• Open Access

Comparison of Polymerase Chain Reaction with Bacterial 16s Primers to Blood Culture to Identify Bacteremia in Dogs with Suspected Bacterial Endocarditis


  • Dr Heaney is presently affiliated with the Alpenglow Veterinary Specialty & Emergency Center, 3640 Walnut Street, Boulder, CO 80301. This study was performed at the Washington State College of Veterinary Medicine. Presented in part at the 2009 ACVIM Forum, Montreal, Canada.

Corresponding author: K. M. Meurs, Department of Veterinary Clinical Sciences, Washington State University College of Veterinary Medicine, Pullman, WA 99164; e-mail: meurs@vetmed.wsu.edu.


Background: Identification of the bacterial organism in dogs with endocarditis is challenging. Human studies have reported the utility of the polymerase chain reaction (PCR) to amplify and identify bacterial nucleic acid from infected valvular tissue and blood.

Hypothesis/Objectives: We hypothesized that PCR using primers designed to amplify the bacterial 16s gene would identify circulating bacteria in dogs with suspected bacterial endocarditis more consistently than standard blood culture techniques.

Animals: Eighteen dogs with suspected bacterial endocarditis based upon clinical and echocardiographic findings. Fifteen clinically normal dogs served as negative controls.

Methods: Prospective study of dogs evaluated for suspect endocarditis at 6 veterinary hospitals. A blood sample was drawn from all dogs and evaluated with both a single-sample PCR and standard 3-sample blood culture techniques.

Results: Blood culture identified noncontaminant bacteria in 6/18 study animals (33%) and 1 control dog; PCR identified noncontaminant bacteria in 7/18 study animals (39%). There were no study animals in which the 2 tests identified different bacteria (κ= 1.0). However, bacteria were identified by both techniques in only 2/18 study animals. When results from both PCR and blood culture were considered together, a noncontaminant bacterial organism was identified in 11/18 study animals (61%).

Conclusion and Clinical Importance: The results of this study suggest that although single sample PCR with 16s primers was not more sensitive than blood culture for detection of bacteremia in dogs with suspect endocarditis, performing both techniques simultaneously did increase the likelihood of identification of bacteria in blood.


polymerase chain reaction

Infectious endocarditis is a serious form of acquired heart disease in dogs that is frequently caused by bacterial infection of the valvular or mural endocardium.1 Morbidity is high and affected dogs can develop septicemia, septic emboli, congestive heart failure, arrhythmias, and sudden death.1 Treatment consists of aggressive long-term antimicrobial therapy directed against the offending bacteria, yet blood culture often fails to identify the causative organism. One study reported that the causative organism was identified by blood culture in only 41/71 dogs (58%).2 Successful identification of the etiologic agent through blood cultures is only slightly better in human medicine, with only 69% of suspected endocarditis cases yielding positive culture results.3 An advantage in human medicine is that the diseased valve is frequently removed and replaced with an artificial valve, allowing histopathologic evaluation and culture of the diseased valve.4 In studies that have used the 16s ribosomal primer based sequencing on whole blood from human patients with infectious endocarditis, polymerase chain reaction (PCR) has significantly greater sensitivity and specificity than conventional microbiological methods.4

We hypothesized that performing PCR with 16s ribosomal primers would be a more sensitive test for the identification of bacteremia than blood culture in dogs with suspected bacterial endocarditis and from control dogs, and that it could be used to identify the genus and species of bacteria involved by PCR-based sequencing. The objective of this study was to compare results from PCR from dogs with suspected infectious endocarditis to those obtained from blood cultures.

Materials and Methods

Eighteen dogs in which bacterial endocarditis was suspected based on an echocardiogram performed by a board-certified cardiologist or a cardiology resident under the direct supervision of a board-certified cardiologist were evaluated. Echocardiographic criteria for suspicion of endocarditis were based on a modified Duke's criteria for diagnosis of infective endocarditis and included discrete, echogenic oscillating mass on the valve or supporting structure, and new or worsening aortic or mitral valvular regurgitation, and thickening of the aortic valve not directly related to a congenital heart defect, and thickening of the mitral valve atypical for mitral valve degeneration, or only thickening of the mitral valve atypical for mitral valve degeneration.5

A second group of 15 adult clinically normal dogs with a normal cardiovascular physical examination and a normal echocardiogram were also included (negative controls).

