Association between Bartonella species infection and disease in pet cats as determined using serology and culture

Abstract

This study's objective was to determine whether a relationship exists between infection or seropositivity to Bartonella species and clinical illness in cats. Blood samples were obtained for Bartonella species isolation and immunofluorescent antibody serology from 298 cats presenting to a tertiary referral hospital. Medical records were searched and the history, physical examination findings and the results of diagnostic testing relating to the visit at which Bartonella species testing was performed were recorded. Fifty-two (17%) samples were seropositive for Bartonella henselae, four (1%) for Bartonella clarridgeiae, and 57 (19%) for both organisms. Nineteen (6.4%) samples were culture positive, 17 for B henselae and two for B clarridgeiae. Gingivostomatitis was associated with Bartonella species isolation (P=0.001), but not seropositivity. There was no association with uveitis, neurologic signs, or chronic kidney disease, and a weak association between seropositivity and idiopathic lower urinary tract disease (feline interstitial cystitis) (P=0.05).

Bartonella species are fastidious, Gram-negative bacteria that reside within erythrocytes and endothelial cells. Domestic cats are the main reservoir for Bartonella henselae, the primary cause of cat scratch disease in humans, and experience recurrent periods of bacteremia following natural infection. The seroprevalence in the United States in one study of 577 cats ranged from 4 to 7% in the Midwest to 60% in the south-east, with an overall seroprevalence of 28%. ¹ Cats may also be infected with Bartonella clarridgeiae, which has also been reported as a cause of cat scratch disease in humans. ^2–4 Cats are generally considered to be subclinical carriers of Bartonella species, although inoculation of specific pathogen-free cats with Bartonella species has been associated with the development of fever, neurological signs, lymphadenopathy, and reproductive failure. ^5–8 The results of necropsy on experimentally infected cats has revealed a variety of inflammatory lesions including peripheral lymph node hyperplasia, splenic follicular hyperplasia, lymphocytic cholangitis, lymphocytic hepatitis, focal lymphoplasmacytic myocarditis, and interstitial lymphocytic nephritis. ^9,10

Because of the high prevalence of seropositivity and bacteremia in apparently healthy cats, it has been difficult to make associations between disease and positive test results for Bartonella species in the field. Conflicting results have been obtained regarding the association of Bartonella species infection and seropositivity and uveitis, neurologic disease, gingivostomatitis, and fever. ^11–18 Three case reports of endocarditis have been reported in cats associated with B henselae infection. ^19–21 In two instances, B henselae type I DNA was detected within diseased heart valve. ^19,20 In one study, sick cats seropositive for B henselae had an increased frequency of stomatitis and a variety of diseases of the urinary tract, although the types of urinary tract diseases were not ascertained. ¹⁵ Another study from the south-eastern United States found an association between seropositivity and hematuria. ²² In Japan, co-infection with Bartonella species and feline immunodeficiency virus (FIV) was associated with gingivostomatitis. ¹⁶ No study has examined whether a relationship exists between feline gingivostomatitis, urinary tract disease, or other clinical manifestations and Bartonella species bacteremia, as determined using bacterial isolation from blood. All previous studies have relied solely on the use of polymerase chain reaction (PCR) or serology for diagnosis of infection.

The purpose of this study was to determine whether a relationship exists between bacteremia and seropositivity to B henselae and clinical illness in cats, in the form of a retrospective, cross-sectional study evaluating cats presenting to a tertiary referral hospital.

Materials and Methods

Ethylenediaminetetraacetic acid (EDTA) anticoagulated blood samples were obtained over a 7-month period (May 2006–November 2006) from 298 cats that presented to the William R Pritchard Veterinary Medical Teaching Hospital (VMTH), at the University of California, Davis, for which a complete blood count (CBC), serum chemistry panel, and urinalysis were performed. Samples for Bartonella species testing were taken from blood remaining in EDTA tubes after blood required for the CBC had been removed. The blood was collected using standard venepuncture methods. Samples were stored at 4°C until processed, within 1–2 weeks of sample collection.

