diagnosis of neutrophilic polyarthritis has also been made in dogs preceding the diagnosis of Bartonella endocarditis.^29,38 Bartonellosis should be considered in the differential diagnoses of dogs presenting for unexplained lameness or neutrophilic polyarthritis. Future studies should focus on the frequency in which these bacteria induce lameness by direct intraarticular infection as compared with indirect immune-mediated damage. As Bartonella spp. appear to be well-adapted organisms or stealth pathogens,⁶² it is possible that these bacteria can induce very chronic, insidious, and slowly progressive joint damage. Additional research efforts, using carefully designed case controlled studies will be necessary to establish the frequency and extent to which Bartonella spp. contribute to ocular, orthopedic, neurologic, cardiovascular, and hematologic abnormalities in dogs.

Recently, both B. vinsonii subsp. berkhoffii and B. henselae have been isolated in our laboratory from transudates, modified transudates and chylous effusions obtained from the thorax or abdomen of dogs with idiopathic cavitary effusions.³⁶B. vinsonii subsp. berkhoffii and B. henselae were also isolated in BAPGM from blood and a massively enlarged seroma after an acute traumatic injury.³⁷ Although a causative role for Bartonella in the development of edema, seroma, or effusive disease has not been established, the presence of viable bacteria in these extravascular fluid compartments could complicate the clinical management of dogs with effusive disease, particularly if immunosuppressive drug therapy is used.

Owing to the relatively recent recognition that dogs can be infected with B. vinsonii subsp. berkhoffii, B. henselae, and potentially other Bartonella spp., seroprevalence data to various Bartonella spp. remains somewhat limited.^1,115–119 As B. vinsonii subsp. berkhoffii was the first documented Bartonella sp. to cause disease in dogs, there is more seroepidemiologic data for this organism. As discussed in the section on canine bartonellosis, for several years B. henselae was thought to infect only cats; therefore, this organism was rarely included in early canine serologic surveys. Seroprevalence by immunofluorescent antibody (IFA) testing, using B. vinsonii subsp. berkhoffii as the antigen source, was determined in 1,920 sick dogs from North Carolina or surrounding states that were evaluated at a tertiary veterinary teaching hospital.¹¹⁵ Using a reciprocal titer of >32, only 3.6% of sick dogs had antibodies to B. vinsonii subsp. berkhoffii. Risk factors that could be associated with seroreactivity included: heavy tick exposure (odds ratio [OR] 14.2), cattle exposure (OR 9.3), rural versus urban environment (OR 7.1), and heavy flea exposure (OR 5.6). These data were interpreted to support the possibility that exposure to B. vinsonii subsp. berkhoffii was more likely in dogs in rural environments that were allowed to roam. In addition, exposed dogs were likely to have a history of heavy tick infestation. Using sera from dogs experimentally infected with R. rickettsii or E. canis, cross reactivity to Bartonella antigens was not detected.¹¹⁵ However, 36% of serum samples derived from dogs naturally infected with E. canis from the southeastern United States were also reactive to B. vinsonii subsp. berkhoffii antigens. As E. canis is transmitted by Rhipicephalus sanguineous, this tick may be involved in the transmission of B. vinsonii subsp. berkhoffii or exposure to this tick is indirectly related to transmission of this bacteria to dogs. The possibility of tick transmission was further supported by 2 additional studies from the same geographic region involving dogs infected with 1 or more Ehrlichia spp., in which seroreactivity to B. vinsonii subsp. berkhoffii antigens was 30% and 89%, respectively.^120,121 In contrast antibodies to B. vinsonii subsp. berkhoffii were not detected in our laboratory in sera from dogs from Canada, using the same indirect immunofluoresent antibody testing platform.¹²² Importantly, the only case of B. vinsonii subsp. berkhoffii genotype IV infection reported to date was in a dog with endocarditis from Canada.⁴⁰ We have also detected genotype IV in a dog with endocarditis from Colorado, suggesting that this genotype may be prevalent in the western United States and Canada. Most likely this distinct genotype has a unique and as yet uncharacterized reservoir host and undetermined mode of transmission to dogs. Seroprevalence, using B. vinsonii subsp. berkhoffii antigens, was 10% (4/40 dogs) in dogs with suspected tick-borne illness from Israel and 36% in dogs with fever and thrombocytopenia from Thailand, many of which had concurrent E. canis antibodies.^123,124 These results suggest that dogs with ehrlichiosis during the Vietnam War may have been concurrently infected with B. vinsonii subsp. berkhoffii. As discussed in the section on canine bartonellosis, the advent of bartonellosis will require some reassessment of ehrlichiosis in dogs. Using an ELISA assay, 35% of 869 samples, derived from coyotes in California, contained antibodies to B. vinsonii subsp. berkhoffii antigens.¹²⁵ As coyotes have spread across much of North America in recent years, these animals may serve as a major natural reservoir for tick vectors and potentially could transmit B. vinsonii subsp. berkhoffii directly to pets or human beings via bites or scratches. In Morocco, the overall B. vinsonii subsp. berkhoffii seroprevalence was 38% (56/147 dogs tested).¹²⁵ Most of the seropositive dogs were stray dogs from Rabat (8/22 [36%]) and Khenifra (47/101 [47%]). In contrast, antibodies against B. vinsonii subsp. berkhoffii antigens were found infrequently among pet dogs from Rabat (1/24 [4%]), again suggesting that dogs with more extensive vector exposure are at risk for infection. B. vinsonii subsp. berkhoffii was also recently isolated from 2 dogs from China.¹²⁶ Based upon PCR amplification and DNA sequencing, B. vinsonii subsp. berkhoffii, B. rochalimae, and a novel Bartonella sp. were detected in blood samples from dogs in Greece and Italy.⁵⁷ Current data indicate that exposure to B. vinsonii subsp. berkhoffii can be found throughout much of the United States and most tropical and subtropical regions of the world and that dogs exposed to E. canis in some regions are more likely to have been exposed to B. vinsonii subsp. berkhoffii as well.

