Bartonella spp. Bacteremia and Rheumatic Symptoms in Patients from Lyme Disease–endemic Region

Prevalence of Bartonella spp. was high, especially among patients with a history of Lyme disease.

From each patient, the attending rheumatologist aseptically obtained anticoagulated blood samples (in EDTA tubes) and serum samples and shipped them overnight to the laboratory. Patient variations included timing of sample collection relative to onset of illness, duration of illness, current illness severity, and prior or recent use of antimicrobial drugs. The samples were then processed in a limited-access laboratory.

Bartonella α Proteobacteria Growth Medium Enrichment Culture
Each sample was tested by PCR amplifi cation of Bartonella spp. DNA before and after enrichment of blood and serum in Bartonella α Proteobacteria growth medium (BAPGM) (10)(11)(12)(13)(14)(23)(24)(25)(26). The BAPGM platform incorporates 4 PCR steps, representing independent components of the testing process for each sample, as follows: step 1) PCR amplifi cations of Bartonella spp. after DNA extraction from whole blood and serum; steps 2 and 3) PCR after whole blood culture in BAPGM for 7 and 14 days; and step 4) PCR of DNA extracted from subculture isolates (if obtained after subinoculation from the BAPGM fl ask at 7 and 14 days onto plates containing trypticase soy agar with 10% sheep whole blood, which are incubated for 4 weeks). To avoid DNA carryover, we performed PCR sample preparation, DNA extraction, and PCR amplifi cation and analysis in 3 separate rooms with a unidirectional work fl ow. All samples were processed in a biosafety cabinet with HEPA (high-effi ciency particulate air) fi ltration in a limited-access laboratory.
Conventional PCR was performed in an Eppendorf Mastercycler EPgradient (Hauppauge, NY, USA) under the following conditions: 1 cycle at 95°C for 2 s, followed by 55 cycles with DNA denaturing at 94°C for 15 s, annealing at 64°C for 15 s, and extension at 72°C for 18 s. The PCR was completed by a fi nal cycle at 72°C for 30 s. As previously described for the Bartonella ITS genus and B. koehleraespecifi c PCRs, all products were analyzed by using 2% agarose gel electrophoresis and ethidium bromide under UV light, after which amplicon products were submitted to a commercial laboratory (Eton Bioscience Inc., Research Triangle Park, NC, USA) for DNA sequencing to identify the species and ITS strain type (13,15,28,30).
To check for potential contamination during processing, we simultaneously processed a noninoculated BAPGM culture fl ask in the biosafety hood in an identical manner for each batch of patient blood and serum samples tested. For PCR, negative controls were prepared by using 5 μL of DNA from the blood of a healthy dog. All controls remained negative throughout the course of the study.

Statistical Analysis
Descriptive statistics were obtained for all demographic variables, self-reported clinical symptoms and concurrent conditions, previous specialist consultation, and self-reported exposures. The χ 2 test was used to assess associations between self-reported clinical symptoms and previous specialist consultation separately with PCR results for B. henselae; B. koehlerae; and B. vinsonii subsp. berkhoffi i genotypes I, II, and III. The Fisher exact test was used when expected cell value was <5. For the initial analysis, a liberal α value (α<0.10) was selected. The effect of each signifi cant variable on the outcome variables was adjusted in separate multivariate logistic regression models controlling for age, sex, and health status. The models were repeated for different possible outcomes: PCR results for B. henselae or PCR results for B. koehlerae. Variables maintaining p<0.05 were considered signifi cant. For some comparisons of potential interest, we were unable to estimate associations with the outcome(s) of interest because of low numbers (e.g., B. vinsonii subsp. berkhoffi i genotypes I, II and III). Statistical analyses were performed by using SAS/STAT for Windows version 9.2 (SAS Institute Inc., Cary, NC, USA).

