Infectious endocarditis (IE), although infrequently diagnosed, is associated with severe morbidity and high case fatality rates in dogs and cats. Bartonella vinsonii subsp berkhoffii was cultured from the blood of a dog with endocarditis1 and, subsequently, other species of Bartonella, including Bartonella quintana, Bartonella clarridgeiae, Bartonella henselae, and Bartonella koehlerae, have been associated with IE or myocarditis in some dogs.2–5Bartonella spp. associated cardiac disease also occurs in cats.6,7 Most dogs diagnosed with Bartonella spp. IE have resided in temperate regions of the United States. This distribution might be because of a true geographic association with the syndrome, or alternately, it might be because of regional differences in frequency of veterinarians testing for Bartonella spp. based on perception that transmission of these bacteria occurs infrequently or not at all in regions with low vector distributions. For example, the seroprevalence and bacteremia rates for B. henselae in feral cats of Colorado, a state with low risk for Ctenocephalides felis infestation, are 6.3% and 0%.8 Thus, B. henselae-associated disease in this state and region is assumed to be rare in companion animals.
The purpose of this retrospective study was to interrogate heart tissues of dogs residing in Colorado and Wyoming that died with suspected IE for the DNA of Bartonella spp. or other bacteria using polymerase chain reaction assays (PCR assays).
The medical records system at Colorado State University's Veterinary Teaching Hospital (CSU-VTH) was searched for dogs with a clinical diagnosis of endocarditis that had been admitted from January, 1990 to June, 2008. The medical records were reviewed to determine which cases had died or were euthanized with a necropsy performed at Colorado State University. The Veterinary Diagnostic Laboratory records were then searched for the cases that had the paraffin embedded tissue blocks available for further testing.
The tissue blocks were shipped to the Intracellular Pathogens Research Laboratory at North Carolina State University (http://www.cvm.ncsu.edu/vth/ticklab.html). Total DNA was assayed for Bartonella spp. DNA using 3 different PCR tests. Conventional PCR assays targeting the ITS gene and Pap31 gene were performed as described previously.9,10 DNA extracted from blood of a healthy dog was used as the negative control and 0.001 pg/μL (equivalent to a 2.5 genomic copies per reaction) of B. henselae genomic DNA was used as the positive control. All amplicon products were sequenced to establish the Bartonella species, strain, or genotype. In addition, real-time PCR using both B. henselae and B. vinsonii subsp. berkhoffii specific probes was performed on all samples using the protocols described elsewhere.11 DNA extracts positive for Bartonella spp. DNA and the tissue blocks were then shipped back to Colorado State University and assayed in an eubacterial 16S rDNA PCR assay in an attempt to determine whether other organisms were present concurrently (J. Veir, M.R. Lappin, unpublished technique, 2010). While this assay can detect 318 × 10−15 g of eubacterial DNA in blood, the analytical sensitivity when used with formalin fixed tissues is unknown.
For those cardiac tissue samples that were positive for Bartonella spp. DNA by PCR, the tissue blocks were recut and 1 slide was stained with hematoxylin and eosin (H & E) stain and another slide stained with Warthin Starry silver stain. All slides were then evaluated microscopically by the same pathologist.
The initial medical records search identified 119 dogs with a presumptive diagnosis of endocarditis. The tissue blocks and medical records were available from 9 dogs. Of these 9 dogs, Bartonella henselae DNA was amplified from the tissues of 7 dogs; B. vinsonii subsp berkhoffii DNA was amplified concurrently from 3 dogs. In the PCR targeting Bartonella spp. ITS gene, 7 of the 9 dogs were positive. Based upon DNA sequencing, all of these dogs were infected with B. henselae (SA2 strain type). Two dogs were also PCR positive when targeting the Pap 31 gene and both were confirmed as B. henselae by DNA sequencing. Based upon real-time PCR using B. henselae and B. vinsonii subsp. berkhoffii probes, all 7 dogs were infected with B. henselae and 3 dogs were coinfected with B. vinsonii subsp. berkhoffii. Eubacterial DNA was not amplified from any of the remaining tissues.
On the initial necropsy or histopathology report, all 7 dogs had vegetative valvular endocarditis. Bacterial colonies were present on histological examination of tissues of 1 dog, but none of the tissues had been cultured. The tissues that remained in the paraffin blocks after evaluation by PCR assays at NCSU and CSU included mitral valve tissue and myocardial tissue (2 cases), aortic valve tissue and myocardial tissue (2 cases), myocardial tissue alone (2 cases), and mitral valve tissue alone (1 case). Valvular endocarditis was documented in the remaining tissues from 3 dogs. Aortic valve endocarditis was confirmed in 2 dogs (1 case having concurrent myocarditis), and the 3rd dog had mitral valve endocarditis with concurrent myocarditis. Neutrophils were the predominant inflammatory cells in these tissues, and there was fibrin accumulation in each. Inflammation was not noted in the other cases with tissues that remained after PCR analysis. The difference between the original and repeat histopathological evaluations likely relates to the loss of tissues during PCR testing.