The extent to which infection with B. vinsonii subsp. berkhoffii influences the pathophysiology of ehrlichiosis, a disease of much longer historical venue, deserves critical reappraisal. For example, infection with B. vinsonii subsp. berkhoffii in dogs concurrently infected with E. canis may contribute to the tendency to develop epistaxis.¹²⁷ Coinfection with these 2 organisms is also of clinical relevance, because doxycycline, the preferred drug for treatment of ehrlichiosis, does not appear to effectively eliminate B. vinsonii subsp. berkhoffii infection in many treated dogs, despite administration of prolonged courses of doxycyline for ehrlichiosis. If R. sanguineus is involved in the transmission of B. vinsonii subsp. berkhoffii among dogs, there appears to be important, and incompletely understood differences in the prevalence of E. canis and B. vinsonii subsp. berkhoffii among tick populations found in various regions of the world. In some tropical and subtropical regions, such as Grenada and southern Brazil, exposure to B. vinsonii subsp. berkhoffii is infrequent, despite high E. canis seroprevalence rates and frequent infestations with R. sanguineus.^90,128 Of similar concern in both canine and human medicine is the finding of cosegregation of B. burgdorferi, Anaplasma phagocytophilum (previously Ehrlichia equi or human granulocytic ehrlichiosis), Babesia microti, and B. vinsonii subsp. arupensis in Ixodes scapularis ticks in the northeastern, northcentral, and northwestern United States. Evolving evidence is increasingly supportive of transmission of B. henselae and potentially other Bartonella spp. by I. scapularis, Ixodes pacificus, and Ixodes ricinus.⁵ The possibility of tick transmission was also supported by studies from California that found a high prevalence of B. vinsonii subsp. berkhoffii bacteremia in coyotes and DNA of the organism in questing I. pacificus in the same region.^125,129 Regional differences in tick species, accompanied by differences in the bacterial, viral, and protozoan organisms that ticks transmit creates substantial challenges for the veterinary clinician in regard to diagnosis and medical management of these vector-borne organisms.

Based upon existing evidence, B. vinsonii subsp. berkhoffii has been considered the most frequent Bartonella spp. that causes disease in dogs. However, this conclusion appears to be inaccurate, as sera from dogs have generally not been screened systematically against a large panel of Bartonella spp. antigens and limited PCR testing or optimized blood cultures have been performed. In most regions, the B. henselae seroprevalence in dogs exceeds B. vinsonii subsp. berkhoffii seroprevalence. Studies from Hawaii,¹³⁰ the United Kingdom,¹³¹ Japan,¹³² and the communal lands of Zimbabwe¹³³ identified B. henselae seroprevalences of 6.5% (2/31 dogs), 3.0% (3/100 dogs), 7.7% (4/52 dogs), and 14% (32/228 dogs), respectively.^18,19,130 In our laboratory, B. henselae seroprevalence was 10% and 28% in healthy and sick dog populations from the southeastern United States and 17% in dogs from Spain, respectively.^134,135 In Italy, B. henselae seroprevalence was 6% (23/381) in dogs from northern Italy and 28% (58/205) in the Sassari district of Sardinia.^136,137 In the study from the Sassari district, there was no statistical difference in the B. henselae seroprevalence in cats as compared with dogs. In a study from Thailand, B. clarridgeiae was isolated from 1 of 296 stray and 54 pet dogs.¹¹⁹ Evolving results suggest that dogs around the world can be infected with B. henselae and B. clarridgeiae. In a study from Korea involving 54 pet dogs, B. henselae DNA was amplified from blood (16.6%), saliva (18.5%), and nail (29.6%) samples, whereas B. clarridgeiae was isolated from 9 of the 54 dogs and 2 dogs were coinfected with B. henselae and B. clarridgeiae.³⁵B. henselae and B. clarridgeiae have also been isolated from 3 dogs from Africa.¹³⁸ Both B. henselae and B. clarridgeiae are flea-transmitted Bartonella spp., suggesting that dogs may be infected by flea infestation, or alternatively by the bite or scratch of a flea infested or Bartonella-infected cat.

