Criteria for Selection of Cases

Records in the electronic database of the University of California Veterinary Medical Teaching Hospital from January 1992 to August 2005 were searched with the terms endocarditis and canine. Records were reviewed by one of the authors (JES), and cases were designated as possible or definite endocarditis on the basis of echocardiographic, microbiologic, and pathologic findings. Cases were assigned to the definite IE group if findings satisfied criteria that were modified from the Duke criteria for IE^21,22 (Appendix) or when the diagnosis was confirmed during necropsy. Possible endocarditis was diagnosed by use of the modified Duke criteria or, in the absence of positive MCB results, if clinical and echocardiographic findings (eg, a new or progressive heart murmur) were suggestive of IE or if there was resolution of lesions seen echocardiographically as a result of antimicrobial treatment. Records of cases designated as possible IE were reviewed by a board-certified cardiologist (MDK) to confirm appropriateness for inclusion in the study. All necropsy findings were reviewed and confirmed by a board-certified pathologist (PAP).

Procedures

Information extracted from the medical records included results of bacterial culture and susceptibility testing of blood, cardiac valve specimens, and urine; serologic assays for Bartonella spp or PCR assay on valvular vegetative lesions; and serologic tests for other canine infectious diseases. Records and necropsy reports were also reviewed for information pertaining to antimicrobial administration prior to examination, potential portals of organism entry, and predisposing causes. The effect of recent antimicrobial drug administration on the likelihood of obtaining a positive blood culture result was evaluated.

Five groups of dogs were designated for further statistical analysis. In 4 groups, a causative organism (Streptococcus spp, Staphylococcus spp, gram-negative organisms not including Bartonellaspp, and Bartonellaspp) was identified, and in 1 group, no organism was identified. Dogs infected with organisms that did not fit into those categories were excluded from analysis because the small numbers precluded analysis. The following variables were evaluated in relation to the infecting organism: age, sex, duration of illness, rectal temperature and fever, valves affected (ie, aortic alone, mitral alone, or multiple valve involvement), thrombocytopenia, high serum creatinine concentration, high serum ALT activity, neutrophils with toxic changes, neutrophilic polyarthritis, renal complications, neurologic complications, thromboembolic complications, cardiac arrhythmias, and CHF. Renal complications were defined as clinicopathologic evidence of renal injury (eg, cylinduria or azotemia and isosthenuria); protein-losing nephropathy (ie, urine protein-to-creatinine ratio > 5.0); lesions consistent with renal infarction observed on abdominal ultrasonographic imaging; or lesions detected at necropsy that were consistent with renal infarction, renal vasculitis, or glomerulonephritis. Neurologic complications were defined as observation of neurologic signs and abnormal results of neurologic examination temporally associated with diagnosis of IE (with or without abnormal findings on CSF analysis), or the finding of inflammatory lesions or infarction in the CNS at necropsy. Dogs with thromboembolic complications had changes consistent with infarction on abdominal ultrasonographic imaging or lesions consistent with thromboembolism at necropsy. Relationships between survival and infection with Streptococcus canis, gram-negative organisms, or Bartonella spp were compared with those in the rest of the population.

Statistical analysis—The c² test was used to determine whether groups of organisms differed with respect to each categoric variable. The Fisher exact test was used to detect differences between 2 groups by use of frequency data (eg, whether streptococci were more likely to infect the mitral valve than other organisms). One-way ANOVA was used to compare continuous variables among groups of organisms, and a Bonferroni test was used to compare selected groups when differences were significant. For survival analysis, KaplanMeier curves were constructed; dogs that were lost to follow up were censored in the analysis. Deaths were classified as related or unrelated to IE. The relationship between infecting organism and survival was determined. Survival curves were compared by use of the log-rank test. Hazard ratios and 95% confidence intervals were also determined. All analyses were performed by use of statistical software.^a Significance was defined as values of P < 0.05.

Results

The database search yielded 91 records. Twenty dogs were excluded on the basis of equivocal echocardiographic lesions and absence of other supportive evidence for IE (eg, negative results of MCB and absence of fever or thromboembolic or immunologic complications) or because a cardiac lesion detected during necropsy was not consistent with a diagnosis of IE. Nineteen of the remaining 71 (27%) dogs were classified as possible endocarditis according to the modified Duke criteria, and 52 (73%) were classified as definite endocarditis. Of the dogs designated as having definite endocarditis, 28 were thus classified on the basis of cardiac histologic findings, either alone or in combination with other major or minor criteria. Subsets of the 71 dogs have been reported in studies^8,17 of IE in dogs in northern California.

A causative organism was identified in 41 of the 71 (58%) dogs (Table 1). Organisms were detected in 40 dogs with definite IE and in 1 dog with possible IE. Causative bacteria were identified via routine culture methods, either antemortem or at necropsy, in 33 of the 64 (52%) dogs in which MCB or culture of a necropsy specimen was performed. In 9 dogs, a causative organism was identified via serologic or molecular methods. Streptococci or staphylococci were cultured from blood or tissue specimens in 21 of the 41 (51%) dogs. In 9 of those 41 (22%) dogs, gram-negative organisms were detected. The group G streptococcal species S canis was the most commonly detected organism (10 [24%] dogs). Other organisms included group D streptococci, which affected 5 (12%) dogs. Organisms from this group included Streptococcus bovis (n = 2 dogs) and Enterococcusspp (Enterococcus faecium [1] and Enterococcus faecium var zymogenes [1]). Additional organisms included coagulase-positive staphylococci (n = 4 dogs [10%]; including Staphylococcus aureus [2] and Staphylococcus intermedius [1]), coagulase-negative staphylococci (2 [5%]), and Escherichia coli (3 [7%]). Actinomyces spp, Pseudomonas aeruginosa, Klebsiella pneumoniae, Erysipelothrixspp, Citrobacter freundii, Salmonella arizonae, a lactobacillus-like organism, an untypable nonenteric gram-negative bacterium, and Pasteurella caniswere also isolated in 1 dog each. In addition, histologic evaluation of specimens obtained at necropsy revealed intralesional organisms by use of H&E or Brown and Brenn staining in 4 other dogs in which MCB results were negative. Of those, gram-positive cocci were observed in 2 dogs, gram-positive rods were observed in 1 dog, and short bacterial rods were observed in 1 dog; those dogs were not included in analyses of cases with known causative organisms.