While none of the remaining tissues had bacteria observed on microscopic evaluation of the H & E-stained sections, Warthin Starry silver stain revealed coccoid bacteria in tissues of 2 cases. In 1 case, the bacteria were noted along the surface of the mitral valve but were not associated with inflammation. This dog was blood culture positive for Streptococcus beta haemolytica during the clinical evaluation, and DNA of B. henselae was amplified from the tissues. In the other case, the bacteria were noted along the mitral valve and myocardium and were associated with inflammation. DNA of B. henselae was amplified from the tissues of this dog. Because of the poor background detail provided by the silver stain, it could not be determined whether or not the bacteria were intracellular or extracellular. As in situ hybridization was not performed, it cannot be determined more accurately whether the Bartonella spp. DNA amplified from tissues was concentrated in the areas of inflammation.
While the data described cannot be used to prove the Bartonella spp. identified were the cause of disease, amplification of Bartonella spp. DNA from tissues from each of the 7 cases suggests the dogs were infected. Of the 7 dogs, 6 were from Colorado and 1 was from Wyoming (B. henselae only) and all were greater than 5 years of age. There were 3 castrated males and 4 spayed females. There was no evidence in the medical records that the dogs had left the region. Two of the 7 dogs were reported to have either fleas or ticks, but the species were unknown. The dog with fleas was solely infected with B. henselae while the dog with ticks was coinfected with both B. henselae and B. vinsonii subsp. berkhoffii. There was no mention of flea and tick preventative use in the medical histories of any of the dogs.
Of the 7 dogs that tested positive for Bartonella spp. DNA in cardiac tissues, all presented to CSU-VTH with nonspecific clinical signs including lethargy, weakness, or anorexia. Four of the 7 dogs presented with fever (rectal temperature ≥102.5°F) and 3 dogs had heart murmurs, DNA of both B. henselae and B. vinsonii subsp. berkhoffii were amplified from cardiac tissues of 1 of 4 dogs with fever and 1 of 3 dogs with a murmur. Echocardiograms were performed on 2 of the 3 dogs with heart murmurs, and neither report listed the suspicion of IE; one of these reports stated that the mitral and aortic valves appeared normal. One of the 3 dogs with a heart murmur had pre-existing heart disease characterized as dilated cardiomyopathy. Routine blood cultures were performed only on the 2 dogs with fever and heart murmurs. One culture was sterile, and the other culture grew Enterococcus faecium and S. beta haemolytica. None of the dogs had been prospectively assessed for Bartonella spp. infection by PCR or culture.
Three of the 7 dogs had concurrent diseases including chronic renal disease, neoplasia, and diabetes mellitus; 2 of these dogs had B. henselae and B. vinsonii subsp. berkhoffii DNA amplified from tissues. Complete blood counts (CBC) were performed on all 7 dogs. Three of the 7 dogs were anemic (range = 2%–437%; reference range >43%) on the 1st CBC completed after presentation to the CSU-VTH; DNA of B. henselae alone was amplified from tissues from each of these dogs. Reticulocytes were counted in 1 dog, which had a nonregenerative anemia with a reticulocyte count of 49,350 cells/μL (reference range > 60,000 cells/μL). Thrombocytopenia without platelet clumps (range = 33,000–155,000 cells/μL; reference range > 200,000 cells/μL) was detected in 4 of the 7 dogs (2 dogs were concurrently anemic); 2 of these dogs had B. henselae and B. vinsonii subsp. berkhoffii DNA amplified from tissues. Three of the 7 dogs had changes interpreted as a stress leukogram while 1 dog had a neutropenia (1,700 cells/μL; reference range > 2,800 cells/μL). The neutropenic dog (B. henselae DNA only) was concurrently lymphopenic (200 cells/μL), but was not anemic. The 3 remaining dogs had a lymphopenia as the only abnormality (range = 100–900 cells/μL; reference range > 1,000 cells/μL). Hypoalbuminemia (range = 2.0–2.5 g/dL; reference range > 2.7 g/dL) was detected in 3 dogs; 2 of these dogs had B. henselae and B. vinsonii subsp. berkhoffii DNA amplified from tissues. Proteinuria (range = 1 + to 4 +; normal = negative) was detected in 4 of the 7 dogs; 2 of these dogs had B. henselae and B. vinsonii subsp. berkhoffii DNA amplified from tissues. Of the 4 dogs with proteinuria, 2 had concurrent hypoalbuminemia. Other serum biochemical and urinalysis abnormalities were more variable among the dogs.
