Literally translated, infective endocarditis (IE) is inflammation of the endocardial surface of the heart due to the invasion by an infectious agent. In reality, almost all endocarditis that is recognized in dogs and cats involves valve tissue and the vast majority of the infectious agents recognized are bacterial. However, in rare situations the mural endocardium can become infected and rarely fungal or rickettsial organisms can cause IE. When a bacterial organism is known to cause IE, the appropriate name for the disease is bacterial endocarditis. When bacteria colonize a heart valve, they generally produce proliferative lesions (vegetations) and/or destroy valve tissue (Figure 1). Vegetations usually result in improper valve coaptation resulting in regurgitation. The vegetations may also result in a narrowed valve orifice, creating stenosis, but this is unusual. Valve destruction results in regurgitation in all instances. Other common names for infective endocarditis are vegetative endocarditis and acute or subacute bacterial endocarditis (SBE). Acute bacterial endocarditis is usually associated with a virulent organism, such as E. coli. The diagnosis is usually made within two weeks of the onset of clinical signs beginning. Subacute bacterial endocarditis is associated with organisms of low virulence and the disease evolves over several weeks to months. Much of the disease seen in dogs and cats appears to be acute and commonly follows a malignant course.

Incidence

Although several studies have been performed to look at the incidence of IE in dogs, most of them have severe limitations and so will not be reported here. Suffice it to say that IE is a rare lesion. The diagnosis of infective endocarditis was made in 45 canine patients in our hospital between 8/1/86 and 8/1/96 or approximately 5 cases per year. This probably means that a veterinarian in private practice will see a case of bacterial endocarditis once every several years, at most. Infective endocarditis is more common in dogs than in cats. Infective endocarditis primarily affects the mitral and aortic valves in dogs and cats. The tricuspid valve is rarely affected and the pulmonic valve is almost never affected. In the 45 canine patients from our hospital, 22 had lesions confined to the aortic valve cusps, 16 had lesions confined to the mitral valve leaflets, one had both the aortic valve and mitral valve involved, and one had tricuspid valve endocarditis. In five the valve involved could not be confirmed from the record.

Infective endocarditis is a disease of medium-sized to large dogs and of purebred dogs. Only 8 of the aforementioned 45 cases from our hospital were non-purebred dogs. The most commonly affected breeds in our hospital included German shepherds or German shepherd crosses, golden retrievers, and Labrador retrievers. The incidence in the retrievers probably reflects the fact that these breeds are popular and therefore we examine many of them. German shepherds appear to be over represented. The only two dogs under 10 kg that we saw in that same period were a Chihuahua and a toy poodle. All other dogs were >15 kg in weight. None of the cases were identified following a dentistry.

Infective endocarditis is rare in cats. During the same period that we identified 45 dogs with IE, we diagnosed 7 cats with IE. Five of these had IE of the aortic valve and two of the mitral valve.

Bacteremia must occur for bacteria to colonize a heart valve. A transient bacteremia apparently is a common event. In humans, bacteremia occurs secondary to such diverse events as defecation and teeth brushing. That transient bacteremia is a normal event in humans is suggested by the fact that circulating antibody titers to normal flora are identified in healthy humans. Bacteria that gain entry to the bloodstream are usually rapidly removed by macrophages in the liver, bone marrow, and spleen and by neutrophils in capillaries.

Physical manipulation of mucosal surfaces (urinary catheterization, dental scaling) produces bacteremia in humans. Dental scaling is a primary concern of most veterinary practitioners for producing bacteremia. Consequently, the administration of prophylactic antibiotics to patients with cardiac disease undergoing dental procedures is commonly recommended. To place this subject in perspective, one investigator has documented that in people, 40% are bacteremic after dental extractions but 38% are bacteremic after hard chewing, and 25% are bacteremic after tooth brushing or oral irrigation. Two recent studies have been performed looking at bacteremia in dogs during dental procedures. In the first, 67% of the dogs had bacteremia. This study identified numerous organisms, including Streptococcus, Staphylococcus, Pasteurella, Acinetobacter, Neisseria, Pseudomonas, and both gram positive and gram negative anaerobic organisms. One-half of these dogs were administered sodium penicillin G prophylactically which did not reduce bacterial numbers. A subsequent study examined two groups of 30 dogs. The first group was anesthetized or sedated only and the second group was anesthetized and underwent dental scaling. Approximately 30% of the dogs in both groups were bacteremic. Five of the dogs that had the dental procedure had Pasteurella spp. cultured from their blood while none of the other dogs did and four of the sedated dogs had Staphylococcus spp. cultured.