One or more MCBs had been performed in 55 of the 71 (77%) dogs, and results were positive for microbial growth in 26 (47%) of those instances. Thirty-four of the 55 (62%) dogs had received antimicrobial treatment within a 2-week period prior to examination at the teaching hospital. One or more antimicrobial agents had recently been administered to 42 of 57 (74%) dogs with a known medication history, and blood cultures had been performed in 34 of those dogs. The most commonly administered classes of antimicrobial agents were fluoroquinolones (enrofloxacin, orbifloxacin, or difloxacin [n = 23 dogs]), penicillin derivatives (ampicillin, amoxicillin, or clavulanic acid-amoxicillin [16]), cephalosporins (cefazolin, cephalexin, cefotaxime, or cefixime [8]), and tetracyclines (doxycycline [9]). Administration of antimicrobial drugs within 2 weeks of examination did not affect the likelihood of positive MCB results (15/34 [44%] vs 11/21 [52%]; P = 0.38), and the prevalence of each group of organisms was the same in dogs that had recently received antimicrobials as it was in those that had not.

The number of blood samples submitted for culture was recorded for 53 of the 55 dogs, including for all of the dogs with positive results. The median number of blood samples submitted was 3 (range, 1 [1 dog] to 6 [2 dogs]). All inoculated bottles yielded growth of a single organism in 20 of 26 (77%) dogs. When all bottles did not have positive results for bacterial growth, > 1 bottle had positive results in 3 dogs. In the remaining 3 dogs, only 1 bottle yielded bacterial growth. In dogs in which not all submitted specimens had positive results, an organism was always cultured from the first blood sample, the first and second blood samples, or the second blood sample but not the first blood sample.

Results of MCBs in 29 of the 55 (53%) dogs in which MCBs had been performed were negative; however, the probable cause of endocarditis was apparent in 13 (45%) of those dogs. Infection with Bartonellaspp was detected by use of serologic assay or PCR assay on valvular vegetations or both in 6 of the 29 (21%) dogs; 13 dogs were not tested for Bartonellaspp, 9 other dogs were seronegative, and 1 dog had negative results of PCR assay for Bartonella spp. In the 7 remaining dogs, IE was thought to be caused by infection with Mycobacterium spp (n = 1 dog), Aspergillus spp (1), streptococci (3), and E coli (2). Mycobacterial infection was detected via PCR assay and sequencing on paraffin sections of valvular tissue in 1 dog with pyogranulomatous inflammation in subcutaneous tissue, skeletal muscles, and lymph nodes, but the species could not be identified because of limited sequence information. Aspergillus flavipes was suspected as the cause of endocarditis in a German Shepherd Dog with a diagnosis of diskospondylitis and growth of the fungal organism from a fine-needle aspirate of disk material, but that dog was not included as a case with known cause in analyses because it was possible that bacterial IE had complicated the disease. In another dog, S canis was considered the likely etiologic agent on the basis of isolation of the organism from synovial fluid. In 2 other dogs, the infecting organism was identified as S canis and a group D Streptococcus spp by bacterial culture of necropsy specimens. One of the 7 dogs had a history of recent, recurrent prostatitis from hemolytic E coli infection, and the other dog had E coli–induced urinary tract infections and pneumonia, but both were receiving antimicrobial treatment at the time of referral to the teaching hospital. Neither of those 2 dogs was included as a case with a known cause for IE. Isolated tricuspid valve involvement was diagnosed in 1 dog with MCB–negative IE, and 2 dogs had IE involving the left ventricular wall; S canis was detected in one of the latter dogs at necropsy. Three dogs had negative MCB results, were seronegative for Bartonellaspp, and had not been receiving antimicrobials prior to examination; necropsies were not performed in any of those dogs.

Results of routine bacterial culture of tissue specimens at necropsy, supported by histologic detection of intralesional organisms, led to a diagnosis of the causative agent in an additional 4 dogs in which MCBs had not been performed. In 3 dogs with supportive histopathologic findings, these were a lactobacillus-like organism, Actinomycesspp, and E coli. Streptococcus canis was thought to be the likely cause of IE in another dog on the basis of results of bacterial culture of urine and histologic detection of intralesional gram-positive cocci, although multiple tissue specimens procured during necropsy yielded growth of E coli; that dog was thought to have mixed S canis and E coli septicemia prior to death. Thus, the total number of dogs in which organisms were identified by use of routine culture methods was 33 (46% of all dogs; 80% of dogs in which an organism was identified).

Aerobic bacterial culture of urine was performed in 42 dogs, including 19 dogs with growth on MCBs, 20 dogs with no growth on MCBs, and 3 dogs in which MCBs were not performed. Five (12%) dogs had bacterial growth in urine, 3 (16%) of which also had growth of the same organism on MCBs; of those 3 dogs, 1 was infected with Enterococcus faecium, 1 was infected with S arizonae, and 1 was infected with S canis. In the dog infected with E faecium, the lower portion of the urinary tract was the likely source of infection. Streptococcus canis was cultured from the urine of another dog with no bacterial growth on MCBs. The fifth urine sample was positive for growth of E coli and came from a dog that had no bacterial growth on MCBs but from which S canis was cultured from synovial fluid. Other antemortem body fluid samples that supported bacterial growth in dogs with positive MCB results were CSF (infected with S canis [n = 1 dog] or with a coagulase-negative staphylococcal organism [1]), synovial fluid (infected with S arizonae [1] or with S canis [1]), and fluid from a subcutaneous abscess (infected with S canis [1]).

