• Open Access

Bartonella DNA in the Blood and Lymph Nodes of Golden Retrievers with Lymphoma and in Healthy Controls

Authors

  • A.W. Duncan,

    1. Intracellular Pathogens Research Laboratory, Center for Comparative Medicine and Translational Research, North Carolina State University, College of Veterinary Medicine, Raleigh, NC,
    Search for more papers by this author
  • H.S. Marr,

    1. Intracellular Pathogens Research Laboratory, Center for Comparative Medicine and Translational Research, North Carolina State University, College of Veterinary Medicine, Raleigh, NC,
    Search for more papers by this author
  • A.J. Birkenheuer,

    1. Intracellular Pathogens Research Laboratory, Center for Comparative Medicine and Translational Research, North Carolina State University, College of Veterinary Medicine, Raleigh, NC,
    Search for more papers by this author
  • R.G. Maggi,

    1. Intracellular Pathogens Research Laboratory, Center for Comparative Medicine and Translational Research, North Carolina State University, College of Veterinary Medicine, Raleigh, NC,
    Search for more papers by this author
  • L.E. Williams,

    1. Department of Clinical Sciences, North Carolina State University, College of Veterinary Medicine, Raleigh, NC,
    Search for more papers by this author
  • M.T. Correa,

    1. Department of Population Health and Pathobiology, North Carolina State University, College of Veterinary Medicine, Raleigh, NC
    Search for more papers by this author
  • E.B. Breitschwerdt

    1. Intracellular Pathogens Research Laboratory, Center for Comparative Medicine and Translational Research, North Carolina State University, College of Veterinary Medicine, Raleigh, NC,
    Search for more papers by this author

  • This work was performed at the Intracellular Pathogens Research Laboratory in North Carolina State University, College of Veterinary Medicine, Raleigh, NC. Abstract presented at the 2006 Joint Meeting: 20th National Meeting of the American Society for Rickettsiology and the 5th International Conference on Bartonella as Emerging Pathogens.

  • The research was funded in part by the American Kennel Club-Canine Health Foundation, Bayer Animal Health, the Golden Retriever Foundation, IDEXX Laboratories, and the State of North Carolina.

Corresponding author: Dr Edward B. Breitschwerdt, Department of Clinical Sciences, College of Veterinary Medicine, North Carolina State University, 4700 Hillsborough Street, Raleigh, NC 27606; email: ed_breitschwerdt@ncsu.edu.

Abstract

Background: Although lymphoma is the most common neoplastic process reported in dogs, its precise etiology is unknown. Golden Retrievers are more likely to develop lymphoma, suggesting a breed predisposition; however, other factors, including environment, immunity, and infection, are likely contributors to oncogenesis.

Hypothesis: We hypothesized that the development of lymphoma in Golden Retrievers may be associated with vector-borne infections, specifically Bartonella, Anaplasma, or Ehrlichia species infections.

Animals: Golden Retrievers with lymphoma and healthy Golden Retrievers from across the United States were recruited for study participation.

Methods: A matched, case-control study was performed to determine the association of lymphoma and the presence of Bartonella, Anaplasma, and Ehrlichia species in serum, blood, and lymph node aspirates.

Results: Using PCR analyses and DNA sequencing, single and coinfections with Bartonella henselae, Bartonella elizabethae, Bartonella quintana, and/or Bartonella vinsonii (berkhoffii) were detected in the blood and lymph node aspirates of Golden Retrievers with lymphoma (5/28 dogs, 18%) and in healthy Golden Retrievers (10/56 dogs, 18%); no Anaplasma or Ehrlichia DNA was detected in any dog. When compared with dogs with lymphoma, a higher (P <.001) proportion of healthy Golden Retrievers were receiving monthly acaricide treatments (2.6 times higher).

Conclusions and Clinical Importance: Bartonella DNA can be detected in blood and lymph nodes; importantly, in this report, Bartonella was detected in the same proportion of clinically healthy dogs and dogs with lymphoma. Longitudinal studies should be conducted to determine the mode of transmission of Bartonella in dogs, whether lymphatic infection is persistent, or whether these bacteria may contribute to the development of lymphoma.