In humans, fewer than 20% of patients with IE have a history of a medical procedure before developing IE. In our experience, most canine patients do not have any history of a previous medical procedure and in many the source of the infection cannot be ascertained. In one study, however, the investigators were able to suggest or define a route of entry or a source of infection in 62% of their cases of IE in dogs.

It is the general impression of the authors that veterinarians commonly administer antibiotics prophylactically to dogs undergoing dental scaling, especially to dogs with myxomatous mitral valve disease. From the data presented above one might conclude that it makes just as much sense to administer antibiotics to all patients that are being sedated. The authors have not witnessed a dog with myxomatous mitral valve disease develop infective endocarditis following a dental procedure. Consequently, we question whether these patients are at a greater risk of developing endocarditis. The evidence for dentistry-induced IE in the literature is lacking. One report has stated that of 61 dogs studied retrospectively, 8 were greater than 7 years of age, and 6 had evidence of myxomatous mitral valve degeneration. All 6 dogs had endocarditis of the mitral valve. There was no mention of a past history of dental procedures in these dogs. One other report from 1979, lists lack of antibiotic prophylaxis after dentistry in 2 poodles diagnosed with infective endocarditis. One of these dogs had a typical murmur of mitral regurgitation, a history of heart failure, and survived. Its other clinical signs were a fever and an elevated white blood cell count. The other dog had no heart murmur, a fever, and an elevated white blood cell count. The diagnosis of endocarditis was based on a positive blood culture. The diagnosis could be questioned in both these dogs. We also do not believe that the antibiotics that are commonly used prophylactically with dental procedures, such as penicillin, streptomycin, and ampicillin, have a broad enough spectrum of activity to be beneficial, even if they were warranted.

The primary difference between the types of organisms isolated in humans and dogs with infective endocarditis is that dogs have a higher incidence of gram negative infections. In four studies, Staphylococcus aureus accounted for approximately 25% of the organisms isolated, hemolytic and non hemolytic Streptococci for 20%, and Escherichia coli for 25%. Corynebacterium was isolated in 10% and Pseudomonas in 6%. Erysipelothrix sp. accounted for 3% of the cases. This unusual organism has been reported elsewhere and has recently been determined to be E. tonsillarum, not E. rhusiopathiae, the swine pathogen. Bartonella vinsonii, a rickettsial organism, has been reported in dogs with culture negative vegetative endocarditis.

Circulating microorganisms must attach to the endocardial surface of a valve for IE to occur. In humans, most (60-80%) patients with IE have an identifiable predisposing cardiac lesion that makes it easier for the organism to attach. This is not be the case in dogs and cats. Most cases of IE in animals occur on apparently normal valves. In one study, out of 61 dogs with IE only four had congenital heart defects. Of these four, three had subaortic stenosis. No evidence exists to support a relationship between myxomatous degeneration of the mitral valve and IE.

In humans, platelets aggregate on valves damaged by congenital or rheumatic disease. Platelet aggregation occurs because of endothelial damage that exposes collagen fibers. Microscopic platelet thrombi may stabilize to form larger lesions (so-called nonbacterial thrombotic endocarditis). Experimentally these lesions can be colonized by circulating bacteria, resulting in endocarditis.