Results of susceptibility testing, including determination of minimum inhibitory concentrations, were available for 22 isolates. Because of predictable susceptibility patterns, several S canis isolates were not subjected to testing, and susceptibility testing was not performed on isolates from necropsy specimens. Isolates with broad susceptibility patterns included S arizonae, P canis, Erysipelothrix spp, S canis(5 isolates), S bovis(2 isolates), a S aureusisolate, a coagulase-negative staphylococcus, S intermedius, a nonenteric isolate, and K pneumoniae. In contrast, the Enterococcusisolates, P aeruginosa, 2 E coli isolates, 1 S aureusisolate, and 1 C freundiiisolate were resistant to multiple antimicrobial drugs. Results of serial MCBs in the dog with P aeruginosasepticemia indicated increasing resistance in the face of antimicrobial treatment. Several of the organisms were resistant to fluoroquinolones and third-generation cephalosporins, whereas all were susceptible to aminoglycosides, and all but C freundii and an E coliisolate were susceptible to ticarcillin-clavulanic acid.

Thirty-two dogs were tested for Bartonella spp infection via serologic tests or PCR assay of material from vegetative lesions, and 8 (25%) of those dogs had positive results. One of those dogs was classified as having possible IE, and the remainder were classified as having definite IE. All of these 32 dogs had been examined during the last 5 years of the study. Serologic testing for Bartonella spp commenced in our hospital in 1999, and from the date of the first test to the end of the study, IE was diagnosed in 45 dogs and a bacterial etiology was confirmed in 28 dogs. Thus, the prevalence of positive test results for Bartonella spp was 18% (8/45), and 29% (8/28) of the dogs in which the causative agent was known had positive test results for Bartonellaspp. Seven dogs had positive test results for Bartonella spp from 1999 to 2001. From 2001 to the termination of the study, seropositivity to Bartonellaspp was detected in only 1 of 28 dogs, and the serum reciprocal antibody titer (128) was of low magnitude. Assuming that IE in that dog was truly caused by Bartonellaspp, the organism was associated with only 1 of the 16 cases with a known cause from 2001 to 2005. However, 7 dogs in which IE was of unknown cause had not been tested for Bartonellaspp.

Serologic testing for Bartonella spp had been performed in 27 dogs, including 13 dogs with negative MCB results, 1 dog with negative MCB results and negative culture results from multiple tissues collected at necropsy (including heart valve lesions), 11 dogs with positive MCB results, 1 dog with positive culture results on tissues collected at necropsy, and 1 dog in which no cultures had been performed. Serologic testing yielded positive results in 7 dogs, including 5 of the dogs with negative MCB results, the dog with negative MCB results and negative culture results from necropsy specimens, and a dog with positive MCB results for S canis. Results of serologic testing for 5 dogs are reported elsewhere.⁸ An additional dog was seroreactive for Bartonella vinsonii subsp berkhoffii with a reciprocal titer of 128, and the dog with positive MCB results for S caniswas seroreactive for Bartonella clarridgeiaewith a reciprocal titer of 256. For statistical analysis, the latter dog was classified as having endocarditis caused by S canisinfection on the basis of isolation of S canis from the blood, spleen, and a cutaneous abscess; histologic detection of intralesional cocci in a spleen specimen; and positive response to treatment with b-lactam antimicrobials.

A broadly reactive PCR assay for Bartonella spp was performed on valvular tissue from 11 dogs. Seven of those dogs had no growth on culture of blood (n = 4 dogs), necropsy specimens (1), or blood and necropsy specimens (2, including the dog in which Mycobacterium spp was detected). Three dogs had microbial growth on culture, and no specimens were submitted for culture in 1 dog. Results of PCR assay for Bartonella spp were positive in 5 dogs with negative results of MCBs, as reported elsewhere,⁸ and in the dog in which microbial culture was not performed. The sequence of the PCR product from the latter case suggested infection with B vinsoniisubsp berkhoffii. Of the 31 dogs in the study in which results of culture of any tissue or body fluid were negative, Bartonella spp infection was diagnosed in 6 (19%).

Serologic tests for antibodies against Anaplasma phagocytophilum were performed in 19 dogs, 6 of which had titers (range of reciprocal serum titers, 80 to 640). All seropositive dogs except one (the dog with the lowest titer) were also seropositive for Bartonellaspp. Not all dogs that were seropositive for Bartonellaspp had antibodies against A phagocytophilum. Seventeen dogs were tested for antibodies against Ehrlichia canis, 16 for antibodies against Rickettsia rickettsii, 13 for antibodies against Borrelia burgdorferi, and 5 for antibodies against Ehrlichia platys. One dog that was seroreactive to Bartonella spp and A phagocytophilumwas also seropositive for E canis, R rickettsii, and Borrelia burgdorferi; 1 was seropositive for R rickettsii and E canis, and another was seropositive for R rickettsii, as previously reported.⁸ Two other dogs had a low titer or equivocal results for B burgdorferi, including the dog with the lowest titer to A phagocytophilumand a dog with endocarditis that had no microbial growth on culture of blood.