During hospitalization, 5 of the 7 dogs were administered antibiotics intravenously that included both enrofloxacina and ampicillin. In addition to enrofloxacin and ampicillin, 1 dog was administered imipenem.b Another dog was administered metronidazole and cefazolin intravenously. One dog was not treated with antibiotics but was administered methylprednisolone sodium succinate.c Ultimately, none of the 7 dogs were discharged from the hospital either because of cardiopulmonary arrest (2 dogs) or euthanasia based on the worsening of clinical signs despite antibiotic therapy (5 dogs).
Care was taken during DNA extraction, and positive and negative controls were performed properly in all 3 Bartonella spp. PCR assays. Bartonella spp. DNA were not amplified from all tissues, and therefore false-positive results were unlikely.12 Antemortem blood cultures were only performed in 2 dogs (with 1 dog being positive). None of the cardiac tissues were cultured at necropsy, and so a eubacterial PCR assay was performed to attempt to assess the remaining cardiac tissues for other bacteria. Results of this assay were negative for all dogs, which suggest that other bacteria were not involved with the clinical disease syndromes noted in these dogs. However, while not optimized for Bartonella spp., based on sequence evaluation, this eubacterial DNA assay should amplify this genus. Thus, the assay was not as sensitive at the 3 Bartonella spp. specific PCR assays. As the analytic sensitivity of the assay is not known for use with formalin-fixed canine tissues, there is the possibility that other bacteria were present but not amplified. While all 7 dogs had Bartonella spp. DNA amplified by PCR, only 2 of 7 dogs had bacteria potentially consistent with Bartonella spp. seen microscopically in Warthin Starry silver-stained sections, further confirming that this technique is not as sensitive as PCR for documenting Bartonella spp. in tissues. Alternately, as multiple assays had been performed before this assay, adequate infected tissues might not have remained by the time this stain was applied.
While travel history was not mentioned within the medical records of these dogs, it is possible that these dogs could have originated from or traveled to geographical locations in which vector density is higher. As some time had passed since the loss of the pets, we chose not to contact the owners directly. Two of the 3 dogs diagnosed with endocarditis and myocarditis were historically infested with ticks or fleas. Evidence for the use of flea or tick control products was not found in the medical records. This study illustrates the importance of having Bartonella spp. infections on appropriate differential lists and that use of flea and tick preventatives is indicated for dogs even in states with a low perceived risk of infestation.
Much of the clinical and laboratory information collected retrospectively was consistent with IE. Results of this study emphasize the limitations of diagnostic tests for documentation of IE in dogs. For example, 1 dog was diagnosed histopathologically with mitral valve endocarditis and myocarditis, yet the dog had a negative blood culture and a normal echocardiogram, which stated that both the “mitral and aortic valves appeared normal.” It is possible to have a normal echocardiogram in the face of cardiac inflammation, as it is also possible to have no growth on a blood culture if the heart valve is not showering bacteria into the blood stream at the time of blood collection. Recently, it was shown that a combination of pre-enrichment liquid culture using an insect cell culture-based growth media and PCR is superior to routine aerobic culture for detection of Bartonella spp. in blood of dogs. Thus, if Bartonella spp. IE is suspected, the combination of culture in liquid media with PCR should be used.d,13
It is difficult to definitively say whether or not Bartonella spp. infection alone was responsible for the clinical signs of these dogs, especially when 3 of the 7 dogs had known concurrent systemic illness. Previously, there has been only one other report of B. henselae causing IE in dogs.4 Of the 7 dogs described here, B. henselae and B. vinsonii subsp. berkhoffii DNA was amplified from 3 dogs. This finding might be important as it is now recognized that coinfections with multiple organisms might increase the likelihood of developing clinical illness.
For 6 of the 7 cases, the primary clinician suspected an infectious disease and prescribed antibiotic therapy. If Bartonella spp. endocarditis or myocarditis is suspected in dogs, concurrent use of a fluoroquinolone with rifampin or an aminoglycoside like amikacin might be indicated. If dogs with Bartonella spp. IE survive to discharge, weeks of antibiotic therapy is indicated, and as with other forms of IE, heart failure might ultimately develop from chronic heart valve damage. We conclude that B. henselae and B. vinsonii subsp. berkhoffii cardiac disease might occur in dogs in regions in which tick and flea exposure is thought to be uncommon.