Lymphoma is a common, life-threatening hematopoietic cancer of dogs, which results from the malignant transformation of lymphocytes. Lymphoma has been estimated to affect between 1.3 and 3.3% of dogs each year1 and accounts for up to 25% of neoplasia in dogs as reported by some veterinary institutions.2 As with other forms of cancer, the etiology of lymphoma is thought to be multifactorial. Although the precise etiology of lymphoma in dogs is unknown, it is hypothesized that various genetic, environmental, immunological, toxic, and infectious factors may all play important roles in etiopathogenesis.3 Recently, studies have focused on the identification of possible predisposing genetic factors that can contribute to the development of cancer. When compared with the general veterinary hospital population, Golden Retrievers are at an increased risk for developing lymphoma (relative risk=2.8, 95% confidence interval=1.8–4.2, P <.001).4 Golden Retrievers develop significantly (P <.001) more B-cell and T-cell lymphoma when compared with other dog breeds.5 Further, unique cytogenetic abnormalities, including chromosomal gains and losses, have been identified in Golden Retrievers with lymphoma, changes that are recognized significantly (P <.001) less frequently or not at all in other dog breeds with lymphoma.5 Although genetics is a prominent factor contributing to lymphoma in Golden Retrievers, it is hypothesized that other interactions between environmental and immunological factors are likely contributors to oncogenesis.

The 1998 Golden Retriever Club of America–National Health Survey demonstrated a statistically significant (P≤ .05) decrease in the frequency of lymphoma among dogs that had been treated with acaricides.6 However, to date, no studies have been conducted to determine whether chronic infection with vector-borne organisms may contribute to the development of lymphoma in dogs. On rare occasions, Ehrlichia canis infection has been associated with the development of clonally expanded T-cell populations in dogs without overt lymphoid malignancies; however, the significance of these lymphocyte populations is currently unknown.7,8

During the past decade, researchers have provided an expanding body of evidence to support a potential role for infectious agents, such as viruses, mycoplasma, bacteria, and protozoa, as cofactors in the development of cancer in humans.9 In fact, the World Health Organization recently estimated that one-fifth (20%) of all cancers worldwide are caused by chronic infection with a virus, bacteria, or protozoa.10 Based on a plausible molecular pathogenesis of endothelial cell proliferation and angiogenesis, several Bartonella species, such as B. quintana and B. henselae, are considered to be bacterial pathogens capable of causing or contributing to carcinogenesis.9,11

Bacteria of the genus Bartonella are fastidious, pleomorphic, Gram-negative, vector-borne (fleas, lice, sandflies, and potentially biting flies and ticks) aerobic bacilli with more than 20 described species or subspecies.12,13Bartonella are highly adapted bacteria that maintain persistent intracellular infections in a wide variety of cell types in humans and animals.14–17 Recognized as important emerging pathogens in human and veterinary medicine, Bartonella species have been implicated in several disease processes in dogs, such as endocarditis, granulomatous lymphadenitis, peliosis hepatis, and granulomatous hepatitis.18–21

In the current study, we hypothesized that the development of lymphoma in Golden Retrievers may be associated with chronic infection with 1 or more vector-borne pathogens, specifically Bartonella, Anaplasma, or Ehrlichia species. We performed a matched, case-control study to determine the association of lymphoma with the presence of these vector-borne organisms in serum, blood, and lymph node aspirate samples. In addition, we determined the association of several purported risk factors for Bartonella infection and the presence of these organisms.

Materials and Methods

Study Participants

Golden Retrievers were identified by veterinarians across the United States. Samples were collected between November 2004 and December 2006 and included a serum sample for serological analyses, an EDTA-anticoagulated whole blood sample, and a lymph node aspirate for DNA extraction and PCR analysis. Participation required informed consent from each owner indicating that he/she understood the goals and procedures for this study. Protocols were conducted in accordance with North Carolina State University (NCSU) Institutional Animal Care and Use Committee approval (protocol number 05-048-B). Testing was conducted by personnel at the NCSU Intracellular Pathogens Research Laboratory (NCSU-IPRL).