Results of valvular histopathologic evaluation were available for 27 dogs. Mineralization of valve tissues was reported in 9 dogs; all except 1 of those dogs had aortic valve involvement. Five of the dogs with valvular mineralization had endocarditis secondary toBartonella spp infection, and the etiology was unknown in 3 dogs. The dog with mitral valve involvement had endocarditis secondary to Streptococcusgroup D (non-Enterococcus) infection, and in that dog, only a small amount of mineralized tissue was noticed. Intralesional organisms were detected in 10 (37%) dogs with either H&E or Brown and Brenn stains; 3 of those dogs had negative culture results, and specimens were not cultured in 1 dog. Convincing light microscopic evidence of Bartonella spp infection was lacking, even when silver stains were used, but organisms were detected in 1 dog by use of electron microscopy. Features of the inflammatory process ranged from chronic (with lymphoplasmacytic and histiocytic cell infiltrates and granulation tissue [11 dogs]) to acute, suppurative, and fibrinous (8 dogs). Four dogs had subacute inflammation with neutrophilic-histiocytic and fibrinous infiltrates, and 2 dogs had chronic-active disease, with neutrophil infiltrates superimposed on granulation tissue. Features were not reported in 2 dogs. Six of 9 dogs tested with chronic endocarditis were infected with Bartonellaspp, whereas none of the 5 dogs tested with subacute or acute endocarditis had positive results for Bartonellaspp. The inflammatory reaction for the 5 dogs infected with Streptococcusspp ranged from acute to chronic in nature.

Review of medical records revealed potential predisposing causes or portals of entry in 33 of the 71 (46%) dogs, although in many cases the causal relationship was not definitive. Two dogs had been treated with balloon dilatation for treatment of subaortic stenosis. Three dogs had a history of recent abdominal surgery, although no evidence of infection in the perioperative period was noted. One of those 3 dogs also had atypical hypoadrenocorticism, megaesophagus, and aspiration pneumonia. Two dogs had undergone a prescrotal urethrostomy procedure; both of those dogs had endocarditis secondary to Enterococcusspp infection and 1 had concurrent cystitis as a result of the infection. The other dog had historical obstructive urate urolithiasis and was referred to the teaching hospital after empiric treatment with antimicrobials; no urinary tract infection was documented in the record for that dog at the time of examination. The dog with endocarditis caused by S arizonaeinfection had a history of eating wild lizards and also had urinary tract infection caused byS arizonae. Two other dogs had streptococcal urinary tract infections diagnosed at the time of initial examination. It was unclear whether those infections preceded or followed development of endocarditis in those dogs. Other potential foci of infection that may have predisposed dogs to development of endocarditis were skin abscesses and wound infections (n = 4 dogs), pyelonephritis (4, including 1 dog with IE secondary to E coliinfection and 1 infected with a gram-positive coccal organism), chronic pyoderma with or without accompanying otitis (4), aspiration pneumonia (2), prostatitis secondary to E coliinfection (1), pyometra (1 dog with IE secondary to K pneumoniae infection), recurrent hemorrhagic gastroenteritis (1 dog with IE secondary toE coli infection), colonic carcinoma (1), and fibrosarcoma involving the oral cavity (1). One of the dogs with otitis had otitis media and IE caused by a coagulase-negative staphylococcus. Two dogs with streptococcal endocarditis had tooth root abscesses (one in conjunction with a dental extraction procedure 6 weeks prior to initial examination), and 1 dog with dental disease had a fractured tooth. The dog with IE secondary to infection with aLactobacillus-like organism had severe dental disease, although periodontal disease per se was not included as a predisposing cause in this study. Two dogs had a sharp-ended gastric foreign body thought to be associated with perforation; one of those dogs had peritonitis, and in the other dog, focal peritonitis from a wire foreign body was suspected on the basis of radiographic findings. Three dogs had diskospondylitis (associated with Aspergillus spp, C freundii, and an unknown agent), one had sternebral osteomyelitis (associated with S bovis infection), and another with mycobacterial IE had severe pyogranulomatous sialoadenitis and cellulitis. Because these sites were not associated with a portal of entry, these dogs were not included as cases with potential predisposing conditions. Twenty-one (30%) dogs had acquired defects of innate or adaptive immunity that could have predisposed them to developing endocarditis, including 6 dogs with malignancies, the 2 dogs with barrier disruption secondary to a prescrotal urethrostomy procedure, 2 dogs with hepatic disease (chronic active hepatitis and portosystemic shunt), and 1 dog with uncontrolled hyperadrenocorticism. Underlying neoplastic diseases consisted of an oral-cavity fibrosarcoma, splenic hemangiosarcoma, pancreatic acinar adenocarcinoma, adrenal neoplasms (in 2 dogs), and a rectal carcinoma in situ. Three dogs were receiving long-term treatment with prednisone for chronic hepatitis, bronchitis, or dermatitis. In 8 dogs, recent treatment with glucocorticoids, alone or in conjunction with azathioprine treatment, may have predisposed to endocarditis, although it is also possible that endocarditis was present prior to initiation of treatment. The most common underlying cardiac valvular disease was thickening of the aortic valve cusps secondary to subaortic stenosis, which was diagnosed in 7 dogs. No apparent link between preexisting cardiac valvular disease and an infecting group of organisms was detected.