Approximately 18 months after sample collection, the electronic database of the VMTH was searched for each patient. Factors recorded consisted of the history (including administration of antimicrobial drugs or glucocorticoids within a week prior to presentation, as well as any history of chemotherapy), retrovirus status, the findings on physical examination, the results of diagnostic testing and medical imaging, and the ultimate clinical diagnosis relating to the visit at which testing for Bartonella species was performed. Chronic kidney disease was diagnosed based on a combination of findings including renal palpation, imaging of the abdomen, and biochemical and urinalysis testing (creatinine>2.2 mg/dl, blood urea nitrogen>33 mg/dl, in conjunction with isosthenuria). Cats diagnosed with polycystic kidney disease, renal lymphoma, and pyelonephritis were excluded from this group. Lower urinary tract disorders included urolithiasis (upper and lower urinary tract stones), or idiopathic feline lower urinary tract disease (hereafter referred to as feline interstitial cystitis [FIC]), which was a diagnosis made based on the presence of lower urinary tract signs and hematuria, and exclusion of other diagnoses including bacterial urinary tract infections, neoplasia, trauma, and urolithiasis based on the results of history, physical examination, urinalysis, urine culture, and radiography, with or without ultrasound imaging of the lower urinary tract. Cats with bacterial urinary tract infections and congenital anatomic abnormalities were excluded from the lower urinary tract group. A diagnosis of gingivostomatitis was based on the presence of chronic oral lesions involving the palate or region lateral to and involving the palatoglossal folds, usually also involving the premolar and molar regions. Periodontal disease did not fall into this category. Neurologic disease was diagnosed based upon the presence of neurologic signs in the absence of other specific diagnoses that might explain the signs, such as cryptococcosis, neoplastic disease, or systemic illness known to be associated with neurologic signs.

EDTA tubes were frozen at −70°C and plated a few days later on to 5% rabbit blood agar plates, as previously described. ²¹ The plates were incubated at 35°C with 5% CO₂ for 4 weeks and were examined 2–3 times a week for bacterial growth. The isolated strains were identified to be B henselae or B clarridgeiae by PCR/restriction fragment length polymorphism (RFLP) of the citrate synthase gene using HhaI, TaqI and MseI endonucleases. ²³ The number of colonies was counted and calculated as colony forming units (CFU/ml). The 16S rRNA gene type (type I or type II) was determined by PCR, using previously designed primers.24

Antibodies against B henselae and B clarridgeiae were detected using IFA testing, as previously described. ²⁵ The intensity of bacillus-specific fluorescence at dilutions of 1:32 and 1:64 was scored subjectively for each serum sample from 1 to 4 in 0.25 unit increments. A fluorescence score of ≥2 at 1:64 was defined as a positive result.26

The group of cats that were culture positive or seropositive for Bartonella species was compared with the group of cats that tested culture negative or seronegative for Bartonella species using Fisher's exact test to detect group differences using frequency data. A Gaussian distribution was determined using the D’Agostino and Pearson omnibus normality test. Continuous variables were compared using the Mann–Whitney U test. Significance was defined as P<0.05. All analyses were performed using standard statistical software (GraphPad Prism version 4.00, San Diego, CA). Where appropriate, results are expressed as mean±standard deviation (SD).

Results

Of the 298 cats, 196 (66%) were listed as domestic shorthair or domestic medium hair, 50 (17%) were domestic longhair and 12 (4%) were purebred crosses. Other breeds included Siamese (11, 3.7%), Persians (seven, 2.3%), Abyssinians (four, 1.3%), Himalayan (three, 1%) and two or fewer of American Shorthair, Birman, Burmese, Japanese Bobtail, Maine Coon, Manx, Norwegian Forest Cat, Ragdoll, Rex, and Scottish Fold. A total of 136 (46%) cats were intact female (one) or female spayed (135), and 162 (54%) were intact male (one) or male neutered (161). Age was known for 291 cats, and ranged from 0.25 to 21 years (median 10 years). Information regarding the results of retrovirus testing was provided in the record for 124 (42%) of the cats. Three (2.4%) of these 124 cats were seropositive for feline leukemia virus (FeLV), and five (4%) were seropositive for FIV.