Case Dogs

Golden Retrievers diagnosed with lymphoma by a board-certified pathologist using cytological or histopathological methods were evaluated as case dogs. Our minimum entry criteria included no prior diagnosis or treatment for lymphoma, no antibiotic administration within 14 days before sample collection (or 30 days for azithromycina), and no treatment with corticosteroids or other immunosuppressive drugs within 14 days of sample collection.

Control Dogs

For each Golden Retriever with lymphoma, 2 healthy Golden Retrievers were sampled in a similar manner and served as unaffected controls. All controls were matched by geographic region (residing within a 100-mile radius of the respective case dog), age (± 24 months), and, when possible, by sex. Controls were clinically healthy and lacked evidence of lymphadenopathy, as determined by a physical assessment by their veterinarian. Control dogs had not received any immunosuppressive drugs or antibiotics within 14 days before sample collection (or 30 days for azithromycin).

Sample Collection

Serum, EDTA-anticoagulated whole blood, and lymph node aspirates were submitted to the NCSU-IPRL. The protocol for obtaining lymph node aspirates was as follows: (1) before obtaining the aspirate, 1 mL of sterile saline was injected into a sterile, unopened, red-topped serum tube,b (2) the lymph node was aspirated using an 18- to 20-G needle and a 12 cm3 syringe, and (3) using the same needle and syringe, the saline was drawn from the tube into the syringe containing the aspirate and rinsed 3–5 times to wash the lymphoid cells from the needle into the collection tube.

Questionnaire

All owners completed a self-administered questionnaire addressing various conditions that were considered as potential risk factors for Bartonella, Anaplasma, and Ehrlichia infection, such as rural/suburban/urban residence and flea/tick/cat exposure. Housing conditions were subjectively assessed by the owners based on 3 categories: indoor only, outdoor only, or indoor/outdoor. Owners were also queried about the use and regularity of acaricide application, along with the dog's travel history. A copy of the questionnaire is available by request through the corresponding author.

Serology

Serum samples were analyzed for IgG antibodies to B. henselae, B. vinsonii (berkhoffii), and E. canis antigens by an indirect immunofluorescence assay (IFA) as previously described (Solano-Gallego et al, 2004). Reciprocal titers ≥64 were considered seroreactive. Serum samples were screened for the presence of Dirofilaria immitis antigens and antibodies to E. canis, Anaplasma phagocytophilum, and Borrelia burgdorferi using a commercial assay kit.c

Preparation of DNA

Total DNA was extracted from blood samples and lymph node aspiratesd according to the manufacturer's instructions. After extraction, DNA concentration and quality were quantified by spectrophotometry.e Blood culture was not performed because of the low sensitivity of traditional methods such as plating to blood agar; further, novel liquid pre-enrichment culture methods were undergoing characterization and optimization during the study period.22

Bartonella Genus-Specific PCR

All blood and lymph node aspirate samples were screened for the presence of Bartonella genus DNA using 2 separate assays. Samples were first screened by real-time PCR, amplifying an approximate 120-base pair fragment of the 16S–23S intergenic transcribed spacer (ITS) region. PCR conditions were optimized using a 50-μL reaction volume containing 25 μL of SYBR Green Master Mix,f 15 pmol of each primer (Table 1), 19 μL of molecular grade water, and 1–5 μL of DNA template, according to the DNA concentration determined for each sample (5–100 ng of DNA/reaction). Real-timeg cycling conditions were 95 °C for 5 minutes, followed by 45 cycles of 95 °C for 30 seconds, 56 °C for 30 seconds, 72 °C for 30 seconds, and a final extension of 72 °C for 5 minutes. In addition to melt-curve analysis, products were visualized by 2.5% agarose gel electrophoresis containing 0.2 μg of ethidium bromide/mL under transilluminating ultraviolet light. The limit of detection for the ITS assay was determined to be 5 copies of target-containing plasmid per reaction.