The frequencies with which various clinical and clinicopathologic findings were reported in the 67 dogs assigned organism groupings were summarized (Table 2). Occurrence and types of renal, neurologic, and thromboembolic complications were described in the accompanying study.²³ Variables that were significantly different among groups of organisms were the involved valve (aortic valve, P = 0.009; mitral valve, P = 0.008), fever (P = 0.009) and mean rectal temperature at initial examination (P = 0.002), neutrophilic polyarthritis (P = 0.03), and congestive heart failure (P = 0.006). Dogs infected with Bartonella spp were significantly more likely to have aortic valve involvement than were other dogs (P = 0.003) and significantly less likely to have mitral valve involvement (P = 0.02; Figure 1). Conversely, dogs infected with streptococci were significantly (P = 0.004) more likely to have mitral valve involvement. Dogs infected with Bartonella spp also had lower mean rectal temperatures at initial examination than dogs infected with gram-negative organisms (P < 0.05), but there were no significant differences when mean rectal temperatures in dogs with gram-negative organisms were compared with temperatures in dogs with disease from other known etiologies (Figure 2). Dogs infected with Bartonellaspp were significantly more likely to be afebrile at the time of initial examination than dogs in other groups (P = 0.03), including dogs infected with streptococci (P = 0.02), gramnegative organisms (P = 0.009), and staphylococci (P = 0.046), but not the group of dogs in which an organism was not identified. Dogs infected with Bartonella spp were also more likely to develop congestive heart failure than those infected with other organisms (P = 0.004), and dogs infected with gram-negative organisms were less likely to develop congestive heart failure (P = 0.02). Dogs infected with streptococci were more likely to have neutrophilic polyarthritis than dogs infected with other organisms (P= 0.004). Synovial fluid analysis in nearly one fourth of these dogs revealed characteristics of septic polyarthritis, including intracellular cocci and degenerate neutrophils.²³ In most instances, large numbers of nondegenerate neutrophils that were characteristic of immune-mediated polyarthritis were observed, and in 1 dog, lupus erythematosus cells were observed in synovial fluid. Most dogs infected with gram-negative organisms had neutrophils with toxic changes, but this finding approached significance only when compared with the other dogs in the analysis (P = 0.06) because numerous dogs infected with other organisms also had toxic-appearing neutrophils.

Summary of groups of infecting organisms and cardiac valves involved in 71 dogs with IE. *Denotes a significant difference for frequency of mitral and aortic valve involvement, compared with that associated with other groups of infecting organisms.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Summary of groups of infecting organisms and cardiac valves involved in 71 dogs with IE. *Denotes a significant difference for frequency of mitral and aortic valve involvement, compared with that associated with other groups of infecting organisms.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

Summary of groups of infecting organisms and cardiac valves involved in 71 dogs with IE. *Denotes a significant difference for frequency of mitral and aortic valve involvement, compared with that associated with other groups of infecting organisms.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

Mean ± SD rectal temperatures in dogs with IE caused by various groups of causative organisms. *Denotes a significant difference in mean rectal temperature, compared with that in dogs with Bartonellaspp infection.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mean ± SD rectal temperatures in dogs with IE caused by various groups of causative organisms. *Denotes a significant difference in mean rectal temperature, compared with that in dogs with Bartonellaspp infection.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

Mean ± SD rectal temperatures in dogs with IE caused by various groups of causative organisms. *Denotes a significant difference in mean rectal temperature, compared with that in dogs with Bartonellaspp infection.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

Fourteen dogs died, 43 were euthanized, 12 were lost to follow-up, and 2 were alive at the time of writing. Additional data relating to survival of this patient population have been described in a separate manuscript.²³ When the relationships between survival and infection with S canis, gram-negative organisms, or Bartonella spp were compared with those in dogs with IE caused by other organisms and dogs with IE of unknown cause, the only significant variable was infection with Bartonellaspp, which was negatively correlated with survival (P = 0.008; hazard ratio, 2.72; 95% confidence interval, 1.57 to 18.82; Figure 3).

Association between Bartonella spp infection and survival in 71 dogs with IE.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Association between Bartonella spp infection and survival in 71 dogs with IE.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

Association between Bartonella spp infection and survival in 71 dogs with IE.

Citation: Journal of the American Veterinary Medical Association 228, 11; 10.2460/javma.228.11.1723

• Download Figure
• Download figure as PowerPoint slide

Discussion

To the authors' knowledge, the present study represents the largest series in which microbiologic findings associated with IE in dogs were evaluated and is the only large series in which clinical features of IE were compared among dogs infected with different groups of organisms. The study is also the first in which infections with Salmonella spp, Citrobacter spp, and Mycobacterium spp were diagnosed as causes of IE. Similar to findings in a review¹⁸ of humans with IE, the most common infecting organisms in dogs in the present population were streptococci (37%). Other predominant organisms were gram-negative bacilli (22%; most commonly E coli) and Bartonella spp (20%); the prevalence of those organisms as causes of IE in humans is much lower.^18,24 Staphylococci were isolated in 15% of dogs. This finding was in contrast with those from studies^3,4,16 published 2 decades ago, in which a higher prevalence of infection with staphylococci, Corynebacteriumspp, and E coliwas reported. This may reflect geographic or temporal differences in the epidemiology of IE in dogs and increased veterinarian awareness of the need to pursue testing for Bartonellaspp. The only other published study of IE in dogs in northern California was a prospective study⁸ in which 18 cases were described during a 2-year period (from 1999 to 2001), including a subset of the dogs included in the present study. In that study, Bartonella spp was the most commonly diagnosed cause of IE, accounting for 5 (28%) cases.