Although the pathogenicity of all Bartonella spp. in dogs is poorly characterized, it is becoming increasingly clear that species other than B. vinsonii subsp. berkhoffii can infect dogs. In fact, B. henselae, not B. vinsonii subsp. berkhoffii is the most frequently isolated or PCR amplified Bartonella spp. detected in sick dogs in the Vector Borne Diseases Diagnostic Laboratory at North Carolina State University. Similar to human reports, B. henselae was amplified and sequenced on 2 independent occasions from the liver of a dog with peliosis hepatis.²⁶ This is a unique pathologic lesion that is induced only by B. henselae infection in humans.¹⁷ Also, B. henselae DNA has been amplified from a dog with granulomatous hepatitis,¹⁰² a histopathologic lesion that is reported with some frequency in children infected with B. henselae.¹⁶ Of pathologic importance, B. clarridgeiae DNA has been amplified and sequenced from the liver of a Doberman Pinscher with copper storage disease and from the aortic valve of 2 dogs with vegetative valvular endocarditis.^21,22,58 As discussed in the section on diagnosis of bartonellosis, standard isolation and PCR techniques are frequently insensitive when attempting diagnostic detection of these Bartonella spp. in a dog, unless PCR is used to target a localized infection, such as endocarditis or the dog is therapeutically immunosuppressed, which appears to increase the level of Bartonella DNA in the blood of infected dogs.^47,102B. elizabethae, a species that infects rodents, was PCR amplified and sequenced from an EDTA-blood sample obtained from a dog that had experienced chronic weight loss culminating in sudden unexplained death.¹³⁹ A recent study reported the occurrence of (Candidatus B. rochalimae) and B. vinsonii subsp. berkhoffii genotypes II and III in dogs in southern Italy and provided DNA sequencing evidence of a potentially new Bartonella sp. infecting dogs in Greece and Italy.⁵⁷ In California, gray foxes appear to be the reservoir host for (Candidatus B. rochalimae), fleas are considered the likely mode of transmission among foxes and as mentioned above, this species has pathogenic potential in humans and dogs.^54,138 We have also isolated a novel Bartonella sp. from dogs in the southeastern United States that is genetically most closely related to B. volans.^b Collectively, these observations indicate that flea-transmitted Bartonella spp. that frequently infect cats, rodents, or other animals may cause disease manifestations in dogs. Whether, while obtaining a blood meal, fleas can infect a dog with any Bartonella spp. has not been proven. Unfortunately, the mode(s) of transmission for any Bartonella spp. to a dog has not been confirmed by laboratory-based vector transmission studies.^1,2 Efforts by the authors of this review to obtain research funding to facilitate vector transmission studies have not been historically successful. Although B. vinsonii subsp. berkhoffii is presumably transmitted to dogs by the bite of an infected tick, this mode of transmission has not been proven.¹¹⁵ Based upon anecdotal clinical evidence, dogs may be infected with B. henselae by a cat bite or scratch, analogous to CSD in humans. Although Bartonella spp. DNA has been documented in salivary samples obtained from cats and dogs,^25,119 neither the presence of viable bacteria nor saliva as a source of infection for animals or humans have been proven. B. vinsonii subsp. berkhoffii has been repeatedly isolated from a dog during a 15-month study period; therefore, chronic infection with B. vinsonii subsp. berkhoffii can occur in dogs.⁶⁵ Assuming that B. vinsonii subsp. berkhoffii causes chronic intraerythrocytic and endothelial cell infections in dogs, this organism is presumably well tolerated by an infected dog for extended periods of time. Similar to other highly adapted intracellular vector-transmitted pathogens, the factors that ultimately result in these bacteria causing disease manifestations are in most instances yet to be determined. However, it is increasingly obvious that the administration of immunosuppressive drugs for the treatment of immune-mediated hemolytic anemia, immune-mediated thrombocytopenia, or SLE is 1 factor that can facilitate the development of pathology in a dog. If similar to babesiosis, a known tick-borne intraerythrocytic pathogen, stress, hard work, parturition, vaccination, or concurrent infection with other organisms may contribute to the development of pathology. Coinfection with B. vinsonii subsp. berkhoffii and Babesia canis was diagnosed in a geriatric dog that presented for a history of chronic thrombocytopenia, increased liver enzyme activities, syncope, and seizures, all of which resolved after treatment was specifically directed at the intracellular bacteria and erythrocytic protozoa, residing within the vasculature.¹⁴⁰ In the context of chronic infection, following experimental inoculation of specific pathogen-free dogs with culture-grown B. vinsonii subsp. berkhoffii, there was sustained suppression of peripheral blood CD8+ lymphocytes, accompanied by an altered cell surface phenotype and an increase in CD4+ lymphocytes in the peripheral lymph nodes.^11,12 Therefore, infection with B. vinsonii subsp. berkhoffii and potentially other Bartonella spp. might induce a degree of chronic immunosuppression that could predispose dogs to infection with other infectious agents or the development of disease manifestations following the stress of vaccination or surgery. As described previously, occult infection with members of the genus Bartonella appear to induce a wide array of clinical manifestations in naturally infected dogs.

Hematologic, biochemical, and urinalysis abnormalities indicative for Bartonella spp. infections in dogs have not been well-characterized to date. Clearly, some infected dogs will have no or only very mild and nonspecific laboratory abnormalities. Thrombocytopenia has been implicated in association with B. vinsonii subsp. berkhoffii and B. henselae infections in dogs.^{100,105,117,140} In a study that included dogs that were seroreactive to B. vinsonii subsp. berkhoffii antigens, thrombocytopenia, anemia (which was frequently an immune-mediated hemolytic anemia), and neutropenia, or neutrophilic leukocytosis were the most commonly detected hematologic abnormalities.¹¹⁶ Thrombocytopenia was found in approximately half of the dogs and eosinophilia in approximately one-third of dogs with disease manifestations. Documentation of eosinophilia in a dog lacking internal and external parasites or evidence of cutaneous or pulmonary pathology should prompt consideration of bartonellosis.¹¹⁶ Monocytosis can also occur in B. vinsonii subsp. berkhoffii-infected dogs, particularly those with endocarditis. Hemoglobinuria, generally unaccompanied by hematuria, was a frequent finding, particularly in dogs with immune-mediated hemolytic anemia.¹¹⁶ Although increased liver enzyme activities have been found in a few dogs, serum biochemical abnormalities are usually very mild or nonexistent.