Patient Characteristics
The age range of the 296 patients was 3-90 years; median ages were 46 years for women and 36 years for men (Table 1). Women made up ≈70% of the study population. Most (68.2%) patients reported that they felt ill, whether chronically or infrequently, and 27.7% considered themselves to be generally healthy. The most common animal exposure reported was dog (n = 252; 85.1%), followed by cat (n = 202; 68.2%) and horse (n = 86; 29.0%). Most patients reported having been bitten or scratched by an animal (n = 202; 68.2%) or exposed to ticks (n = 229; 77.4%) and biting fl ies (n = 160; 54.0%). Hiking was the predominant outdoor activity reported (52.0%). Most (273 [92.2%]) patients reported having had a condition diagnosed before visiting the rheumatologist. Previously diagnosed conditions included Lyme disease (46.6%), arthralgia/arthritis or osteoarthritis/rheumatoid arthritis (20.6%), chronic fatigue (19.6%), and fi bromyalgia (6.1%) (Figure 1). For B. vinsonii subsp. berkhoffi i, 148 (50.0%) patients were seroreactive by IFA testing to at least 1 of 3 genotypes, and 10 (3.4%) had a positive PCR. Of these 10 patients, 3 were infected with genotype I, 6 were infected with genotype II, and for 1 patient the genotype could not be defi ned on the basis of readable DNA sequence. Seroreactivity to genotypes I, II, and III was found for 77 (26.0%), 102 (34.5%), and 82 (27.7%) patients, respectively. There was no association between B. vinsonii subsp. berkhoffi i seroreactivity and bacteremia. Combined PCR and IFA assay results are summarized in Table 2. Of the patients with a positive PCR, 65% reported a prior diagnosis of Lyme disease (n = 138), bartonellosis (n = 29), or babesiosis (n = 14). Among the 138 patients with a prior diagnosis of Lyme disease, the prevalence of Bartonella PCRs indicated the following: B. henselae positivity was associated (p<0.05) with blurred vision and numbness (Table 3), patients who had visited a neurologist were more likely than those who had not to be B. henselae positive, older median age was signifi cantly associated with B. koehlerae positivity, and patients who reported paralysis were more likely to be positive for B. vinsonii subsp. berkhoffi i. No associations were found for self-reported exposures (e.g., insect or animal exposure) and positive PCR for B. henselae, B. koehlerae, or B. vinsonii subsp. berkhoffi i.

Logistic Regression Analysis
To identify factors associated with PCR positivity for B. henselae or B. koehlerae, we adjusted the models for 3 biological confounders: age, sex, and health status (