A total of 113 (38%) of 298 samples were seropositive for Bartonella species. Of these, 52 (46%) were positive for B henselae alone, four (4%) were positive for B clarridgeiae alone, and 57 (50%) were seropositive for both organisms. There was no association between seropositivity, age, sex and retrovirus status (Table 1).

Table 1.

Comparison of signalment and retrovirus status between cats testing positive for Bartonella species using serology (fluorescence score≥2 at 1:64) and isolation, and those testing negative.

	Seropositivity		Isolation
	Positive	Negative	P	Positive	Negative	P
Age	9±4.8 (108)*	10±4.8 (183)	0.29	10.0±4.8 (18)	7±5.3 (273)	0.25
Male sex	59 (113)	103 (185)	0.63	10 (19)	152 (279)	1.00
FeLV	1 (47)	2 (77)	1.00	0 (10)	3 (114)	1.00
FIV	1 (47)	4 (77)	0.65	1 (10)	4 (114)	0.35

_*Number in parenthesis refers to the total number of cats for which the results of that variable were known.

B henselae or B clarridgeiae were isolated from 19 (6.4%) cats. Culture contamination did not prevent successful isolation of Bartonella species from any sample. Seventeen cats were infected with B henselae and two cats with B clarridgeiae. Of the B henselae isolates, 10 were B henselae type 2, six were B henselae type 1 and one cat was co-infected with type 1 and type 2. The magnitude of bacteremia ranged from 8 to 176,000 CFU/ml (median, 176 CFU/ml). There was no difference in the age, sex, or retrovirus status of culture positive and culture negative cats. Fluorescence scores were significantly higher in cats from which B henselae was isolated (median, 3) than cats from which Bartonella species were not isolated (median, 0) (P<0.001). Negative serologic test results were present in one cat infected with B henselae and one cat infected with B clarridgeiae. Fluorescence scores at 1:64 of <2 but ≥1 to both B henselae and B clarridgeiae were present in one cat infected with B henselae and one cat infected with B clarridgeiae.

Variables examined for an association with seropositivity to, and isolation of Bartonella species are shown in Table 2. The presence of gingivostomatitis was strongly associated with isolation of Bartonella species (odds ratio [OR]=14.6, 95% confidence interval [CI] 3.6–60.1), but not with seropositivity to Bartonella species. The association between gingivostomatitis and Bartonella species isolation was independent of seropositivity to FIV (P=0.009, Fisher's exact test, OR=10.8, 95% CI 2.4–49.5). Only 1/4 cats that had gingivostomatitis and was bacteremic was co-infected with FIV, this was also the only FIV infected cat that was seropositive for Bartonella species. None of the cats with gingivostomatitis had signs of, or a reported history of, upper respiratory disease. A diagnosis of FIC was weakly associated with Bartonella species seropositivity, but not bacteremia. There was no significant association between seropositivity and bacteremia and a diagnosis of uveitis, neurologic signs, urolithiasis, or chronic kidney disease (Table 1). One cat with neurologic signs had B clarridgeiae bacteremia but was only weakly seropositive. This cat presented with seizures and underwent extensive diagnostics, including magnetic resonance imaging and cerebrospinal fluid analysis, which revealed only a lesion consistent with an infarct in the cerebellum, which was felt not to explain the presenting signs. The other cat that was bacteremic but only weakly seropositive had bilateral anterior uveitis with secondary lens luxation in the left eye, and was infected with B henselae. Bloodwork, including retrovirus testing, was unremarkable. Cytology of an aqueocentesis sample showed mild mixed, including lymphoplasmacytic, inflammation. Two cats were bacteremic but seronegative; one had congestive heart failure associated with unclassified cardiomyopathy, and the other had gingivostomatitis.

Table 2.

Association between diagnosis and results of immunofluorescent antibody testing (fluorescence score≥2 at 1:64) or culture for Bartonella henselae and Bartonella clarridgeiae in 298 cats.