Necropsy and histopathologic findings associated with Bartonella spp. infection in dogs include endocarditis, myocarditis, granulomatous lymphadenitis, granulomatous hepatitis, cutaneous panniculitis, bacillary angiomatosis, and peliosis hepatitis. Multifocal areas of severe myocardial inflammation can be found in dogs with B. vinsonii subsp. berkhoffii endocarditis.^100,101,110 Although not specific for Bartonella infections, organisms can be detected in diseased tissues using silver stains, particularly in acute Bartonella infections, which would be analogous to acute regional lymphadenitis (CSD) in human patients. During chronic infections, organisms are presumably too few in number to be detected in tissues by silver staining, unless the fulminating infection is localized to heart valves or the patient is immunosuppressed by drug administration or concurrent disease and develops bacillary angiomatosis.¹¹⁴ Infection with a Bartonella spp. should be considered in dogs with granulomatous inflammation of undefined etiology or in dogs with vasoproliferative lesions indicative of bacillary angiomatosis or peliosis hepatis. Recent isolation of B. vinsonii subsp. berkhoffii from a dog with a hemangiopericytoma and a boy with epitheliod hemangioendothelioma³¹ suggests that this subspecies should be investigated as a cause of vasoproliferative tumors in dogs and human beings.

Diagnosis of Canine, Feline, and Human Bartonellosis

As is true for other intracellular pathogens that induce chronic infection in dogs after vector-borne transmission, including E. canis, B. canis, Babesia gibsoni, and Leishmania infantum, diagnostic confirmation of active infection with a Bartonella spp. can be extremely challenging. Because Bartonella spp. induce chronic intraerythrocytic, intravascular, and potentially lymphatic infection,^36,141 the diagnosis of canine bartonellosis (ie, active infection with a Bartonella spp.) should be considered in association with a diverse spectrum of clinical and pathologic abnormalities. Bacterial isolation, ELISA or IFA detection of Bartonella spp. antibodies and PCR amplification of Bartonella spp. DNA directly from patient samples all have substantial diagnostic limitations. As antibodies to B. vinsonii subsp. berkhoffii antigens are infrequently detected (<4%) in a sick referral or healthy (<1%) dog population, detection of B. vinsonii subsp. berkhoffii antibodies in a sick dog provides strong clinical evidence for prior exposure to and potentially active infection with this organism.¹¹⁵ As further support for the specificity of the B. vinsonii subsp. berkhoffii IFA test, antibody reactivity was not found in sick dogs from Canada¹²² or in dogs exposed to E. canis in Brazil.⁹⁵ For this reason, treatment of B. vinsonii subsp. berkhoffii seroreactive dogs or dogs from which Bartonella spp. DNA is detected in blood or tissue samples would be recommended. However, whenever possible, culture and PCR should be used to document active infection in seroreactive dogs. A reciprocal titer of 64 or greater is considered indicative of prior exposure to or active infection with B. vinsonii subsp. berkhoffii or B. henselae in dogs. Importantly, there is evidence to support serologic cross reactivity between B. henselae and Rickettsia spp. antigens in dogs and in humans.^134,142,143 This observation has proven to be of diagnostic relevance during consultation calls in which veterinarians report detection of R. rickettsii seroreactivity in dogs tested by commercial laboratories that do not also include testing for Bartonella spp. antibodies. Frequently, these dogs will respond initially to doxycycline, as a recommended treatment for Rocky Mountain spotted fever, but will then have a relapse of clinical or hematologic abnormalities during or after cessation of treatment. Recent serologic and molecular evidence indicates that coinfection in dogs with Ehrlichia, Babesia, Rickettsia, and Bartonella spp. may be more frequent than previously realized.^{95,121,135,140} As certain Borrelia, Ehrlichia, Babesia, and Bartonella spp. can cause chronic, insidious infection in dogs, the relative role of each organism to the pathogenesis of specific disease manifestations in a sick, naturally infected dog will remain difficult to establish in the clinical setting; however, each genus is associated with specific therapeutic considerations. When dealing with sick dogs that have a history of tick exposure, clinicians should screen using serologic and molecular testing modalities, whenever possible. Diagnostically, this approach allows for the simultaneous evaluation of exposure to and active infection with a spectrum of tick-transmitted pathogens. Because confirmatory tests are not 100% sensitive, serology remains an important diagnostic test that informs the clinician of prior exposure to a pathogen.

Confirming a diagnosis of feline bartonellosis can also be difficult. While most flea-exposed cats develop serum antibody responses and bacteremia as documented by culture or amplification of Bartonella spp. DNA from blood, positive serology, culture, and PCR test results can be detected in both healthy and clinically ill cats and test results can vary in an individual cat over time. As previously discussed, these findings lead to low positive predictive value for individual test results or combinations of tests, when attempting to confirm Bartonella spp. as the cause of specific disease entities. In sick cats, the American Association of Feline Practitioners Panel recommended use of positive diagnostic assay results combined with clinical parameters and response to therapy to make a presumptive diagnosis of bartonellosis.⁸⁴ However, as vague clinical signs of disease can resolve spontaneously, antimicrobials can have effect against other undiagnosed organisms, and some antimicrobials like doxycycline and azithromycin have anti-inflammatory effects, response to therapy cannot be used to definitely diagnosis feline bartonellosis. It is currently unknown whether there is clinical utility to following Bartonella spp. assay results in successfully treated cats. In 1 study, fever rapidly resolved on administration of 28 days of doxycycline or orbifloxacin but Bartonella spp. antibodies or DNA was still detected in the majority of the cats several weeks after finishing therapy.¹⁴² It is recommended that flea control be maintained to attempt to avoid new infections.⁸⁴ It appears that most cats that are concurrently seronegative and have negative direct culture or PCR assay results on blood have not been exposed to B. henselae or B. clarridgeaie but further information using more sensitive techniques is needed.