The fact that most dogs with IE do not have underlying cardiac disease makes the canine disease distinctly different from the human disease. Why do bacteria lodge on apparently normal heart valves more readily than on other normal structures in the cardiovascular system? Multiple factors likely exist. That heart valves are subject to constant trauma as they open and close with each cardiac cycle. This trauma probably results in very minute damage to the endothelial surfaces of the valves. This may explain why valves are more prone to IE. That trauma to valve leaflets is part of the pathogenesis of IE is suggested by the fact that the incidence of IE is directly related to the force placed on each valve. The mitral valve is affected in 86% of human patients and is subjected to approximately 116 mmHg systolic pressure, the aortic valve is affected in 55% of human cases and is subjected to approximately 72 mmHg diastolic pressure, the tricuspid valve is affected in 20% of cases and is subjected to approximately 24 mmHg systolic pressure, and the pulmonic valve is affected in 1% of cases and is subjected to approximately 5 mmHg diastolic pressure. Normal valves may also develop nonbacterial thrombotic endocarditis. However, in humans this is more common in patients that are systemically ill such as those with advanced malignancy, uremia, systemic lupus erythematosus, etc.

Bacteremia is common and IE is rare. Why then do some dogs and cats with apparently normal valves develop IE while others do not? This is unknown. Some factors that might contribute are as follows. First, the type of organism is probably an important factor. Some virulent bacteria, such as Staphylococcus, Streptococcus, and Pseudomonas, can adhere to normal endocardial surfaces. In gram negative bacteremia, serum bactericidal activity appears to play an important role such that only serum-resistant isolates of E. coli, Pseudomonas aeruginosa, and Serratia marcescens produce bacterial endocarditis in rabbits. Of course, many dogs are probably exposed to bacteremia with similar organisms and most do not develop endocarditis. It seems likely that immune factors could play a major role in determining if a dog or cat develops IE. Rabbits homozygous for C₆ deficiency develop IE more readily than normal rabbits. Antibodies are known to protect against endocarditis by either inhibiting adhesion of the bacteria or clearing the blood of the organism more rapidly. Abnormalities in antibody production or function could predispose to endocarditis. German shepherds were more commonly affected with bacterial endocarditis in the southeastern United States in the 1980s. Some type of inherited immune system abnormality might have produced this phenomenon. When one examines the breeds reported in the literature with IE, very few mixed breed dogs are represented. This could reflect the demographics of a referral population or it could suggest that hereditary abnormalities, such as an inherited immune system abnormality, might be involved.

Dogs with subaortic stenosis are one group of dogs with pre-existing cardiac disease at risk for developing IE. In subaortic stenosis, turbulent, high velocity blood flow develops in the subvalvular region. This jet traumatizes the aortic valve leaflets during each systolic interval. This damage is apparent from the fact that these otherwise normal valve leaflets consistently become insufficient early in life. Presumably, this injury to the valve makes it easier for bacteria to colonize on the leaflet endothelium. Dogs with subaortic stenosis should be administered prophylactic antibiotic therapy during procedures known to produce bacteremia.

Administering prophylactic antibiotics to dogs with other congenital cardiac abnormalities also may be rational based on experience in humans. One author has suggested that dogs that are at risk be administered penicillin G, ampicillin, or amoxicillin based on one study that showed that the oral flora of normal dogs and dogs with dental disease consisted primarily of anaerobic bacteria. However, anaerobic bacteria have been reported only once as an isolate from a dog with IE and another author has identified aerobic organisms from dog and cat mouths. Anaerobic organisms have not been isolated from any dog in the four most recent reports of dogs with IE. Because of this and because of the high incidence of gram negative cases of IE in dogs, we do not feel that the aforementioned antibiotics are the best choices.

The vegetations in IE vary in size and shape from small, warty nodules to large cauliflower-like masses. Histologically, vegetations consist of colonies of bacteria imbedded in a fibrin-platelet matrix. This matrix protects the bacteria from being phagocytized by leukocytes. The bacteria in the matrix are packed tightly (e.g., 10⁹ - 10¹⁰ organisms per gram). Consequently, they metabolize and reproduce slowly. Slow reproduction and protection from phagocytosis make it difficult to kill bacteria in vegetations, especially with a bacteriostatic antibiotic.