Discussion
We identifi ed unexpectedly high serologic and molecular prevalence for B. henselae, B. koehlerae, and B. vinsonii subsp. berkhoffi i in patients who had been examined by a rheumatologist, of whom more than half reported a prior diagnosis of Lyme disease, bartonellosis, or babesiosis. However, the diagnostic criterion upon which these infections were based was not available for review because all prior diagnoses were self-reported. Overall, 185 (62.5%) of 296 patients had antibodies to B. henselae, B. koehlerae, or B. vinsonii subsp. berkhoffi i, and 122 (41.1%) were positive for Bartonella spp. according to PCR. In most instances, DNA sequencing of the amplifi ed product facilitated identifi cation of the infecting species. The prevalence of antibodies against Bartonella spp. (93 [67.4%]) and bacteremia [57 [1.3%]) among 138 patients with a prior diagnosis of Lyme disease did not differ from that of the overall study population. Because our analysis was restricted to patients selected by a rheumatologist practicing in a Lyme disease-endemic region, extrapolations to other regions or other rheumatology practices might not be applicable. Also, because the survey was self-administered, objective confi rmation of symptoms, conditions, and diagnoses was not always possible; therefore, responses might have been subject to respondent bias. Similarly, because responses associated with symptoms, conditions, and exposures might have occurred over a protracted time, survey responses might also be subject to recall bias.
Despite these study limitations, B. henselae infections seemed to be more common in patients who reported blurred vision, numbness in the extremities, and previous consultation with a neurologist before referral  to the rheumatologist. In a case series of 14 patients, the following were reported by 50% of patients infected with a Bartonella species, specifi cally B. henselae, B. vinsonii subsp. berkhoffi i, or both: memory loss, numbness or a loss of sensation, balance problems, and headaches (10). Another 6 B. henselae-bacteremic patients reported seizures, ataxia, memory loss, and/or tremors; 1 of these patients was co-infected with B. vinsonii subsp. berkhoffi i, and another was positive for B. henselae by PCR after enrichment of cerebrospinal fl uid in BAPGM (23). An enrichment culture approach also identifi ed an association between intravascular infection with B. vinsonii subsp. berkhoffi i genotype II and B. henselae and neurologic symptoms in a veterinarian and his daughter (12). Symptoms in the father included progressive weight loss, muscle weakness, and lack of coordination; symptoms in  the daughter were headaches, muscle pain, and insomnia. For each patient, after repeated courses of antimicrobial drugs, blood cultures became negative, antibody titers decreased to nondetectable levels, and all neurologic symptoms resolved. Although no symptoms were statistically associated with B. koehlerae infection, patients infected with B. koehlerae were more likely to have previously consulted an infectious disease physician. Of the 54 B. koehlerae patients with a positive PCR result, 54% reported a prior diagnosis of Lyme disease (n = 25), bartonellosis (n = 3), or babesiosis (n = 1). Fatigue, insomnia, memory loss, and joint and muscle pain were frequent complaints among those with a positive PCR result for B. koehlerae, but these symptoms did not differ in frequency from those in patients with negative PCR. Similar symptoms were previously reported in a small case series involving B. koehlerae-bacteremic patients (13). Peripheral visual defi cits, sensory loss, and hallucinations resolved in a young woman after antimicrobial drug treatment for B. koehlerae infection (30). Because of the small number of patients with positive PCR results for B. vinsonii subsp. berkhoffi i, we restricted the multivariate analysis to those with positive results for B. henselae and B. koehlerae. Because limited sample size affected our ability to conduct multivariate analysis to control for potential confounders for B. vinsonii subsp. berkhoffi i positivity, the χ 2 associations with B. vinsonii subsp. berkhoffi i positivity should be interpreted with caution.
Although the pathogenic relevance of the high Bartonella spp. seroprevalence and bacteremia in this patient population are unclear, these results justify additional prospective studies involving more narrowly defi ned patient and control populations. Of the 92 patients infected with B. koehlerae, B. henselae, or B. vinsonii subsp. berkhoffi , 69 (75%) had at least 1 discordant IFA assay result for Bartonella spp. antigen seroreactivity and only 34 (30.6%) had a concordant species-specifi c PCR and IFA result. Also, consistent with previous study fi ndings (15), the PCRs depicted in Figure 2 illustrate an increased likelihood of positivity if blood, serum, and enrichment blood cultures are independently tested. According to these and previous results (7,18,31,32), a subset of Bartonella spp.-bacteremic patients could be anergic and might not produce a detectable IFA response, or alternatively, the substantial antigenic variation among various Bartonella strains might result in false-negative IFA assay results for some patients. In a study on Bartonella serology conducted by the Centers for Disease Control and Prevention, IFA cross-reactivity among Bartonella species occurred in 94% of patients with suspected catscratch disease (33). Despite the lack of concordance between serologic results and BAPGM enrichment PCR results, most (185 [62.5%]) patients in our study were seroreactive to Bartonella spp., suggesting prior exposure to >1 Bartonella spp. Because serologic cross-reactivity to Chlamydia spp. and Coxiella burnettii antigens has been reported, exposure to these or other organisms might have contributed to the high seroprevalence. In a previous study involving 32 healthy volunteers and patients at high risk for Bartonella spp. bacteremia, seroprevalence rates for B. henselae, B. koehlerae and B. vinsonii subsp. berkhoffi i genotypes I and II were 3.1%, 0%, 0,%, and 50%, respectively, for the healthy population compared with 15.6%, 9.2%, 19.8%, and 28.1%, respectively, for the high-risk population (15). Although in that study and the study reported here, the same test antigens and identical IFA assays were used and the same research technologist interpreted the results, the overall seroprevalence in the study reported here was higher than that among highrisk patients with extensive arthropod or animal contact (49.5%) and differed substantially from serologic results from healthy volunteers (15). However, in the study reported here, a large portion of the population (34.5%) was also seroreactive to B. vinsonii berkhoffi i genotype II. Immunophenotypic properties giving rise to seroreactivity to this particular antigen among healthy control and patient populations have not been clarifi ed but could be related to polyclonal B-cell activation, commonly found in patients with rheumatologic or chronic infl ammatory diseases.
It is becoming increasingly clear that no single diagnostic strategy will confi rm infection with a Bartonella sp. in immunocompetent patients. Before the current study, we primarily used BAPGM enrichment blood cultures and PCR to test symptomatic veterinarians, veterinary technicians, and wildlife biologists, who seem to be at occupational risk for Bartonella sp. bacteremia because of animal contact and frequent arthropod exposure (10)(11)(12)(13)(14)(15)23). Cats are the primary reservoir hosts for B. henselae  (4,6,29,34). Although infrequent when compared with cat transmission of B. henselae resulting in classical cat-scratch disease, dogs have been implicated in the transmission of B. vinsonii subsp. berkhoffi i and B. henselae to humans (35,36). The predominant symptoms reported among occupationally at-risk patient populations have included severe fatigue, neurologic and neurocognitive abnormalities, arthralgia, and myalgia (10)(11)(12)(13)23). In the study reported here, dog (85%) and cat (68%) contact were reported by most respondents; however, no associations were found between infection with a Bartonella sp. and contact with a specifi c animal. Similarly, exposure to mosquitoes, ticks, fl eas, and biting fl ies were all reported by >50% of the study population. The results of this study support documentation of Bartonella spp. bacteremia in patients seen by a rheumatologist in a Lyme diseaseendemic area and provides the basis for future studies to ascertain the prevalence of Bartonella spp. in patients with rheumatic and neurologic symptoms. . is chief scientifi c offi cer for Galaxy Diagnostics, a newly formed company that provides diagnostic testing for the detection of Bartonella species infection in animals and human patients. R.G. Maggi has lead research efforts to optimize the BAPGM platform and is the scientifi c technical advisor for Galaxy Diagnostics. R. Mozayeni was the attending physician for the patients described in this study and has recently joined Galaxy Diagnostics as the chief medical offi cer. All other authors have no potential confl icts.