In contrast to the majority of cats, documenting active infection with a Bartonella spp., direct culture of blood or other diagnostic specimens (CSF, joint fluid, or cavitary effusions) onto blood agar plates from sick dogs and human patients has proven to be very insensitive.⁹⁴ Also, for reasons that remain unclear, antibody reactivity to the respective antigens is detected by indirect IFA testing in only 50% of dogs and humans in which active infection with B. vinsonii subsp. berkhoffii and B. henselae can be documented.^36,141Bartonella spp. are typically grown in various cell lines for IFA testing purposes, a process that might alter surface antigen recognition for this organism when testing patient samples. Regardless of the cause, antibody testing in dogs and human patients is highly insensitive, and if detected, the presence of antibodies can only be used to infer prior exposure to a Bartonella spp. Although less frequently encountered cats can also be seronegative despite infection with B. henselae.⁷⁵ The extent to which seroreactive dogs are actively infected is unknown, as studies to definitively characterize the association between seroreactivity and active infection in dogs are currently lacking. Similar to the poor sensitivity associated with culture onto blood agar plates, PCR amplification of Bartonella spp. DNA following direct extraction from patient blood or diagnostic fluid samples is also relatively insensitive, primarily due to the low level of bacteria typically found in diagnostic samples (Table 4).^36,94 Prior administration of immunosuppressive drugs or cancer chemotherapeutic agents appears to increase Bartonella bacteremia to a level that improves the likelihood of detecting organism-specific DNA sequences by PCR testing whereas administration of some antimicrobials before obtaining samples for BAPGM culture will decrease detection.^29,114,139 Owing to the limitations associated with standard diagnostic modalities, our research group optimized a novel diagnostic platform for the molecular detection and microbiologic isolation of Bartonella spp. from animal and human patient samples.^{14,15,31,61,94} The use of this diagnostic approach continues to change our understanding of the role of various Bartonella spp. as pathogens of human and veterinary importance.

Bartonella Preenrichment Culture

Ideally, the diagnosis of Bartonella infection should be confirmed by culturing the organism from aseptically obtained patient samples, including blood, CSF, lymph node or other tissue aspiration samples, cavitary effusions, joint or ocular exudates, and from surgical biopsies. Alternatively, PCR can be used to directly amplify organism-specific DNA sequences from various diagnostic sample sources and tissues, when obtained as surgical biopsies. Although it is clear that Bartonella DNA can be amplified from tissues, particularly heavily infected heart valves obtained from dogs or humans with Bartonella endocarditis, the sensitivity of PCR is poorly characterized when targeting a potentially low number of these bacteria in parenchymal tissues such as liver, spleen, kidney, or muscle, tissues that contain high host DNA concentrations that interferes with DNA amplification of the Bartonella target. In addition, as mentioned above, cross contamination with Bartonella spp. DNA in necropsy rooms or histopathology processing laboratories, where a large number of samples from various animal species are processed daily, is a recently recognized problem for veterinary pathologists and infectious disease researchers seeking to use archival samples and surgically obtained paraffin embedded diagnostic biopsy samples for PCR testing.¹¹³ As Bartonella spp. bacteremia is highly prevalent (ranging up to 100% of the study population) in cats, cows, and various small mammal and rodent species, all of which are routinely necropsied by veterinary pathologists, transfer of Bartonella-infected blood or fluids from 1 necropsy case (or biopsy) to the next is an inherent problem that will have to be addressed prospectively, if tissues are to be considered for molecular diagnostic purposes.

When culturing cat blood samples, B. henselae and B. clarridgeiae can often be isolated effectively using standard blood agar plates, a high CO₂ incubator, and prolonged incubation times (2–6 wk); however, isolation of Bartonella spp. from dog, horse, or human blood samples using the same approach is very insensitive mainly due to the low level of bacteremia that exists during persistent infection in these species (Table 4). In 2005, we described a novel, chemically modified, insect-based liquid culture medium (BAPGM) that supports the growth of at least 7 Bartonella spp.¹⁴⁴ This medium also supported cocultures consisting of different Bartonella spp. Subsequently we devised a unique diagnostic platform that combines preenrichment culture utilizing BAPGM, followed by a highly sensitive PCR assay (which has a sensitivity of 0.5 bacterial genome copies per microliter of sample DNA template) targeting the 16S-23S ITS region or Pap31, a bacteriophage-associated gene.^145,146 This approach has allowed us to characterize and quantify Bartonella infection in blood,^14,15,94 CSF,¹⁵ aqueous,¹⁴⁴ and joint fluids,³⁷ in addition to seroma fluid,³⁷ transudates, and modified transudates from dogs with idiopathic cavitary effusions³⁶ and tissue biopsy samples obtained at surgery. There are an increasing number of potential diagnostic indications for the clinician when considering Bartonella testing (Table 5). Most importantly, the use of this enrichment growth medium before PCR testing has allowed our research group to confirm that immunocompetent human patients, in particular veterinarians and veterinary technicians, can have chronic intravascular infections with Bartonella spp.^14,15,21,55

As currently used in our laboratory, the BAPGM diagnostic platform is cumbersome and time consuming; however, as the dividing time for Bartonella spp. is approximately 24 hours, enrichment culture is often necessary to increase bacterial numbers to detectable levels. For each sample, DNA is extracted directly from the sample for PCR amplification, and at the same time, a portion of the aseptically obtained sample is inoculated into BAPGM liquid preenrichment growth medium. Following incubation for at least 7 days in liquid culture, DNA is again extracted from the preenriched sample for PCR testing and at the same time a subculture is established by subinoculation from the liquid medium onto an agar plate. The agar plate is then incubated for up to 5 weeks. If bacterial growth is obtained, DNA is extracted from the isolate for Bartonella spp. PCR. When attempting to document Bartonella infection in several animal species and in immunocompetent humans, the combined BAPGM culture/PCR assay has become the preferred research testing platform utilized by the Intracellular Pathogens Research Laboratory, at North Carolina State University and the recommended diagnostic platform offered by Galaxy Diagnostics Inc (Research Triangle Park, NC, USA). Diagnostic testing (serology, PCR, and BAPGM blood culture/PCR combination testing) for Bartonella spp. using BAPGM is only available through: Galaxy Diagnostics Inc (http://www.galaxydx.com).