Diagnosis	Seropositivity			Isolation
	Positive (n=85)	Negative (n=213)	P value	Positive (n=19)	Negative (n=279)	P value
Stomatitis	5	4	0.13	4	5	0.001
Chronic kidney disease	21	46	0.65	1	66	0.09
Urolithiasis	5	10	0.77	0	15	0.61
FIC	5	3	0.045	1	7	0.41
Uveitis	1	4	1.00	1	4	0.29
Neurologic signs	7	9	0.17	1	15	1.00
Diabetes mellitus	4	6	0.47	0	10	1.00
Hyperthyroidism	11	13	0.06	1	23	1.00
Idiopathic hypercalcemia	2	6	1.00	0	8	1.00
Lymphosarcoma	11	21	0.42	2	30	1.00
Mast cell neoplasia	4	5	0.28	0	9	1.00
Sarcoma	7	8	0.14	0	15	0.61
Squamous cell carcinoma	5	7	0.33	0	12	1.00
Recent antimicrobial therapy	15	20	0.07	4	31	0.26
Recent glucocorticoid administration	16	22	0.06	3	35	0.72
Chemotherapy	5	10	0.77	0	15	0.61

Recent administration of antimicrobial drugs was not associated with seropositivity or Bartonella species isolation. Thirty-five of the 298 cats had a history of receiving antimicrobial drugs. Of these, 17 had been treated with a penicillin derivative, 10 with a fluoroquinolone, five with a cephalosporin, four with doxycycline, two with azithromycin and one each with clindamycin and metronidazole. Antimicrobial drugs used in bacteremic cats immediately prior to isolation of Bartonella species included clavulanic acid–amoxicillin (three cats), cefazolin (one), and metronidazole (one). No association was noted between seropositivity or bacteremia and recent administration of glucocorticoids or chemotherapeutic drugs.

Discussion

The results of this study support previous observations made using IFA serology that detected a possible link between infection with Bartonella species and feline gingivostomatitis. ^15,16 In one study, gingivostomatitis was associated with positive serology for both Bartonella species and FIV. ¹⁶ In the other study, many of the cats with positive Bartonella species serology were also FIV seropositive, although the association was not significant (P=0.06). ¹⁵ A confounding effect of FIV seropositivity was not investigated. In the study reported here, the association between gingivostomatitis and Bartonella species infection applied only to bacteremic cats and not to seropositive cats. A lack of correlation between seropositivity to Bartonella species and feline chronic gingivostomatitis was noted in another recent study, which employed enzyme-linked immunosorbent assay (ELISA) and Western immunoblotting techniques, but not isolation. ¹⁷ This study also found no association between PCR positivity on blood and oral swabs for Bartonella species and gingivostomatitis. The population of cats in the study came from a single, communally-housed, mixed source colony of 45 cats, nine of which had stomatitis. Not all cats with gingivostomatitis in the study reported here were bacteremic or seropositive for Bartonella species or FIV, supporting the involvement of other etiologies in the pathogenesis of this condition. Infection with feline calicivirus (FCV), but not Bartonella species, has also been linked to chronic gingivostomatitis in cats based on the use of PCR on gingival biopsy specimens. ^18,27,28 Concurrent shedding of FCV and feline herpesvirus 1 (FHV-1) was associated with feline chronic gingivostomatitis in another study from our institution. ²⁹ Although none of the cats with gingivostomatitis had evidence or past history of upper respiratory tract disease in this study, this does not eliminate the possibility of concurrent infection with upper respiratory viruses, and testing for the presence of FHV-1 and FCV was not performed.

In the study reported here, we did not detect an association between uveitis and bacteremia or seropositivity to Bartonella species, although the number of cats with uveitis was low. In another study of naturally infected cats, an association was found between the presence of uveitis, positive serum IgM titers, and antibodies to B henselae in aqueous humor, but no association was noted between uveitis and positive aqueous humor PCR results or positive serum ELISA results. ^11,13

Experimental inoculation of specific pathogen-free cats with Bartonella species results in the development of a variety of inflammatory renal, myocardial and hepatic lesions. ^9,10 We did not find an association between bacteremia or seropositivity to Bartonella species and chronic kidney disease. The diagnosis of chronic kidney disease was made on the basis of abdominal palpation, laboratory testing, imaging, and in some cases, necropsy. Additional cats with occult renal disease may have been overlooked in this study. Exclusion of cats with known diagnoses may have impaired our ability to identify Bartonella species as a co-factor for development of certain diseases. We did not evaluate associations for some diseases, such as inflammatory intestinal, hepatic, and myocardial disease because of the large number of underlying etiologies for these diseases and the fact that many cats did not receive diagnostic work-ups for these problems that included pathology.