Vegetative lesions commonly break free and are carried to distal sights within the arterial tree where they lodge (embolization). In humans, 15-35% of patients have clinical evidence of embolization while 45-65% have evidence of embolization at an autopsy. In one report where 44 dogs with IE were examined at necropsy, 84% had evidence of systemic embolization. The kidney had evidence of infarction in 64% and the spleen was infarcted in 45%. Clinical signs of embolization develop more commonly in humans infected with organisms that tend to produce large vegetations that are more prone to dislodge. These organisms include Staph. aureus, streptococci, and fastidious gram-negative organisms. Emboli that originate from the mitral or aortic valves lodge in numerous places in the body including myocardium, brain, limbs, kidneys, spleen, bowel, and iliac artery. Resultant clinical problems can include neurologic signs, arrhythmias, lameness (fixed or shifting), renal failure, gastrointestinal signs, abdominal pain, and posterior paresis and pain. The emboli are septic and so produce regions of infection and inflammation. This may result in exacerbation of nonspecific clinical signs at the time of embolization, especially fever.

Infective endocarditis usually destroys valve tissue or interferes with valve function, resulting in valvular regurgitation (Figures 2, 3, 4, and 5). Rarely it will result in clinically significant stenosis (Figures 6 and 7). Aortic valve endocarditis often results in severe aortic regurgitation. Both acute and chronic severe aortic regurgitation are poorly tolerated by the cardiovascular system. Dogs with severe aortic regurgitation secondary to IE commonly present in left heart failure and usually die of this complication within a short time. In dogs with acute bacterial endocarditis, the left heart may not have time to compensate for the severe regurgitation. Consequently, bacterial endocarditis should always be a differential diagnosis when pulmonary edema is identified radiographically without left heart enlargement (Figure 8). Mild to moderate mitral regurgitation can be tolerated for prolonged periods. Even severe mitral regurgitation may be successfully managed medically for months.

Persistent bacteremia in IE stimulates both the cell-mediated and the humoral immune systems. This has not been well studied in dogs or cats although hyperglobulinemia and glomerulonephritis have both been reported in dogs with IE. In humans, circulating immune complexes (containing IgG, IgA, IgM, and complement) are found in almost all patients with IE that have positive blood cultures. These complexes are deposited along the glomerular basement membrane and in joint capsules where they cause glomerulonephritis and arthritis. Clinical signs associated with IE in dogs can sometimes mimic those of immune-mediated disorders. Arthritis, either septic or non septic, is a common sequela to IE in dogs. Occasionally an antinuclear antibody test or Coombs’ test will be positive in a dog with IE.

Renal failure can be a devastating complication to IE. It can occur secondary to immune complex glomerulonephritis or secondary to renal infarction. Glomerulonephritis may result in significant protein loss and hypoalbuminemia (nephrotic syndrome).

The vast majority of dogs with IE are large dogs. Less than 10% of the dogs with IE reported in the literature are <15 kg. Most dogs are more than four years of age. The male-to-female ratio was 2:1 in one study. Most are purebred dogs. The disease is rare in cats

A history of a predisposing factor (previous bacterial infection, recent dental procedure, immunosuppressive drug therapy, intravenous or urinary tract catheterization, trauma) is uncommon in canine or feline patients with IE. Patients are most commonly presented because they appear systemically ill to the owner. They are often depressed and weak. They commonly are not eating well or eating at all. Weight loss is frequent. Other dogs have signs of left heart failure. The owners often note respiratory abnormalities (tachypnea, dyspnea, cough) in this situation. Concomitant discospondylitis is a common problem.

Systemic septic emboli commonly lodge in various regions of the body and can create a multitude of clinical signs. Very small emboli may take the first exit that they can off the aorta and so travel down a coronary artery, creating a small, septic myocardial infarct. This can cause sudden death or an arrhythmia. Slightly larger emboli may take the first large exit off the aorta, which is the brachiocephalic trunk. They then take the first exit off the brachiocephalic trunk (the right subclavian artery) and lodge in the right front leg. This causes a right front leg lameness. Although right front leg lameness is most common in the authors’ experience, other legs are commonly involved. Larger emboli can travel to numerous regions of the body creating infarcts in kidney, brain, spleen, or bowel. Appropriate clinical signs are produced. Lameness is a common clinical sign in dogs with IE. It may be due to septic embolization of skeletal muscle or it may be due to arthritis. Two types of arthritis have been described in dogs with IE. One is septic arthritis, presumably secondary to bacteremia or embolization. The other is a sterile arthritis, thought to be due to immune complex deposition that occurs secondary to the bacterial antigenemia. The lameness may be stable or shift from leg to leg.