When compared with more traditional methods, this combinational approach has facilitated the detection of active infection in dogs with at least 4 Bartonella spp. (B. henselae, B. quintana, B. vinsonii berkhoffii, and B. bovis), but of perhaps greater comparative microbiologic importance, this approach has resulted in the successful isolation of B. henselae (among the only canine derived clinical isolates to date) from numerous sample sources obtained from sick dogs. In some dogs, Bartonella DNA can be amplified from EDTA-anticoagulated blood samples or from other diagnostic samples, but inoculation of BAPGM does not result in preenrichment growth or in the subsequent isolation of the infecting Bartonella spp. by agar plate subculture. Therefore, direct PCR from the original patient sample is always used as a component of our testing platform. Most often, this scenario appears to be related to administration of antimicrobials before submission of the diagnostic sample for BAPGM enrichment culture. In other instances, an insufficient quantity of viable bacteria may be present in the sample resulting in failure of the bacteria to proliferate in BAPGM to levels that can be detected by PCR (ie, failure to achieve a minimum of 500 bacteria per milliliter of liquid medium would result in a negative PCR). These observations emphasize the importance of proper aseptic sample collection and rapid refrigeration and storage of samples before testing in BAPGM. It is also important to mention that BAPGM will grow numerous other bacteria, including skin flora, fastidious environmental bacteria and potentially highly fastidious uncharacterized intravascular bacteria, for which the pathogenic potential remains unknown.^147,148 In a recent study that predated our use of BAPGM, there was no serologic or molecular evidence to support a role of Bartonella spp. in dogs with lymphocytic plasmacytic rhinitis.¹⁴⁹ Although this conclusion may continue to be accurate, it is important to note that approximately 50% of B. vinsonii subsp. berkhoffii and B. henselae infected dogs do not have detectable antibodies by IFA testing, BAPGM enrichment culture was not used before PCR testing, and other Bartonella spp. known to infect dogs do not cross react with B. vinsonii subsp. berkhoffii and B. henselae antigens, which are the standard test antigens used in our laboratory. As described in ‘Introduction,’ improvements in test sensitivity continue to modify our understanding of the role of Bartonella spp. as animal and human pathogens. Finally, due to the prolonged incubation time required for the successful use of this diagnostic testing platform, it is critical that the clinician or technician obtain diagnostic samples for BAPGM culture with the utmost attention to aseptic collection (clipping hair, new cotton balls, fresh surgical preparation solutions, no contact with the needle or collection site) and before antimicrobial administration, whenever possible.

Treatment of Bartonella spp.

To date, treatment recommendations and approaches applied to cats or dogs infected with a Bartonella spp. have been derived predominantly from the experiences generated in human medicine. Human Bartonella spp. treatment experiences can be placed into 1 of 3 major categories: (1) treatment of individuals with CSD^150,151 and (2) treatment of immunodeficient individuals (transplant recipients, immunosuppressive drug therapy recipients, humans infected with human immunodeficiency virus),^17,152 and (3) treatment of B. quintana and B. henselae endocarditis.¹⁵³ Because of disparate results among studies and an overall lack of microbiologic data in clinical therapeutic trials, numerous issues related to treatment of human, canine, and feline Bartonella infections remain controversial and essentially unstudied. In contrast to the apparent lack of response to antimicrobial treatment in human CSD patients, bacillary angiomatosis, parenchymal bacillary peliosis, and acute Bartonella bacteremia generally respond to antimicrobial treatment, even when individuals are immunocompromised. In these individuals the vasoproliferative lesions or febrile illness resolve. However, there are numerous case reports involving immunocompromised humans in which relapses are documented following cessation of antimicrobial treatment. Based upon in vitro testing, numerous antimicrobials appear to be effective for the treatment of Bartonella infections.¹⁵³ However, as Bartonella spp. induce both intracellular, as well as extracellular infection, in vitro test results can confirm those antimicrobials that are not effective and can identify antimicrobials that should be tested in clinical trials for in vivo efficacy. Doxycycline, erythromycin, and rifampin are the most frequently recommended antimicrobials used for treating Bartonella spp. infection in humans, but clinical improvement has been reported following the use of penicillin, gentamicin, ceftriaxone, ciprofloxacin, and azithromycin.^{150–152,154} In human medicine, treatment for 2 weeks in immunocompetent and 6 weeks in immunocompromised humans is generally recommended. Relapses in bacteremia, fever or other disease manifestations have been reported in immunocompromised humans, despite a 6-week treatment regimen.