Because testing for Bartonella spp at our institution only commenced in 1999 and because not all dogs with IE have been tested since that time, the study reported here underestimates the prevalence of Bartonella spp infection in dogs. Since the initiation of testing in 1999, Bartonellaspp infections have composed 29% of cases in which the cause was known. However, between 2001 and the termination of the present study, Bartonellaseropositivity was detected in only 1 of 28 dogs (4% of cases with known cause). Seven dogs in which an etiologic agent was not identified were not tested for Bartonellaspp; 1 of those dogs had isolated aortic valve endocarditis, and 2 of those dogs had aortic valve endocarditis in conjunction with mitral valve involvement. Knowledge of the predilection of Bartonella spp for infection of the aortic valve may be a reason for lack of testing in those cases. Because only a few cases of endocarditis secondary to infection with Bartonella spp have been reported, clinicians should be encouraged to test dogs with negative MCB results for Bartonella with serology regardless of the site of valve involvement or geographic location of the dog so that our understanding of the epidemiology of Bartonella infection and subsequent IE in dogs might be improved. In a series of IE secondary to Bartonella spp infection in humans,²⁴ 91% (20/22) patients had aortic valve involvement; in a subsequent series involving 48 patients,²⁵ the prevalence of aortic valve involvement was only 60%, a value that was not significantly different from the prevalence of endocarditis caused by other pathogens. Temporal variations in the prevalence of Bartonella spp endocarditis may reflect differences in ectoparasite activity associated with different climactic conditions.

In the present study, dogs with positive test results for Bartonella spp were significantly more likely to be afebrile than other dogs with IE, had predominantly aortic valve involvement, and had a higher prevalence of CHF than was observed in dogs infected with other pathogens. Because histopathologic assessment of valvular tissue in those dogs revealed chronic inflammation,¹⁷ we also hypothesized that the duration of illness prior to initial examination in dogs with Bartonella spp infection might have been longer than in dogs infected with other pathogens. However, the mean duration of recorded illness for dogs with Bartonella spp infection was not significantly longer than that in the other groups. Bartonella spp are well recognized causes of endocarditis in humans with negative MCB results, accounting for 10/299 (3%) of cases of IE reported in 1 series²⁴ and 99/348 (28%) cases in another.²⁶ To the authors' knowledge, an association between IE resulting from Bartonella spp infection and congestive heart failure has not been investigated in humans. It has been reported²⁵ that there is no significant difference in the prevalence of fever in human patients with Bartonella spp endocarditis and patients with other causes of IE, but the magnitude of fever has not been studied. The study reported here corroborates findings data from an earlier study⁸ in which a negative association between infection with Bartonella spp and survival was observed.

Of dogs infected with streptococci, S canis, a group G (pyogenic) streptococcus that is most commonly isolated from the oral cavity and anal region of dogs,²⁷ was the predominant causative organism. In humans, a-hemolytic streptococci belonging to the viridans group are the causative agents in most streptococcal infections,²⁸ whereas endocarditis caused by group G streptococci is uncommon to rare.²⁹ The major portal of entry for group G streptococci in humans is skin; 3 of the 10 dogs infected with S canis in the present study had evidence of skin infections, and 1 dog had a tooth root abscess.

Several dogs in the present study were infected with group D streptococci, a group that includes Enterococcus spp and S bovis. These organisms are part of the normal flora of the gastrointestinal tract of humans and other animals.³⁰ Recently, IE caused by S boviswas reported in a small-breed dog with myxomatous mitral valve degeneration after routine dental prophylaxis.¹⁴Streptococccus bovis infection causes 5% to 19% of IE in humans,^31,32 commonly develops in elderly patients and patients with colonic disease, frequently involves multiple valves, is associated with embolic complications, and has a good short-term prognosis.³¹ In 1 study³¹ of humans, endocarditis resulting from S bovis infection was associated with large valvular vegetative lesions, whereas in another study,³² small and fixed lesions were associated with the organism. All 3 dogs with streptococcal infections in the present study that were infected with nonenterococcal group D streptococci had mitral valve involvement, but valvular lesions were not particularly large; in fact, the dog with the unspeciated infection had a normal echocardiogram, and the diagnosis was only made at necropsy. One dog had a history of tooth root abscessation and recent dental extraction combined with a gastric foreign body and peritonitis, and 1 dog had sternebral osteomyelitis. Embolism to multiple sites was observed in 1 dog. The 2 dogs with enterococcal endocarditis had both undergone a prescrotal urethrostomy surgery, suggesting a possible relationship between genitourinary tract abnormalities or surgery and Enterococcus spp endocarditis, as has been reported³³ in humans. Both isolates were resistant to multiple antimicrobial drugs. In human patients, prior antimicrobial treatment is an important risk factor for infection with antimicrobial-resistant enterococci.³⁴ Both of the dogs with enterococcal infections had received multiple antimicrobials prior to initial examination, but whether this occurred before or after development of the resistant infection was not known.

In the present study, dogs infected with streptococci were more likely to have mitral valve involvement than dogs infected with other organisms. Dogs with streptococcal infections also had a higher prevalence of neutrophilic polyarthritis (including septic polyarthritis and, in 1 dog in which lupus erythematosus cells were observed in synovial fluid, immunemediated disease). The high prevalence of IE secondary to streptococcal infection and data from the present study should make Streptococcus spp high on the differential diagnostic list of causative agents in dogs with polyarthritis and mitral valvular endocarditis, at least in northern California.

