Risk Factors for Bartonella species Infection in Blood Donors from Southeast Brazil

Bacteria from the genus Bartonella are emerging blood-borne bacteria, capable of causing long-lasting infection in marine and terrestrial mammals, including humans. Bartonella are generally well adapted to their main host, causing persistent infection without clinical manifestation. However, these organisms may cause severe disease in natural or accidental hosts. In humans, Bartonella species have been detected from sick patients presented with diverse disease manifestations, including cat scratch disease, trench fever, bacillary angiomatosis, endocarditis, polyarthritis, or granulomatous inflammatory disease. However, with the advances in diagnostic methods, subclinical bloodstream infection in humans has been reported, with the potential for transmission through blood transfusion been recently investigated by our group. The objective of this study was to determine the risk factors associated with Bartonella species infection in asymptomatic blood donors presented at a major blood bank in Southeastern Brazil. Five hundred blood donors were randomly enrolled and tested for Bartonella species infection by specialized blood cultured coupled with high-sensitive PCR assays. Epidemiological questionnaires were designed to cover major potential risk factors, such as age, gender, ethnicity, contact with companion animals, livestock, or wild animals, bites from insects or animal, economical status, among other factors. Based on multivariate logistic regression, bloodstream infection with B. henselae or B. clarridgeiae was associated with cat contact (adjusted OR: 3.4, 95% CI: 1.1–9.6) or history of tick bite (adjusted OR: 3.7, 95% CI: 1.3–13.4). These risk factors should be considered during donor screening, as bacteremia by these Bartonella species may not be detected by traditional laboratory screening methods, and it may be transmitted by blood transfusion.


Introduction
Bartonella species are fastidious alpha-proteobacteria with worldwide distribution. To date, at least 15 species have been associated with human infections, with at least eight species also capable of infecting dogs and cats. Several blood-sucking arthropods have been suggested or confirmed as vectors for this genus, including sandflies, body lice, fleas, ticks, and keds [1]. In humans, select species of Bartonella were confirmed as etiologic agents of cat scratch disease (CSD), trench fever, bacillary angiomatosis and Oroya fever [2]. Moreover, Bartonella infections have also been documented by culture or molecular methods in human cases of endocarditis, myocarditis, polyarthritis and granulomatous inflammatory disease [1,2]. In countries throughout the world, most diseases associated with Bartonella species infection are not reportable; therefore, incidence data is scarce. One study performed at the end of the 20 th century estimated that 22,000 new cases of cat scratch disease appear every year in the United States, and roughly 10% of these infections were thought to require hospitalization [3].
Bartonella species cause chronic and intermittent intra-erythrocytic bacteremia and infect endothelial cells of both incidental and natural reservoir hosts. The establishment of chronic, stealth infection is achieved by evasion of innate immune responses. These include resistance to complement activation, antigenic variation of surface proteins, and inhibition of host cell apoptosis [4]. Consequently, subclinical bloodstream infection in humans has been reported [5], supporting the fact that these bacteria will not follow Koch's postulates for disease causation [6].
Asymptomatic Bartonella species infection poses a hazard for blood recipients because fast, sensitive and specific diagnostic tests are not currently available for donor screening. To the authors' knowledge, no prophylactic measures are currently in place to prevent collection and transfusion of blood and blood products contaminated with species of Bartonella worldwide. Recently, using a combination of culture methods and PCR assays, we documented Bartonella henselae and Bartonella clarridgeiae infection of blood samples from 16 asymptomatic blood donors at a large blood bank center in Southeast Brazil [5]. Risk factors for Bartonella species exposure in blood donors were initially evaluated in one study from Sweden using serology methods [7]. However, antibody detection has limited predictive value in confirming or excluding Bartonella species bacteremia in humans and animals [6,[8][9][10]. Furthermore, higher vector activity is expected in tropical and sub-tropical regions of the world. Therefore, this study objective was to determine which risk factors are associated with Bartonella species infection in asymptomatic blood donors from a population in Campinas, Sao Paulo State, Brazil.