Eight milliliters of blood were drawn from each dog aseptically 3 times, at least 1 hour apart and from 3 sites by standard technique for blood culture.1 All cultures were performed in the diagnostic laboratory of the hospital in which the dog was examined. Briefly, after skin preparation, blood samples were drawn with a needle and syringe and placed into 3 mL blood culture bottlesa and maintained at room temperature until arrival at the laboratory. Each tube was processed and used to inoculate 4 agar plates, including 2 Brucella blood agar platesb for anaerobic incubation and 2 Chocolate agar platesb for incubation in air supplemented with 5% CO2. Plates were observed daily and specimens were considered negative if no bacterial growth was observed after 7 days of incubation. Plates were considered positive if at least 2 of the 3 blood culture samples were positive for the same bacteria.

Two milliliters of the 8 mL blood sample taken for the first blood culture sample were placed in an EDTA tube for the PCR and were sent overnight with a cold pack to Washington State University for PCR analysis. Within 60 minutes of arrival they were treated with the MolYsis Plus kitc to extract bacterial DNA according to the manufacturer's recommendations. PCR was performed by standard technique with proprietary primersc targeted at the 16s conserved ribosomal region of bacterial DNA. The PCR was performed using a mastermix with a total volume of 24 μL (2.5 μL NH4 buffer, 1.0 μL dNTPs, 15.75 μL dH2O, 0.25 μL Platinum Taq,d 2.5 μL MgCl2, and 25 pmol in 1.0 μL each of the forward and reverse primers). One microliter of DNA was added, and the reaction mix amplified with the program of 94°C for 4 minutes and 36 cycles of 94°C for 30 seconds, 55°C for 60 seconds, 72°C for 60 seconds, and 72°C for 10 minutes. Amplification reactions were analyzed by agarose gel electrophoresis, and amplicons were sequenced with primers targeted at the 16s conserved ribosomal region of the bacterial DNA (F—5′-TCC TAC GGG AGG CAG CAG T, R—5′-TCC TAC GGG AGG CAG CAG T). Sequences were analyzed and the genus of bacteria was identified by the microbes section of BLAST (National Center for Biotechnology Information, http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi). Bacteria that were known to be common contaminants (Bacillus spp., Corynebacterium spp., Propionibacterium acnes, Clostridium perfringens) were disregarded.6

Positive-control samples of Staphylococcus aureus (ATCC 29740) and Escherichia coli (ATCC 25922) were obtained from bacterial colonies and extracted in a similar fashion to the blood samples. DNA extracted from each control bacteria was evaluated by PCR using the 16s primers to ensure amplification and identification of the correct DNA.

Statistical Analysis

The difference between the proportion of endocarditis cases in which a noncontaminant bacteria was identified by each technique (single sample PCR and 3 sample blood culture) was evaluated with a McNemar's test. A P-value of <.05 was considered to be significant. Agreement between the 2 techniques was evaluated with a κ statistic in 2 ways: agreement regarding whether or not bacteremia was identified; and agreement regarding the species of bacteria that was identified when both tests were positive.


The 18 dogs with suspected endocarditis and the 15 clinically normal dogs were evaluated by both techniques.

The average age of the suspected endocarditis group was 8 (range 2–13) years. Eleven were male (5 intact, 6 castrated), 7 were female (2 intact, 5 spayed). They included several breeds: Boxer (5), mixed breed (2), and 1 each of Alaskan Malamute, Doberman Pinscher, Weimaraner, German Shepherd, Labrador, Brittany, Greyhound, Hovawart, Papillion, Boston Terrier, and Maltese. Based on echocardiographic examination, 7 had aortic valve lesions, 9 had mitral valve lesions, and in 2 dogs both valves were affected.