Antimicrobial efficacy has not been established for any antimicrobial that might be used to eliminate B. henselae or B. vinsonii subsp. berkhoffii bacteremia in cats or dogs. Incomplete treatment responses have been reported in experimentally infected cats treated for 2 or 4 weeks with doxycycline or enrofloxacin¹⁵⁵ and in client-owned cats administered doxycycline or orbifloxacin.¹⁴² Ampicillin and doxycycline appeared to be effective in treating cats experimentally infected by needle inoculation of culture-grown B. henselae; however, the duration of bacteremia was short even in untreated control cats.^156,157 Results derived using inoculation of culture-grown Bartonella spp. should be viewed with caution as the biological behavior of B. henselae following in vitro culture appears to differ from the biological behavior of these organisms if transmitted by blood transfusion or by flea infestations.^59,70 Cats with flea-induced transmission of B. henselae had fever rapidly resolve after oral administration of 5 mg/kg of enrofloxacin.⁷² However, there was not an untreated control group and so it is unknown whether the antimicrobial therapy was effective or whether the fever resolved spontaneously. In the context of protective immunity following natural infection, cats experimentally infected with B. henselae do not develop protective immunity to heterologous challenge with other B. henselae genotypes or when challenged with B. clarridgeiae.⁸⁵ Therefore following successful treatment, a cat could become reinfected with a different B. henselae strain or a different species. Flea control may be critical to prevent reinfection following successful treatment. Similar to cats, we have documented coinfections in dogs with more than 1 Bartonella spp. or strain type.^36,95 Although untested experimentally, it is likely that dogs and human beings can be reinfected with heterologous strains of the same Bartonella spp. or with other Bartonella spp. following therapeutic elimination of an initial infection. Considering the large number of reservoir hosts in nature, the spectrum of diverse arthropod vectors, and the tendency for some dogs to roam the environment, exposure to more than 1 Bartonella spp. during a lifetime is perhaps the norm, rather than the exception. Based upon experimental infection studies in cats there appears to be variability in virulence among B. henselae strains found in nature, suggesting that more virulent strains will induce more severe disease in dogs or humans.⁷¹ In addition, alterations in virulence can be induced in the laboratory during in vitro culture of the organism, which in most instances contributes to a decrease in pathogenicity when culture-grown organisms are used as inoculums in experimental infection studies.^68–71,73 Therefore, when performing experimental infection studies in animals to address pathogenicity, virulence, treatment or vaccine strategies, the use of the natural vector for transmission of the particular Bartonella spp. is recommended.

For the past several years and based upon extrapolation from the human literature, the Intracellular Pathogens Research Laboratory has recommended treatment with azithromycin for dogs with microbiologic documentation of active Bartonella spp. infection. A standard treatment regimen using azithromycin (5–10 mg/kg daily for 7 d followed by every other day administration for an additional 5 wk) has proven effective for most, but not all cats and dogs. Fluoroquinolones alone, or in combination with amoxicillin, have also elicited a positive therapeutic response in dogs, which is accompanied by a progressive decrease in B. vinsonii antibody titers.¹¹⁶ Doxycycline may or may not be effective for treatment of B. vinsonii subsp. berkhoffii, but data from cats experimentally or naturally infected with B. henselae or B. clarridgeiae indicate that a high dose of doxycyline (10 mg/kg, q 12 h, for 4–6 wk) may be necessary to eliminate Bartonella infection in cats, dogs, or other animal species.¹⁵⁰ In sick dogs in which antibodies to B. henselae or B. vinsonii subsp. berkhoffii can be detected before antimicrobial treatment, antibody titers decrease in a rapid, decremental fashion following therapeutic elimination of the organism.¹¹⁶ Surprisingly, many dogs that experience resolution of disease manifestations will no longer have detectable antibodies as early as 3–6 months following treatment. Therefore, posttreatment serology may be a useful adjunct to BAPGM/PCR to determine if therapeutic elimination of bartonella infections has been achieved. Unfortunately, as stated above, only approximately half of the B. henselae or B. vinsonii subsp. berkhoffii infected dogs (as documented by BAPGM culture/PCR and DNA sequencing) have detectable antibodies at the time of initial testing. In some dogs, bactermia persists despite treatment with the azithromycin regimen. Dogs that fail the initial 6-week course of azithromycin have been subsequently treated with a combination of azithromycin and rifampin (5 mg/kg daily) or doxycycline and rifampin for an additional 6 weeks. This approach has resulted in disease resolution and negative posttreatment blood cultures in the small number of dogs for which we have follow-up data. Recent unpublished data from our laboratory indicate that B. henselae and B. vinsonii subsp. berkhoffii can induce sustained intravascular infections in human patients, despite administration of antimicrobials such as azithromycin and rifampin that achieve high intracellular concentrations and are used for a 4–6-week treatment duration. We have also reported therapeutic failure of multiple courses of azithromycin and marbofloxacin to eliminate infection with B. vinsonii subsp. berkhoffii and B. henselae from the blood and joint fluid of a dog with chronic, debilitating arthritis.³⁷ In previous work, azithromycin was shown to be ineffective for the treatment of feline hemoplasmosis.¹⁵⁸ Outdoor cats with fever of unknown origin could be ill from Bartonella spp., hemoplasmas, Ehrlichia spp., or Anaplasma spp. depending on geographical location.^159,160 As azithromycin is not effective for hemoplasmas, Ehrlichia spp., or Anaplasma infections, azithromycin may not be the optimal empirical antimicrobial choice for cats with fever suspected to be associated with a vector-borne disease. Regardless of the antimicrobial that is used for treatment, a long duration of antimicrobial administration (4–6 wk) may be necessary to eliminate Bartonella infections in cats, dogs, or human beings. Induction of antimicrobial resistance may be responsible for treatment failures in some Bartonella-infected patients.^161–167 Recently, specific antimicrobial resistance genes have been characterized in B. bacilliformis, B. henselae, and B. quintana by in vitro serial passage.^161–163 In data from our laboratory, azithromycin was active only until the second passage for B. henselae isolates obtained from cats.¹⁶⁸ This finding is disconcerting as failure of azithromycin to eliminate infection in a cat could predispose to the development of a antimicrobial resistant strain of B. henselae that could potentially be transmitted to the owner or an animal health professional by a bite or scratch. This is perhaps another reason that healthy, seroreactive cats should not be automatically treated with antimicrobials. In a prospective, randomized, double-blind and placebo-controlled study that addressed antimicrobial treatment of CSD in humans,¹⁵¹ azithromycin taken orally for 5 days was considered to be effective in decreasing lymph node size within the first 4 weeks following therapy. However, enlargement of new lymph nodes or an increase in size of the original lymph node occurred in some study subjects despite azithromycin therapy. Our in vitro results might explain relapses or treatment failures observed in vivo when using azithromycin for the treatment of Bartonella-related infections. As described above, due to difficulties associated with obtaining stable isolates from patient samples, very few clinical isolates have been tested for antimicrobial sensitivity or resistance.