On physical examination, most canine patients will have a fever or a recent history of a fever. Patients that are already being administered antibiotics or corticosteroids or patients that are in severe heart failure may be afebrile. A heart murmur is also a common finding. Discovering a new heart murmur in a patient that is febrile is the classic finding to make one suspicious of IE. Of course, most clinicians realize that classic findings usually do not occur in most cases. In one study, only 25 out of 61 cases of dogs with IE had a new or latent heart murmur. A systolic heart murmur is the most common. In IE it can be heard with either mitral or aortic involvement. The murmur associated with mitral valve endocarditis is that of mitral regurgitation. In one report of 24 dogs with aortic valve endocarditis, 23 had a systolic heart murmur although the intensity was not reported. Presumably at least some of these systolic murmurs were due to some degree of aortic stenosis. Others may have been due to concomitant disease, such as myxomatous AV valve degeneration, or may have been flow murmurs secondary to fever-induced high cardiac output. The significance of a systolic heart murmur may be difficult to ascertain in an individual patient. If the murmur is grade III or louder and can be documented to be new and the patient is febrile, IE must be a primary differential diagnosis. Being certain that a systolic heart murmur has only occurred recently, however, is often difficult. A loud heart murmur in a young dog with a fever examined previously by a veterinarian, however, should be considered a new murmur until proven otherwise. Even in an older, large breed dog, a loud systolic heart murmur is not common. Consequently, it can be considered highly suspicious for recent valve tissue destruction if the patient has or has had a fever. A loud systolic murmur in a small breed, geriatric dog (even one with a fever) is most commonly due to myxomatous mitral valve degeneration, not to IE. A soft systolic murmur in large dog can also be due to IE but can also occur with numerous other cardiovascular lesions. It can also be due to an increased stroke volume secondary to fever.

A diastolic heart murmur due to aortic regurgitation is commonly identified in dogs with aortic valve endocarditis. The murmur of aortic regurgitation is often soft and so difficult to identify. Consequently, one must auscult any patient suspected of having IE very carefully in a quiet room. The murmur is heard best over the left base. It is blowing in character. It starts immediately after the second heart sound and decreases in intensity (decrescendo) through diastole. Infective endocarditis is by far the most common cause of an audible diastolic heart murmur secondary to aortic regurgitation in the dog and cat. Consequently, identification of a diastolic heart murmur heard best over the left base in a dog or cat, especially in one with a fever, should be regarded as caused by IE until proven otherwise.

Dogs with aortic regurgitation often have an increase in pulse pressure (bounding pulse; Figure 9). Systolic pressure is commonly increased because of the increased stroke volume that the left ventricle must generate in systole to compensate for the leak. The diastolic pressure is decreased because of blood flowing back into the left ventricle in diastole.

In humans, the diagnosis of IE, before the arrival of echocardiography, was typically based on clinical signs of infection (fever, leukocytosis, etc.) and positive blood cultures combined with a new regurgitant heart murmur. Alternatively, it was based on positive blood cultures and predisposing heart disease and evidence of embolic disease. In patients with negative blood cultures, the patient had to have a fever, a new regurgitant heart murmur, and evidence of embolic disease to be definitively diagnosed with IE. When transthoracic echocardiography became available, it was used as an adjunctive diagnostic tool. Because transthoracic echocardiography lacked sensitivity and specificity for diagnosing IE in humans, it was not included in listings of diagnostic criteria. Many false negative findings in humans with IE can be attributed to the relatively poor transthoracic image produced in many adult humans. Others can be attributed to the need to use a low frequency transducer in adult humans. The resolution of a low frequency transducer is poorer than that of a higher frequency transducer resulting in a reduced ability to identify smaller vegetations. More recently, the use of transesophageal echocardiography has markedly increased the image quality in humans and the sensitivity of identifying IE has improved to approximately 90%. Because of this documented improvement in sensitivity, more recent recommendations for diagnosing IE have included a positive echocardiographic (transthoracic or transesophageal) finding as a major criterion for diagnosing IE. However, even this has been disputed.