Table 1.    Bartonella genus- and species-specific primer sequences for the 16S–23S intergenic transcribed spacer (ITS) and heme-binding protein (Pap31) regions, and Anaplasma and Ehrlichia genus-specific primer sequences for the 16S rRNA region.
Primer NamePrimer Sequence
  • a

    Primer designed to amplify other Bartonella species, but to exclude B. henselae amplification.

Bartonella genus-specific, screening level primers
Bartonella genus ITS forward primer5′ - AGATGATGATCCCAAGCCTTCTGG - 3′
Bartonella genus ITS reverse primer5′ - GATAAACCGGAAAACCTTCCC - 3′
Bartonella species-specific primers
B. henselae-specific ITS forward primer5′ - CAAGCCTTCTGGCGATCTAG - 3′
Common Bartonella species ITS forward primera5′ - CAAGCCTTCGGGCGATCTCT - 3′
Bartonella species ITS reverse primer5′ - GATAAACCGGAAAACCTTCCC - 3′
B. henselae-specific Pap31 forward primer5′ - TGGGCTGACAGAGAAGACG - 3′
Bartonella species Pap31 reverse primer5′ - CACCACCAGCAACATAAGGC - 3′
Anaplasma and Ehrlichia genus-specific primers
Anaplasma and Ehrlichia genus reverse primer5′- TATAGGTACCGTCATTATCTTCCCTATTG - 3′
Anaplasma and Ehrlichia genus forward primer5′ - GCAAGCYTAACACATGCAAGTCGA - 3′

The 2nd screening for Bartonella genus DNA utilized a conventional PCR targeting the heme-binding Pap31 gene as described elsewhere23; master mix volumes were as described for the ITS PCR. The limit of detection for the conventional Pap31 method is 10 copies of target-containing plasmid per reaction.

Bartonella Species-Specific PCR

Samples that were positive for Bartonella DNA using either of the genus-level analyses were further analyzed using real-time PCR designed to detect coinfections with more than 1 Bartonella species in the same DNA sample (Table 1). One ITS primer was designed to detect B. henselae only; a 2nd ITS primer was designed to amplify other common Bartonella species, but to exclude amplification of B. henselae. Similarly, a Pap31 assay was designed to detect only B. henselae. All reaction master mixes, cycling conditions, and methods of amplicon detection were similar to those used in the ITS genus-specific amplification. The limits of detection for these species-specific assays were determined to be 5 copies of target-containing plasmid per reaction.

Anaplasma and Ehrlichia Genus-Specific PCR

All blood and lymph node aspirate samples were tested for the presence of Anaplasma and Ehrlichia DNA using 16S rRNA as a gene target (Table 1). Reaction master mixes were prepared in 25 μL volumes with 1–5 μL of DNA template as described above for Bartonella PCR. Conventional PCR conditionsh were 95 °C for 10 seconds, followed by 55 cycles of 94 °C for 15 seconds, 63.8 °C for 15 seconds, 72 °C for 18 seconds, and a final extension of 72 °C for 2 minutes. PCR products were visualized using 2% agarose gel electrophoresis containing 0.2 μg of ethidium bromide/mL under transilluminating ultraviolet light. The limit of detection is 5 copies of the target-containing plasmid per reaction.

PCR Controls

In order to prevent PCR amplicon contamination, DNA extraction, reaction setup, PCR amplification, and amplicon detection were all performed in separate areas. Positive and negative controls were used in all processing steps, including DNA extraction. Amplification of a fragment of the glyceraldehyde-3-phosphodehydrogenase (GAPDH) pseudogene was performed, as previously described, to demonstrate integrity of the DNA and the absence of PCR inhibitors.24 All samples had to be GAPDH-PCR positive to be included in the subsequent PCR analyses for Bartonella, Anaplasma, or Ehrlichia.