Based upon the approach used for the treatment of human endocarditis patients,¹⁵⁴ dogs with acute or severe life-threatening infections (endocarditis, pneumonitis, meningoencephalitis, and systemic granulomatous disease) an aminoglycoside is recommended during the initial management of the patient, assuming that renal function is stable and renal perfusion will be maintained through the administration of IV fluids. Aminoglycosides are the only antimicrobial class reported to be bactericidal against Bartonella spp.^152,154 Administration of an aminoglycoside to human endocarditis patients decreases morbidity, shortens the hospitalization period, and lessens the need for heart valve replacement. Highly structured prospective treatment studies of cats and dogs infected with a Bartonella spp. are needed to clarify appropriate antimicrobial selection and optimal treatment durations. In addition, further evaluation of posttreatment serologic and BAPGM culture findings in patients are needed.

Application to Veterinary Emergency and Critical Care

In the context of this review and the current understanding of the genus Bartonella, the zoonotic potential for transmission of these bacteria to veterinary emergency and critical care professionals is of greatest concern. Transmission via a bite, a scratch, an arthropod vector, or by direct contact with blood or other bodily fluids from animals could be facilitated in the emergency room situation, when caring for a severely ill, debilitated, or traumatically injured animal infected with a Bartonella spp. In addition, B. henselae is commonly present within fleas and flea dirt where it survives for at least 9 days.^164,165 As Bartonella spp. have been isolated from cat, dog, or human blood, CSF, joint fluid, aqueous fluid, seroma fluid and from pleural, pericardial and abdominal effusions, a substantial number of diagnostic biological samples collected on a daily basis in veterinary practices could contain viable Bartonella organisms. The increasing number of defined Bartonella spp., in conjunction with the high level of bacteremia found in reservoir-adapted hosts, which represent the typical veterinary patient population for many practices, ensures that most if not all veterinary professionals will experience frequent and repeated exposure to animals harboring these bacteria. Therefore, personal protective equipment, frequent hand washing, and avoiding cuts and needle sticks have become more important as our knowledge of this genus has improved and various modes of transmission have been defined.

In the context of veterinary care for cats, dogs, and other animals, evolving case-based and limited experimental evidence supports the concept that Bartonella spp. are among the most efficient stealth bacterial pathogens defined to date.^1–3,62 Because these bacteria can invade erythrocytes, macrophages, dendritic cells, and endothelial cells, dissemination to any location within the animal's body may be feasible. For the emergency care veterinarian, obtaining and properly storing serum and aseptically obtained blood and other diagnostic fluid samples before initiation of antimicrobials is of critical importance to achieve diagnostic documentation of infection. On a practical basis, this approach is easier to propose in a review manuscript than to accomplish in a busy emergency setting. However, as described in detail above, obtaining pretreatment diagnostic specimens is of critical importance when assessing the Bartonella infection status of a non–reservoir-adapted host. Unfortunately, as discussed in the section on diagnosis, these stealth pathogens are difficult to detect under optimal clinical and laboratory circumstances. Prior administration of antimicrobials, even if not therapeutically effective in eliciting a cure, can result in a rapid decrease detectable antibody levels and can eliminate the potential for BAPGM enrichment culture to increase organism numbers to detectable levels before performing PCR. Based upon a growing body of literature, a large number of clinical or hematologic abnormalities might prompt consideration of Bartonella spp. testing (Table 4).

Finally, veterinarians play an increasingly important role in advising the public as to the epidemiologic and zoonotic implications associated with vector-borne pathogens. As discussed above, numerous nondomestic animal species frequently serve as the primary reservoir for Bartonella spp. Based upon scientific evidence generated during the past several decades, vector-transmitted pathogens can induce clinical manifestations ranging from acute fatal illness (eg, Rocky Mountain spotted fever, ehrlichiosis, babesiosis, and bartonellosis) to chronic debilitating disease states (eg, ehrlichiosis, babesiosis, borreliosis, and bartonellosis). Therefore, minimizing or eliminating flea and tick exposure is perhaps of greater veterinary and public health importance today, than during any previous time in the history of veterinary medicine. When rigorous flea and tick control measures are instituted, it is highly probable that transmission of Bartonella spp. will be greatly reduced or eliminated.⁷² Recently, we reported a high molecular prevalence of Bartonella spp. infection in healthy Golden Retrievers and Golden Retrievers with lymphoma (18% of both populations were actively infected based upon PCR testing).²⁴ Interestingly, an earlier study also found that routine application of acaracides decreased the risk of lymphoma in Golden Retrievers.¹⁶⁹ Of comparative biomedical importance, dogs and humans develop similar disease manifestations and pathology following infection with a Bartonella spp.^58,170 Therefore, research observations derived in 1 species has proven to be of medical benefit to the other species. Although recent research findings have substantially improved our understanding of the clinical, microbiologic, and zoonotic aspects of diseases caused by Bartonella spp., the exact mode of transmission, the relative role of various insect vectors such as fleas and ticks, the identification of potential reservoir hosts, and the spectrum of animal and human illnesses caused by these organisms remains largely undetermined. Particularly lacking are evidence-based studies that define duration of infection, that characterize hematologic and pathologic consequences on natural infection and studies that confirm causation of disease induced by infection with specific Bartonella spp.

Footnotes

^aBreitschwerdt EB, Maggi RG: Unpublished data.

^bMaggi RG, Breitschwerdt EB: Unpublished data.

^cChomel B, Sykes J: Unpublished data.