Blood culture identified noncontaminant bacteria in 6/18 dogs (33%). PCR identified noncontaminant bacteria in 7/18 dogs (39%) and a likely contaminant (based on the weakness of the PCR product, and documentation as a contaminant in the literature6), Propionibacterium acnes, in 1 case (Table 1). Statistically, the proportion of times each test identified bacteria was not different. An organism was identified by one or both of the tests in 11/18 of the dogs (61%) but there were only 2 dogs in which both tests identified bacteremia (κ agreement of −0.8 for the detection of bacterermia). In the 2 dogs in which both blood culture and PCR identified bacteremia (Enterococcus faecalis, S. aureus) the species of bacteria was the same for both assays. There were no dogs in which both tests identified an organism but the organism identified was not the same.

Table 1.   Blood culture and polymerase chain reaction (PCR) results for dogs with suspected endocarditis.
BacteriaPCRBlood CultureValveTotal Cases
  • a

    One of the cases of Staphylococcus and the case of Enterococcus was identified by both tests.

  • b

    Identified as either Brucella abortis or mellitensis by sequencing.

Staphylococcus aureus2a2aMitral3
Streptococcus species12Mitral, aortic3
Brucella speciesb10Aortic1
Enterococcus faecalis1a1aAortic1
Bartonella vinsonii10Mitral, aortic1
Granulicatella adiacens10Aortic1
Actinomyces viscosus01Aortic1

There were 5 dogs in which PCR identified an organism that was not identified by blood culture, the bacteria identified included Staphylococcus, Granulicatella, Streptococcus, Bartonella, and Brucella. Likewise, there were 4 dogs in which blood culture identified an organism that was not identified by PCR, the bacteria isolated in those instances included 2 dogs of Streptococcus spp., 1 S. aureus, and 1 Actinomyces viscosus.

An organism was not identified in 7 dogs involving 3 aortic and 4 mitral valves.

The clinically normal population had an average age of 6 (range 2–11) years. Eleven were male (2 intact, 9 castrated), 4 were female (all spayed). The population included the following breeds: mixed breed (5), Labrador (2) and one each of Boston Terrier, Vizsla, Great Dane, Dalmatian, Dachshund, Jack Russell Terrier, Miniature Schnauzer, and Australian Cattle Dog. Blood culture identified a noncontaminant bacteria (S. aureus) in 1/15 dogs. PCR did not identify bacteria in this dog or any others.

DNA extracted from each positive-control bacteria amplified, and the sequence correctly matched the control samples.


We compared the identification of bacteria by PCR using 16s ribosomal bacterial DNA primers to standard blood culture technique to determine the optimal way to identify circulating bacteria in dogs with suspected bacterial endocarditis. Although PCR did not amplify bacterial DNA in more cases than did blood culture (39 and 33% of the cases, respectively), performing both techniques simultaneously did increase the likelihood of detection and identification of bacteria (61% of the cases).

The prognosis for dogs with endocarditis is generally poor.1 It is possible that more accurate and earlier identification of the offending bacteria could lead to an improved clinical outcome. Bacterial culture to identify bacteremia is considered a standard diagnostic test in the evaluation of dogs with endocarditis but it has limitations. Dogs with suspected endocarditis are commonly administered antibiotics before a blood culture can be performed. Current or recent administration of antimicrobial therapy could affect bacterial viability and reduce the likelihood of a positive bacterial culture. Additionally, some causative agents such as Bartonella spp. and Coxiella burnetii are difficult to culture.7 The observation that Bartonella species can account for up to 19% of dogs with endocarditis in some geographic regions underscores the importance of these organisms in dogs and the need for methods that can more accurately detect these organisms.2 This could have been true in our study as well. Bartonella vinsonii was identified in 1 dog and it was identified only by PCR.