DNA extracted from the whole blood of a Golden Retriever that was consistently Bartonella, Anaplasma, and Ehrlichia PCR-negative was used as a negative control. Plasmid clones of a partial sequence of (1) the ITS region of B. henselae San Antonio 2 strain (GenBank Accession number: AF369529), B. quintana Fuller strain (M11927), or B. vinsonii (berkhoffii) Type I (AF167988); (2) the Pap31 region of B. henselae San Antonio 2 strain (DQ529248); or (3) the 16S rRNA region of E. canis (CP000107) were resuspended in Bartonella-, Anaplasma-, and Ehrlichia-negative canine DNA and were used as positive controls for the Bartonella ITS, Bartonella Pap31, and Anaplasma/Ehrlichia 16S PCR analyses, respectively.

Sequencing of Amplicons

All amplicons were sequenced, in forward and reverse directions, to identify species.i Sequence analysis and alignment with GenBank sequences were performed.j

Statistical Analyses

Descriptive statistics were obtained for the following variables: exposure to ticks, fleas, and/or cats, each coded as no/yes; monthly acaricide use coded as no/yes; indoor/outdoor status versus indoor only; and urban/suburban or rural residence. Putative risk factors were evaluated using 2 different statistical models, one assessing the differences between dogs with lymphoma and healthy controls, and another considering the differences between PCR-positive and PCR-negative dogs. The PCR status variable was created by combining the PCR results of blood and lymph node aspirates; dogs were considered to be positive for Bartonella, Anaplasma, or Ehrlichia if either the blood or lymph node aspirate sample was PCR positive.

We developed a conditional logistic regression main-effects-only modelk and utilized a hierarchical backward elimination algorithm to determine the association between the previously described risk factors and (1) lymphoma status and (2) Bartonella, Anaplasma, and/or Ehrlichia PCR status. As an alternative statistical analysis, we evaluated the data using the Mantel-Haenszel method for multiple matched controls to test whether the proportions of dogs with lymphoma and healthy dogs differed with respect to the previously described risk factors; this method analyzes the proportions for only 1 variable at a time.25 Statistical significance was considered at a P-value of ≤ .05.

Results

Study Participants

Complete sample sets were submitted for 28 Golden Retrievers with lymphoma and their 56 age-, sex-, and geographically matched, healthy, Golden Retriever controls. Samples were submitted from veterinarians across the United States, with 11 of the 28 dogs with lymphoma residing in North Carolina or Virginia.

Of the 28 dogs with lymphoma, 12 were female (10 spayed, 1 intact, 1 unspecified) and 16 were male (14 neutered, 2 intact); median age was 7.5 years (range, 3–12 years). Of the 56 control dogs, 26 were female (17 spayed, 9 intact) and 30 were male (26 neutered, 4 intact); median age was 6.8 years (range, 2–13 years). No differences were observed between the groups when evaluating spay/neuter status or travel history.

Descriptive data, as provided by the owners in a completed questionnaire, for the dogs with lymphoma (n = 28) were as follows: 15 dogs with previous tick exposure and 13 without a history of tick exposure; 9 dogs with previous flea exposure, 18 without a history of flea exposure, and 1 was not reported; 14 dogs received monthly acaricide treatments and the remaining 14 did not receive monthly acaricide treatments; 9 dogs were described as indoor only, 1 dog was outdoor only, and 18 dogs had free access to indoor/outdoor environments; and 8 dogs had at least 1 cat in the household, 18 reported no cat(s) in the household, and 2 questionnaires had no responses. Data for the clinically healthy dogs (n=56) were as follows: 19 dogs with previous tick exposure and 37 without a history of tick exposure; 8 dogs with previous flea exposure and 48 without a history of flea exposure; 36 dogs received monthly acaricide treatments, 19 dogs did not receive monthly acaricide treatments and 1 was not reported; 13 dogs were described as indoor only, 42 dogs had free access to indoor/outdoor environments, and 1 was not reported; and, lastly, 23 dogs had at least 1 cat in the household, 31 reported no cat(s) in the household, and 2 questionnaires had no responses.