The small number of dogs with staphylococcal endocarditis in the present study precluded meaningful analysis of the relationship between staphylococcal infection and clinical features of IE. Four dogs in which IE was caused by coagulase-positive staphylococci and 2 associated with coagulase-negative staphylococci were identified. The latter were thought unlikely to be contaminants because the organism was isolated from 3 MCB bottles in 1 dog and was isolated from blood, CSF, and valvular tissues at necropsy in the other dog. In the latter dog, gram-positive cocci were also observed during histologic evaluation of valvular lesion tissue. In humans with IE, coagulase-positive staphylococcal infections outnumber those caused by coagulase-negative staphylococci, accounting for 10% to 27% and 1% to 3% of cases, respectively.^18,20 Staphylococcus aureus is an important cause of IE in human intensive care units.³⁵ A cutaneous portal of entry and concurrent disease, such as diabetes mellitus and neoplastic disease, are more common in humans with S aureus IE than in patients with IE caused by other pathogens. The organism affects the right side of the heart more frequently than other pathogens, frequently causes neurologic complications, and is associated with a high (34%) in-hospital mortality rate.²⁰ In contrast, although they were severely ill at the time of initial examination, the 2 dogs with S aureus infections in the present study had good long-term outcomes; the valvular lesion resolved in 1 dog, and the other dog was euthanized more than 2 years after IE was diagnosed because of persistent chylothorax. Neither of those dogs had neurologic complications, and both had involvement of the left side of the heart. This may reflect a difference in virulence of coagulase-positive staphylococci in dogs with IE or increased susceptibility to antimicrobial drugs; alternatively, those dogs may not have been representative of IE secondary to coagulase-positive staphylococcal infections in dogs.

A variety of gram-negative bacilli were isolated from dogs in the present study, most commonly E coli, but also S arizonae, P canis, K pneumoniae, C freundii, and P aeruginosa. The predominance of female dogs may reflect a sex-related association with urinary tract infections, but case numbers were insufficient to confirm the importance of this observation. Several of those organisms were resistant to antimicrobial drugs, and the P aeruginosaisolate acquired additional resistance during the course of treatment. In humans, the prognosis for recovery from IE caused by P aeruginosa is poor because of rapid induction of antimicrobial resistance, the need for frequent or continuous drug administration, and impaired clearance of mucoid strains of the organism from vegetative lesions.¹⁸ The dog in the present study was euthanized 14 days after diagnosis because of neurologic complications, which develop in 53% of humans with this infection.¹⁸Citrobacter freundiiis a rare cause of bacteremia and IE in humans and typically infects neonates, the elderly, and debilitated or immunocompromised individuals; a poor prognosis is associated with these factors rather than the virulence of the organism, per se.³⁶ The dog with C freundii infection in the present study was a 2-year-old Rottweiler with concurrent diskospondylitis; the diskospondylitis and endocarditis resolved with amikacin treatment. Infective endocarditis secondary to E coli infection is uncommon in humans, but a recent study³⁷ involving 7 cases revealed an association between E coli urinary tract infections and diabetes mellitus with underlying heart disease. One of the 3 dogs with IE from E coliinfection in the present study had evidence of pyelonephritis at necropsy; another dog was suspected to have E coli IE on the basis of a history of recurrent E coli prostatitis. Neurologic complications occurred in almost half of the dogs with gram-negative bacillary endocarditis in the present study, although this rate was not significantly different from that in the other groups. Congestive heart failure is a common complication of gram-negative bacillary IE in humans, but CHF was not observed in any of the dogs with gram-negative bacillary IE. Although not significant, the magnitude of the febrile response in dogs with E coli infections was the highest of all groups, and the response was significantly higher than that associated with Bartonella spp infection. This finding seems logical considering the profound and rapid ability of gram-negative bacterial lipopolysaccharide to stimulate release of endogenous pyrogens such as interleukin-1 and tumor necrosis factor-a.³⁸

Other organisms that were detected in the present study included Mycobacterium spp, Actinomyces spp, Erysipelothrix spp, and a Lactobacillus-like organism, all of which have been associated with IE in humans.^39–42 Pyogranulomatous inflammation in multiple tissues, including severe pyogranulomatous sialoadenitis, was observed in the dog with mycobacterial IE. The nature of the inflammatory response was most suggestive of a nontuberculous, rapidly growing mycobacterium, but insufficient sequence data were available for determination of the infecting species. Mycobacterial IE in human patients is rare, nearly always involves prosthetic valves, and is usually associated with the rapidly growing mycobacterial species Mycobacterium fortuitum or Mycobacterium chelonae.³⁹ Mitral valve endocarditis from Actinomyces spp infection was recently reported in a Labrador Retriever¹; the Rottweiler in the present study had isolated aortic valve involvement, and diagnosis was made at necropsy. Studies of Erysipelothrix spp isolates from dogs with IE suggest that most, if not all, of such cases are caused by Erysipelothrix tonsillarum serovar 7,¹⁰ which has been isolated from healthy pigs and chickens.^43,44 The isolate in the present study was reported as Erysipelothrix rhusiopathiae, but typing to rule out E tonsillarumwas not performed. Endocarditis from Lactobacillus spp infection in humans is most commonly caused by Lactobacillus casei; two thirds of patients have underlying structural heart disease, and in nearly 50% of patients, dental procedures or conditions are reported as possible predisposing causes.⁴² The dog reported in the present study with Lactobacillus-like IE had severe diffuse gingivitis and dental calculus deposition, although evidence of abscess formation was not observed.

In dogs with microbial growth on cultures of blood, all bottles yielded bacterial growth in 77% of dogs, and when bacteremia was detected, the etiologic agent was identified from the first 2 cultures in all dogs. This finding was similar to findings in humans with IE, in which all MCB bottles are positive for bacterial growth in two thirds of cases, and the first 2 cultures yield the diagnosis in > 90% of cases.¹⁹ On the basis of that information, veterinary clinicians should perform at least 2, and preferably 3, MCBs within a 24hour period in dogs with suspected IE. In humans, antimicrobial administration within the previous 2 weeks decreases the rate at which MCBs yield bacterial growth,⁴⁵ leading to the recommendation¹⁸ that at least 3 blood samples be submitted for culture if the patient recently received antimicrobials.