We detected a weak association between FIC and seropositivity, but not bacteremia. Another study associated seropositivity with hematuria using multiple logistic regression analysis. ²² The association with FIC may reflect the presence of unidentified confounding factors that predispose to exposure to Bartonella species and development of lower urinary tract signs or hematuria. Alternatively, Bartonella species may play a causative or contributory role to signs of lower urinary tract disease in some cats. The number of cats with FIC in this study was low.

Experimental inoculation of research cats has been associated with fever, lymphadenopathy, and in some cats, neurological signs, such as disorientation, aggression, and an inappropriate response to stimuli. ^10,11 B henselae can infect feline fetal microglial cells. ³⁰ Studies examining naturally infected cats have shown no associations between seropositivity and neurologic signs. ¹⁴ Bartonella species DNA was amplified from the cerebrospinal fluid of cats with neurologic signs, ³¹ but this study lacked a control group to determine the significance of this finding in relationship to the presence of neurologic signs. We did not detect an association with neurologic signs, although some cats with neurologic signs did not have necropsies performed to evaluate for underlying causes such as neoplasia or ischemic lesions.

The prevalence of bacteremia was 6.4%. This was lower than the 24% reported in pet cats <2 years of age, the most likely vectors for zoonotic transmission, ³² but similar to the 4.4% reported in a previous study of pet cats tested at the same veterinary teaching hospital. ²³ The seroprevalence of 38% in this study was consistent with results obtained in a previous study of pet cats in California, ¹ but was higher than the prevalence of 14.7% for B henselae found in a recent study of cats presenting to our hospital from 2001 to 2003. ³³ Positive serology is associated with increasing age, ^23,15,32,33 an important confounder when assessing the relationship between age-related diseases such as chronic kidney disease and hyperthyroidism. The population of cats examined in this study had a high median age of 10 years, which may explain the lack of correlation between seropositivity and age.

Administration of penicillin derivatives in the week prior to illness did not prevent isolation of Bartonella species from some cats, despite the results of previous studies showing clearance of bacteremia in cats following treatment with amoxicillin and clavulanic acid–amoxicillin. ³⁴ Nevertheless, previous antimicrobial therapy may have suppressed detection of bacteremia in other cats in the study that had previously received antimicrobial drugs. Four of the 19 bacteremic cats were seronegative, with two having no detectable antibodies to B henselae or B clarridgeiae, emphasizing that the results of serology are not predictive of bacteremia, as noted previously for both IFA and enzyme immunoassay methodology. ^3,23,32

A limitation of this study was its retrospective nature, and medical records may not have included all necessary information regarding each cat. The history of, or presence of flea infestation was not evaluated because records often lacked this information. In general, the prevalence of flea infestation in our population of pet cats is quite low, which is illustrated by the low prevalence of Bartonella species bacteremia in the pet cat population (4%) compared with that in shelter cats (>50%), where flea infestation is quite common. ²³ Thus, flea infestation would be an unlikely confounding variable in our study. We chose not to evaluate indoor/outdoor status as many cats from indoor-only households lived with multiple other cats, the indoor/outdoor status of which was unknown. Testing for the presence of FHV-1 and FCV was not performed in the cats with gingivostomatitis. Multivariate analysis was not performed, because of the low number of cats in each group.

Despite these limitations, samples were collected from cats that presented to a tertiary referral institution, so records for each cat contained extensive and detailed information regarding history, physical examination findings, diagnostic test results and treatments administered. Records had been reviewed by residents or faculty members. Many cats had undergone extensive diagnostic investigations to establish the underlying cause of their illness. The retrospective review of the medical record 18 months after sample collection increased the likelihood of a diagnosis being made, as many cats returned for follow-up evaluations. This study supports the need for prospective studies including multivariate analysis to further investigation of a connection between Bartonella species bacteremia and gingivostomatitis in cats.