Ethics statement
The

Sample collection and diagnostic techniques
We collected blood from five hundred apparently healthy voluntary blood donors in this cross sectional study. Sample size was determined with an alpha of 0.05, power of 0.8, and estimated prevalence of 5%, as previously reported in a serology-based study [11]. Donors were selected through convenience sampling, with inclusion and exclusion criteria following the current international standards for blood bank donor selection [12]. Bartonella species infection from the bloodstream was detected based on enrichment blood culture in a liquid growth medium (Bartonella alpha-Proteobacteria growth medium-BAPGM), coupled with isolation in solid medium and Bartonella-specific DNA amplification by PCR, followed by DNA sequencing to confirm species identification [8].

Epidemiological data collection
A standardized epidemiological questionnaire was delivered to each blood donor participant. The interviewer had no knowledge of the diagnostic test results, as Bartonella species testing was performed subsequent to interviews. The following information was captured: gender; selfreported ethnicity as African-American, native Indian, Caucasian, Asian or multi-racial; average monthly income represented by multiples of Brazilian monthly minimal wage; occupational animal exposure (veterinary professionals and others with direct animal exposure such as veterinary assistants, ranchers, biologists, and volunteers at shelters); contact with cats, dogs, other companion animals, livestock or wildlife (including handling animals, bedding, waste or sharing the same environment); bites from dogs, cats, and other animals; arthropod bites caused by ticks, fleas, or other insects; previous blood transfusion; previous history of blood donation; and presence of tattoos. Past history of contact with animals, animal bites, and/or arthropod bites (defined as more than one year after contact) were also recorded.

Data analysis
Potential risk factors were first compared in a univariate analysis using Fisher's exact test or the Fisher-Freeman-Halton test. All risk factors significant at the p< 0.25 level were entered into a stepwise logistic regression model, and variables significant to p< 0.05 were retained. Univariate odds ratios (OR), adjusted odds ratios (aOR), and 95% confidence intervals (CI) were calculated. Variables with collinearity were removed from the multivariate analysis. Statistical analyses were performed using JMP Pro 10 for Windows (SAS Institute Inc., Cary, NC).

Results
Bartonella species bloodstream infection was detected in 16/500 blood donors (3.2%). DNA amplification and sequencing identified B. henselae in 15 blood donors (3%) and B. clarridgeiae in one donor (0.2%), which was previously reported [5,13]. B. henselae bacteremia was also confirmed in six donors by bacterial isolation. The univariate and multivariate analyses of risk factors between blood donors infected with Bartonella species and uninfected subjects are provided in Tables 1 and 2. With univariate analysis, a professional with animal exposure was seven times more likely to be infected with Bartonella species than blood donors working in all other professions. When significant variables were entered into the multivariate logistic regression model, collinearity between animal-related professions and cat contact was documented; therefore, profession was not maintained in the final model. Adjusted odds ratio indicated that subjects with cat contact, or past history of tick bite, were approximately 3 to 4 times more likely to have a Bartonella species infection than donors without cat contact or lack of history of tick bite (Table 2).