Serology

Of the 84 Golden Retrievers surveyed, 4 (5%) were seroreactive to Bartonella antigens, 1 dog with lymphoma (1/28, 4%) and 3 healthy controls (3/56, 5%). The dog with lymphoma was seroreactive to B. henselae and B. vinsonii (berkhoffii) at reciprocal titers of 2,048 and 64, respectively. Among the 3 healthy dogs, 2 were seroreactive to B. henselae only, whereas the remaining dog was seroreactive to both antigens; reciprocal titers were 64 or 128. The remaining 80 dogs were not seroreactive to B. henselae and B. vinsonii (berkhoffii).

Of the 82 dogs surveyed for which there was sufficient serum for ELISA testing, 3 dogs with lymphoma (3/27, 11%) and 6 control dogs (6/55, 11%) were seroreactive to B. burgdorferi; all dogs had either resided in or regularly traveled to northeastern Lyme-endemic states. One dog with lymphoma (1/27, 4%) and 4 healthy dogs (4/55, 7%) were seroreactive to A. phagocytophilum antigens, and all 5 currently reside or have previously resided in Connecticut, an A. phagocytophilum-endemic state. D. immitis antigens were not found in any of the 82 dogs. E. canis antibodies were not detected by IFA (n=84) or by ELISA (n=82).

Bartonella PCR Analyses

Five of 28 dogs with lymphoma (18%) and 10 of 56 healthy dogs (18%) were positive for Bartonella by PCR (Table 2). Bartonella DNA was detected in 2 blood samples from dogs with lymphoma and in 2 blood samples from the healthy Golden Retrievers; in contrast, Bartonella DNA was detected in 5 and 8 lymph node aspirates from dogs with lymphoma and healthy Golden Retrievers, respectively.

Table 2.    Bartonella PCR and DNA sequence data for all 15 PCR-positive Golden Retrievers in the study.
No.SexAge
(year)
City, StateSample TypeBartonella ITSBartonella
henselae
ITS
Pap31B. henselae
Pap31
  1. Blood and lymph node aspirate samples were screened for Bartonella bacteria by targeting a fragment of the 16S–23S intergenic transcribed spacer (ITS) region and the Pap31 gene. All PCR-positive samples were confirmed by DNA sequencing.

  2. LNA, lymph node aspirate; M, male, F, female, I, intact, N, neutered, S, spayed; QNS, quantity not sufficient for testing.

Dogs with lymphoma
1M-N8Fairfield, NJBloodB. henselae, B. vinsonii (berkhoffii)B. henselaeB. henselaeB. henselae
   LNAB. henselaeB. henselaeB. henselaeB. henselae
2M-I4Dallas, TXBloodB. henselaeB. henselaeB. henselaeB. henselae
   LNAB. henselae, B. elizabethaeB. henselaeB. henselaeB. henselae
3M-N9Raleigh, NCBloodNegativeNegativeNegativeNegative
   LNANegativeNegativeB. henselaeB. henselae
4F-S9Chatham, VABloodNegativeNegativeNegativeNegative
   LNAB. henselaeB. henselaeNegativeNegative
5F-S8Raleigh, NCBloodNegativeNegativeNegativeNegative
   LNAB. quintanaNegativeNegativeNegative
Healthy dogs
1M-N4.5Medina, OHBloodNegativeNegativeNegativeNegative
   LNAB. quintanaNegativeQNSNegative
2M-N6Akron, OHBloodB. henselaeB. henselaeB. henselaeB. henselae
   LNANegativeNegativeNegativeNegative
3F-S3Castro Valley, CABloodB. henselaeB. henselaeNegativeNegative
   LNANegativeNegativeNegativeNegative
4M-N5Rochester, NYBloodNegativeNegativeNegativeNegative
   LNAB. henselae, B. quintanaB. henselaeQNSB. henselae
5F-I7Prince George, VABloodNegativeNegativeNegativeNegative
   LNAB. henselaeB. henselaeNegativeB. henselae
6M-N2Richmond, VABloodNegativeNegativeNegativeNegative
   LNAB. quintanaNegativeQNSNegative
7M-N10Waterford, CTBloodNegativeNegativeNegativeNegative
   LNAB. henselaeB. henselaeB. henselaeB. henselae
8M-N10Lyme, CTBloodNegativeNegativeNegativeNegative
   LNAB. henselae, B. quintanaB. henselaeB. henselaeB. henselae
9F-S7Clayton, NCBloodNegativeNegativeNegativeNegative
   LNAB. henselaeB. henselaeNegativeNegative
10M-N2Greenwich, CTBloodNegativeNegativeNegativeNegative
   LNAB. henselaeB. henselaeNegativeB. henselae