In small animal veterinary medicine, higher frequency transducers are routinely used and echocardiographic image quality is usually very good to excellent. In fact, the image quality is usually so good that, in our experience, transesophageal echocardiography usually provides no distinct advantage over transthoracic echocardiography in cats and only provides more detailed information about heart base structures in dogs. We feel that our ability to examine the aortic and mitral valves with transthoracic echocardiography in dogs and cats rivals that of transesophageal echocardiography in adult humans. Because of this and other factors mentioned later in this paragraph, echocardiography has been the major tool used to diagnose IE in dogs and cats for the past 10 years at our institution. Most patients that we diagnose with IE are presented with clinical signs that suggest IE and the diagnosis is made using echocardiography. Blood cultures are then used to attempt to identify the offending organism. Blood cultures are almost never used to make the definitive diagnosis of IE because they lack sensitivity and specificity in dogs and cats. Of dogs with IE, blood cultures are positive in 50-90% of the cases (lack of sensitivity). The variability in the percentage of positive cases is probably due to the variability in skill and diligence of various microbiology laboratories as well as the criteria used for diagnosing IE. In one previous report of dogs examined before the widespread use of echocardiography (before 1982), blood cultures were positive in 88% of cases diagnosed with IE. However, one of the primary diagnostic criteria for the diagnosis of IE in this report was a positive blood culture. Currently, most of our cases are initially diagnosed based on clinical signs (usually fever, leukocytosis with a heart murmur or signs consistent with systemic embolic disease) and echocardiography and then blood cultures are obtained. In this situation, we estimate that our laboratory’s success rate at identifying the offending organism is closer to 50%. This poor success rate may be related to prior antibiotic therapy (a common problem), fastidious organisms, laboratory technique, or noncontinuous shedding of organisms in dogs and cats with IE (species difference from humans). Since the advent of the use of echocardiography as a major diagnostic tool for diagnosing IE in humans, the number of culture negative cases has increased to as high as 41% in one series of cases.

Blood cultures are never used definitively to make a diagnosis of IE because blood cultures are also positive in other diseases (lack of specificity). In one study, of 165 dogs with positive blood cultures, only 45 were diagnosed as having IE. A similar number (42) had discospondylitis. Presumably the rest were septicemic secondary to other diseases. Critically ill animals frequently develop sepsis and have positive blood culture results. Dogs with discospondylitis can have clinical signs that closely mimic IE.

Echocardiography, however, is not 100% sensitive or specific. In humans, vegetations less than 3 mm in diameter usually go undetected. Modern two-dimensional echocardiography equipment can resolve down to less than 1 mm. It is difficult to believe that in a small patient, such as a cat, that a lesion 2 mm in diameter or smaller could not be visualized since a normal left ventricular wall in a cat can be 3 mm in diameter (Figure 10). Consequently, we believe that it is incorrect to transfer statements regarding the ability to resolve or detect a lesion based on size from human texts directly to the situation in veterinary medicine. Instead, we believe that we can identify smaller lesions that are present in smaller animals with modern equipment. Besides identifying vegetations, dogs with destructive lesions of their valves but without vegetations can be diagnosed with IE based on the presence of a regurgitant lesion and the echocardiographic appearance of the valve, especially when the aortic valve is involved (Figures 11 and 12). It has been stated in the human literature that the echocardiogram is positive in up to 80% of patients with IE but the usual sensitivity is 60%. We believe that our sensitivity is >90%. In fact, we think that missing the diagnosis of IE when echocardiography is used is very unusual in our clinic. Specificity (false positive diagnoses) can sometimes be a problem. This occurs primarily in the older small breed of dog with myxomatous mitral valve disease. The degenerative lesions that involve the mitral valve in these dogs can be mistaken for vegetations. In dogs with no history of fever, lameness, etc., the abnormal appearance to the valve is almost always due to myxomatous degenerative change. Differentiation between mitral valve degeneration and IE in an older, small breed dog with a fever or other evidence of infection may be impossible, however. Blood cultures do not necessarily make the distinction possible either, however, if the patient is bacteremic or septic from another cause. In some cases, the echocardiographic appearance of a vegetative lesion is different from that of a myxomatous lesion (Figure 13). These cases may have a lesion or lesions that appear more isolated and echodense than myxomatous lesions. False negative findings may be a problem with tricuspid valve lesions. In one report, endocarditis lesions on the tricuspid valve were not detected in the two dogs examined.