Discussion
This study identified two risk factors associated with subclinical Bartonella species bloodstream infections in blood donors. Bartonella species infection was three times more likely to be diagnosed in blood donors who had contact with cats compared to blood donors with no contact. Similarly, blood donors working in animal-related professions were seven times more likely to be infected with these pathogenic bacteria, although this variable was not included in the multivariate model due to collinearity. In Swedish blood donors, contact with cats was also a risk factor for B. elizabethae seropositivity [7]. Our findings indicate that cat contact also increases the risk of subclinical bloodstream infections with B. henselae or B. clarridgeiae. Previously, 24% of 192 non-immunocompromised Americans with frequent exposure to cats, cat scratches or fleas had detectable Bartonella species DNA in blood specimens using the same diagnostic approach as used in our study [9]. Cats are natural reservoir hosts for B. henselae, B. clarridgeiae, and Bartonella koehlerae, all of which are important zoonotic species [2]. After fleatransmitted infection with most Bartonella species strains, cats rarely develop clinical manifestations but remain persistently infected with high levels of intravascular bacteria, which facilitates pathogen acquisition by blood-sucking vectors [10]. Recently, artificial feeding of Ctenocephalides felis with B. henselaeor B. clarridgeiaeinfected blood demonstrated that these pathogens can persist in C. felis and be excreted in flea feces [14]. Cat nails can be contaminated with infected flea feces, where B. henselae can survive for several days [15]. Needle stick transmission of Bartonella species has also been reported [16]. Bacterial transmission is less likely to occur by cat bite, since shedding of Bartonella species in cat saliva has not been clearly documented [17]. Interestingly, a history of cat bite, cat scratches, or flea bite was not significantly associated with Bartonella species infection in our study. Possible explanations include donors avoiding cat bites and scratches during "rough play", lack of identifying fleas as the source of an insect bite, cat confinement to the house, and routine use of parasiticides.
In Southeast Brazil, the combined Bartonella prevalence reported in domestic and stray cats by five reports was 38.5% (112/291), but ranged from 4.3% to 97% among feline populations tested [18,19,20,21,22]. Such wide variation may be associated with level of flea infestation or analytical sensitivity of PCR assays used. B. henselae and B. clarridgeiae were the only two species detected by molecular methods from cats in these studies. Furthermore, we have tested 112 cats from the same county of the blood donors (Campinas, Brazil) for Bartonella bacteremia, using the same diagnostic approach used in the present report (BAPGM culture and PCR), where we detected bacteremia in 90% (101/112) of cats and B. henselae was confirmed by DNA sequencing and BLAST analysis from three culture isolates [23]. Therefore, similar to reports from other countries, cats may be the main reservoirs of B. henselae and B. clarridgeiae in Southeast Brazil, and pose as a risk factor for subclinical infection in humans in this region.
A previous history of tick bite was also determined to be a risk factor for Bartonella species infection in this human population. While cat fleas are a well-established vector, the capability of ticks to transmit these organisms has been the subject of substantial debate. It has been demonstrated that seven Bartonella species are capable of growing in an Amblyomma americanum tick cell line [24]. Recently, the vector competency of Ixodes ricinus for transmission of Bartonella birtlesii was confirmed using a mouse model [25]. Although. Ixodes ricinus and Amblyomma americanum are not present in Brazil, other ticks such as Ixodes loricatus, Ixodes didelphidis, Amblyomma cajennense, Amblyomma aureolatum, Amblyomma ovale, and Rhipicephalus sanguineus have been isolated from marsupials and rodents living in a locality where other tick-borne organisms had been previously identified [26]. Therefore, while the mode of transmission of any Bartonella species to humans is still unclear, the potential role of tick transmission should not be ignored. Cat-scratch disease, bacillary angiomatosis and endocarditis are the most frequently reported manifestation of Bartonella infection in both immunocompetent or immunocompromised Brazilian populations [27][28][29], although ocular, neurologic and dermatologic abnormalities were also reported [30]. In addition, Correa et al. reported that the risk of Bartonella DNA bloodstream infection was 45 times higher in arrhythmic patients from Brazil or Argentina when compared to controls [31]. Our results expand the current understanding about Bartonella species infection in humans in Brazil by reinforcing the ecological role of cats and ectoparasites in the transmission of Bartonella. Risk of human infection can be minimized by implementing year-round ectoparasite control in domestic animals, by avoiding cat bites and scratches, and by keeping cats indoors to minimize exposure to vectors [32]. The medical impact of Bartonella species occurrence in the blood supply is still unknown and should be carefully investigated.
Supporting Information S1 Dataset. Complete data set of Brazilian blood donors who participated in the study. Donors are identified by a unique numerical ID, and significantly associated variables (Tables  1 and 2) are provided with responses for occupational animal exposure, contact with cats, and recall of a past tick bite. Column labeled "status" refers to presence or absence of Bartonella infection following blood sample analysis described in Materials and Methods. (XLSX)