Anaplasma and Ehrlichia PCR Analyses

All blood and lymph node aspirate samples screened for Anaplasma and Ehrlichia DNA were negative.

PCR Controls

Contamination was not detected in any of the negative control samples at any stage of processing or at any time during the study. As determined by the successful amplification of GAPDH, no PCR inhibitors were present in any of the samples that were negative by PCR.

Questionnaire

Because of the distribution of the data, we were unable to develop a conditional logistic regression model to evaluate the association between lymphoma status and the proposed risk variables. Using the Mantel-Haenszel method, no differences were observed in the proportions of dogs with lymphoma and previous tick exposure; however, when considering the monthly use of acaricide as reported by the owners, the proportion of healthy dogs receiving acaricide was approximately 2.6 times higher (P <.001) than dogs with lymphoma. Further, the proportion of healthy dogs classified by their owners as indoor/outdoor dogs was approximately 2.3 times higher (P <.001) than dogs with lymphoma.

In the 2nd statistical model, the results of conditional regression suggest (P value for the model=.046) that dogs that were PCR positive for Bartonella DNA were more likely to have previous tick exposure (odds ratio: 3.4; 95% confidence interval for the odds ratio: 0.7–17.1; P-value = .13) and were classified by their owners as indoor/outdoor dogs (odds ratio: 3.1; 95% confidence interval for the odds ratio: 0.9–10.3; P-value=.07) when compared with dogs that were negative for Bartonella DNA. Despite the statistical model retaining these factors, the confidence intervals include one, and therefore they were considered to be marginally significant.

As executed, this study had a power (1 −β=0.80) to detect a 20% difference in the presence of Bartonella DNA in dogs with lymphoma and healthy dogs; the fact that we did not detect a difference between the proportions of Bartonella DNA in the 2 groups would suggest that Bartonella infection is not a major cofactor in the development of lymphoma in Golden Retrievers. However, when considering the monthly use of acaricide as reported by the owners, the proportion of healthy dogs receiving acaricide was approximately 2.6 times higher (P <.001) than acaricide use reported in dogs with lymphoma.

Discussion

Using PCR and DNA sequencing, we detected 4 different Bartonella species in blood and lymph node aspirates of the Golden Retrievers surveyed: B. henselae, B. elizabethae, B. quintana, and B. vinsonii (berkhoffii). Although there were no differences in the prevalence of PCR-positive dogs or the particular Bartonella species detected in Golden Retrievers with lymphoma when compared with age-, sex-, and geographically matched, healthy Golden Retrievers, the high prevalence of PCR-positive dogs (18%) was unexpected. In contrast, no Anaplasma or Ehrlichia species DNA was detected in any sample.

Based on the results obtained, there was no association between lymphoma and the evidence of vector-borne infection with Bartonella, Anaplasma, or Ehrlichia species as determined by serologic testing. There was also no difference in seroreactivity to A. phagocytophilum and B. burgdorferi among dogs with lymphoma and healthy dogs. Seroreactivity to A. phagocytophilum and B. burgdorferi was detected only in dogs that reside, resided, or regularly traveled to the northeastern United States, suggesting that these regionally defined pathogens would be less likely candidates if vector-borne organisms contribute to the development of lymphoma.