In the present study, the MCB or MCBs yielded no growth in 29 of the 55 (53%) dogs in which MCB was performed. That value is higher than the values of 2.5% to 31% reported¹⁸ in human patients and higher than the value of 19% reported in an earlier study⁴ of dogs. The high prevalence of Bartonella spp infection in dogs with IE represents recently developed information, and lack of uniform testing in that population of dogs could be one of the reasons for the large number of dogs with negative results of MCB; IE secondary to Bartonellaspp infection was confirmed in 6 of the 29 dogs, and 11 dogs were not tested for Bartonella spp infection. One dog had a diagnosis of IE resulting from mycobacterial infection at necropsy. Other fastidious organisms that have been detected in humans with IE that could also play a role in dogs with IE include Coxiella burnetii, Chlamydiaspp, Brucellaspp, various anaerobes, Mycoplasmaspp, spirochetes, and nutrionally variant streptococci such as Abiotrophia spp.^18,26 Fungal infection may have been the cause of endocarditis in 1 dog in which A flavipes was isolated from a intervertebral disk aspirate specimen. Aspergillus spp is a common cause of fungal IE in humans.¹⁸ Serologic testing for Aspergillus spp infection could be considered in dogs with no growth on MCBs because it has been suggested^b that serologic testing may be more sensitive and specific than was previously thought, at least when used for diagnosis of nasal aspergillosis. Recent administration of antimicrobial drugs was not a significant contributor to negative MCB results in the present study. The lack of significance may be accurate or a result of insufficient case numbers or inaccurate information in the medical record regarding the timing of antimicrobial drug administration. Infective endocarditis caused by E coli infection was suspected in 2 dogs with recent histories of urinary tract infection that were referred for evaluation after antimicrobial drugs had been administered. Extrapolation of results of cultures of extracardiac sites to the cause of IE should be undertaken with caution, as concurrent infection with other bacteria is also possible. One dog in the present study had IE secondary to streptococcal infection and concurrent E coliurinary tract infection. Mural or right-sided heart lesions are another reason why results of MCB may be negative in human patients with IE, possibly because bacteria are released into and filtered by the pulmonary system; such lesions were observed in 3 dogs with MCB-negative IE in our study. Microbial cultures of valvular tissue at necropsy assisted in diagnosis in 3 dogs. Only 3 (4%) dogs had negative MCB results, were seronegative for Bartonella spp, and had not been receiving antimicrobials at the time of initial examination. None of those dogs had isolated mural lesions or IE involving the right side of the heart, and none underwent necropsy.

The number of dogs in the present study was insufficient to detect relationships between predisposing causes and infecting organisms. Some dogs had multiple potential predisposing causes. Whether infection of extracardiac sites (such as the lungs, lower portion of the urinary tract, peritoneal cavity, intervertebral disks, and sternebrae) developed prior to or after development of IE could not be ascertained. Interestingly, no diabetic dogs were included in the present study. Underlying malignancy was a more common association with IE, and all such dogs had a carcinoma or a sarcoma; no dogs had a hematologic malignancy such as lymphoma. Only a few dogs had underlying cardiac valvular malformations or disease; those have been described in a separate manuscript.²³ No obvious correlations between underlying valvular disease and IE resulting from infection with a specific infectious agent were observed.

Most of the organisms that were identified by use of routine culture methods in the present study were susceptible to a broad range of antimicrobial drugs.

Multiresistant organisms included Enterococcus spp isolates from 2 dogs, a P aeruginosa isolate from 1 dog, E coli isolates from 2 dogs, an S aureusisolate from 1 dog, and a C freundii isolate from 1 dog. Many isolates were resistant to enrofloxacin. All streptococci except for the E faeciumisolate were susceptible to penicillin; the latter isolate was susceptible to high concentrations of gentamicin. Low-level aminoglycoside resistance is intrinsic to enterococci because of their anaerobic metabolism.³⁴ Guidelines for the treatment of endocarditis in humans have been published recently.⁴⁶ A combination of ampicillin and an aminoglycoside or vancomycin is recommended for treatment for IE in humans with penicillin-resistant streptococcal or enterococcal infections. Penicillinase-resistant penicillins such as oxacillin, together with an aminoglycoside, are recommended for treatment for staphylococcal IE. The resistant S aureus isolate in the present study was resistant to oxacillin but produced b-lactamase and was sensitive to clavulanic acid and sulbactam. Penicillinaminoglycoside combinations are also recommended for treatment for IE secondary to gram-negative bacterial infections in humans, but in many patients, valve replacement is essential for resolution.^18,46 The C freundiiand P aeruginosaisolates were the most widely resistant organisms, but both remained susceptible to aminoglycosides and imipenem. Only aminoglycosides appear to have bactericidal activity against Bartonella spp.⁴⁷ Combination treatment with aminoglycosides (for 2 weeks) and penicillin, doxycycline, or a macrolide (for a minimum of 4 weeks) is recommended in humans with IE secondary to Bartonella spp infection,^46,47 although valve replacement is also necessary in many patients.

Limitations of the present study include its retrospective nature and the small numbers of cases in each group of infecting microorganisms. To maximize case numbers in each group, causative agents were grouped in categories that encompassed several species. Relationships between clinical features and the infecting species may have been obscured when the effect of infection with 1 species within a group was different from the effect of another. In humans, enterococcal endocarditis is more likely to be associated with CHF than is endocarditis caused by other streptococci.⁴⁸ Analyses involving large numbers of human cases have recently been facilitated by the creation of the International Collaboration on Endocarditis Merged Database, which includes prospectively collected data from cases from 5 countries.⁴⁸ This has permitted analysis of clinical characteristics associated with IE caused by uncommon organisms. Nevertheless, the number of cases in the present study was sufficient to reveal several important clinical features associated with various groups of causative organisms, findings that may aid in diagnosis, treatment, and prognostication in affected dogs.