PCR amplification and sequencing appears to have a number of advantages over blood culture for detection and identification of circulating bacteria. Even bacteria no longer viable because of antimicrobial use may be identified if the bacteria are still present in the blood. Additionally, even for more unique and difficult to culture bacteria, the conserved regions of the bacterial DNA are present and can be detected with PCR using the 16s ribosomal primers. Although there may be concerns that the high sensitivity of PCR may result in false positives, studies of humans with endocarditis have demonstrated very few false positive results.8 This would appear to be true in this study as well because PCR did not amplify bacterial DNA in any of the control dogs. The PCR does, however, have limitations. In our study there were 4 dogs in which bacteria were identified by blood culture but not by PCR. The ability of PCR to amplify bacterial DNA can depend on a number of factors including the minimum number of bacterial cells in the sample and the presence of inhibitory factors. Potential inhibitory factors include constituents of bacterial cells and common laboratory items such as glove powder and laboratory plasticware.9 Additionally, although the 16s gene is believed to be highly conserved across all bacteria species, it is possible that some species have subtle sequence differences that could prevent their amplification with the primers used in this study. Although the 16s primers used here have been validated with many different bacterial species including Staphylococcal spp. (aureus, intermedius, coagulase positive, coagulase negative), Streptococcus spp. (canis, bovis, and b-hemolytic), Escherichia coli, Pseudomonas, Pasteurella, and Corynebacterium, they have not been specifically validated for Actinomyces viscous. A. viscous was one of the organisms identified by blood culture but not by PCR. PCR cannot provide the information on sensitivity to antibiotics that a bacterial culture can; however, in the absence of a successful culture the ability of PCR to identify a bacterium can allow therapy to be chosen based on the type of bacteria. Finally, although the 16s primers were useful to amplify a bacteria and in most cases the sequence produced was sufficient to identify both the genus and species of bacteria, in 1 dog, Brucella, the sequence could not differentiate between 2 species. In this situation additional PCR primers would need to be designed in order to identify the species.

In the study performed here, if both PCR and bacterial culture were performed, an infectious agent was identified in 61% of the dogs with suspect bacterial endocarditis. This might appear to be only slightly better than blood culture alone.2 However, in that study all but one of the dogs were definitively diagnosed with bacterial endocarditis. In the present study, a clinical diagnosis of endocarditis was suspected based on clinical suspicion. It is possible that some samples found to be negative for PCR and culture were negative because the dogs did not have a circulating infectious agent and did not have endocarditis. Confirmation would have been facilitated by a necropsy. Additionally, although we did identify noncontaminant bacteria in 61% of the dogs, the lack of postmortem culturing of the lesions prevents us from knowing if these bacteria were associated with bacterial endocarditis. Regardless, in a disease where successful treatment could depend upon the early identification of a bacterial agent, use of both PCR with 16s primers and blood culture could be clinically beneficial.

A limitation of this study is the possible inconsistency in blood culture technique because the blood cultures were performed at the 6 different institutions from which the dogs were recruited rather than by 1 microbiology laboratory. There could have been variability in methodology from institution to institution, which could have impacted the culture results. However, all clinical diagnostic laboratories used the standardized culture techniques as described above. It could have been optimal to have all cultures performed within 1 laboratory but would have required shipment of the blood culture samples, and exposure to environmental factors that could impact bacterial viability and the ability to grow the bacteria on a culture. All PCR reactions were done in a single laboratory after overnight shipment. Although these samples would have been exposed to the same environmental risks as mentioned above, PCR does not need the bacteria to be viable for analysis.

The results of this study suggest that performing both PCR with 16s primers and blood culture could increase the likelihood of identifying a bacterial agent in dogs with suspected bacterial endocarditis.


aInverness Medical, Princeton, NJ

bHardy Diagnostics, Santa Maria, CA

cMolYsis Plus kit, Molzym, Bremen, Germany

dPlatinum Taq, Invitrogen, Carlsbad, CA


This work was graciously funded by the Morris Animal Foundation.