No differences were observed between dogs with lymphoma and those that were PCR positive for Bartonella or between dogs with lymphoma and dogs with previous tick exposure. When compared with dogs with lymphoma, more healthy dogs (P <.001) were receiving monthly acaricide treatments; importantly, this finding is similar to previous observations indicating a decrease (P≤ .05) in the frequency of lymphoma among Golden Retrievers treated with acaricides.6 These results suggest that another tick-associated factor may be involved in the development of lymphoma within this breed. Additionally, when considering the risk factors for detectable Bartonella DNA in the current study, tick exposure and owner's classification of indoor/outdoor status were similar to putative risk factors previously described for B. vinsonii (berkhoffii) seroreactivity.26 Although these 2 factors were marginally significant in the current conditional regression model, it should be noted that the lower side of the 95% confidence intervals for both odds ratio estimates was marginal in value, a finding likely to be associated with the small sample size and large variability in the data.

Before the development of PCR, diagnosis of Bartonella, Anaplasma, or Ehrlichia infection was dependent on microbiological isolation, blood smear examination, or serological analyses. However, these methods of diagnosis can be relatively insensitive when compared with molecular-based assays, and in certain instances confirm only prior exposure (serology) as compared with active infection (DNA detection by PCR). In the present study, seroreactivity by IFA did not correlate with the presence or absence of Bartonella as detected by PCR amplification and DNA sequencing. Although a lack of correlation between serology and PCR and/or blood-culture positive patients has been demonstrated in humans27,28 and dogs22,29 infected with Bartonella species, additional studies will be necessary to define discrepancies between Bartonella serology and PCR analyses.

To our knowledge, this is the 1st report in which Bartonella DNA was detected in the lymph nodes of healthy dogs or dogs with lymphoma. Regardless of case/control status, Bartonella species were detected more frequently in lymph node aspirates than in whole blood samples. It is possible that after intradermal inoculation of a Bartonella species by an arthropod vector or a scratch, the organisms enter dendritic cells, traffic to regional lymph nodes, and induce a chronic infection within the lymphatic system.17 In the context of disease causation, it is important to note that this study did not incorporate a longitudinal design, and therefore the authors are unaware of whether the healthy dogs seroconverted, subsequently developed Bartonella-related disease manifestations, were treated by their veterinarians, or progressed to lymphoma. Based on these data, longitudinal studies should be conducted to determine the mode of transmission of Bartonella in dogs, whether lymphatic infection is persistent, or whether these bacteria may contribute to the development of lymphoma.

Footnotes

aZithromax, Pfizer, New York, NY

bVacutainer, BD, Franklin Lakes, NJ

cCanine SNAP 4Dx Test, IDEXX Laboratories, Westbrook, ME

dQIAamp DNA Blood Mini-Kit, QIAGEN Inc., Valencia, CA

eND-1000, NanoDrop Technologies, Wilmington, DE

f2X SYBR GREEN Master Mix, Applied Biosystems, Foster City, CA

giCycler iQ Real-Time PCR System, BioRad, Hercules, CA

hMasterCycler epS gradient, Eppendorf, Westbury, NY

iDavis Sequencing, Davis, CA

jAlignX, Vector NTI Suite 6.0, InforMax, Inc., Frederick, MD

kEgret, Cytel Inc, Cambridge, MA

Acknowledgments

The authors gratefully acknowledge the assistance of the veterinarians who provided samples and the owners who allowed participation of their dogs in this study. Our appreciation is extended to Valerie Hendrickson, Dr Pedro Paulo Diniz, Julie Bradley, Dr Hunter Blanton, Dr Andrew Vaughan, Dr Neil Marrinan, Dr Julie Levy, Natalie Cherry, and the Golden Retriever Club of America for their assistance with the study.

Ancillary