We have devised a list of diagnostic criteria used in diagnosing IE in dogs and cats (Table 1). From these criteria we have devised a diagnostic scheme for diagnosing IE in dogs and cats (Table 2).

Blood cultures must be obtained to try to identify the offending organism and subsequently to identify the appropriate antibiotic in patients with IE. Although bacteremia is continuous in bacterial endocarditis in humans and so thought to be continuous in animals, this may not be true in dogs and cats. It certainly does not translate into a 100% success rate at identifying the offending organism in veterinary patients with endocarditis. The bacterial numbers are usually small (<100 bacteria/ml of blood) and can be very small (<10 bacteria/ml of blood) in humans and are probably the same or less in veterinary patients. The bacteria are collected, in this situation, in blood that contains elements that normally inhibit bacterial growth (for example, complement) or destroy bacteria (for example, granulocytes). Consequently, expecting to be able to identify all patients that are bacteremic is unreasonable. If one combines the data from the four largest and most recent reports of IE in dogs where a positive blood culture was not a prerequisite for the diagnosis, out of 78 dogs, 57 (73%) had a positive blood culture and 21 (27%) had a negative blood culture. Because a positive blood culture strengthens the diagnosis and not all these dogs had echocardiograms, we believe that these data are probably optimistic if one uses them to predict the number of veterinary patients with IE that will have a positive blood culture. Even so, this leaves at least 25% of patients that must be treated for their disease without the benefit of a positive culture and a determination of microbial antibiotic sensitivity.

To maximize the chance of identifying the bacteremia through the use of blood cultures in a patient with IE, the veterinarian must pay attention to technique. More than one blood culture should always be taken as this increases the chance of success. Preferably the patient should not be on antibiotic therapy at the time of culture but positive blood cultures can still be obtained and so antibiotic therapy is not an absolute contraindication. It is often stated in the veterinary literature that presence or absence of fever at the time of culture makes no difference. However, in one study 88% of the febrile canine patients had a positive blood culture while only 50% of afebrile patients had a positive blood culture. Obtaining blood cultures from an afebrile patient at the time of echocardiographic diagnosis and then waiting seems prudent. One can then take further cultures if the patient develops a fever, if immediate antibiotic treatment is not required. Commercial vacuum bottles for collecting the blood samples, supplied by a laboratory, should be used. These bottles usually contain a tryptic digestion of protein. An additive such as the anticoagulant sodium polyanetholsulfonate increases the chances of isolating bacteria by anticoagulating the sample as well as interfering with complement and granulocyte activity. The venipuncture site must be prepared for aseptic collection by clipping the region over the vein and alternately wiping it with povidone iodine and alcohol. The area cannot be touched with an ungloved finger during venipuncture or the chances of culturing a skin contaminant increases greatly. As much blood as possible should be collected because the more blood that there is in the bottle, the more organisms that are present. This can be a problem in a cat or small dog. At least nine volumes of medium per 1 volume of blood should be used to dilute the complement and the granulocytes. The blood should be placed in 2 bottles after collection. One should be left unvented for anaerobic culture while the other must be vented to produce an aerobic environment. In patients with subacute endocarditis that are not in immediate danger, three separate venous blood cultures should be taken on the first day, at least 1 hour apart. If no growth is obtained by the second day, two more should be taken. If there still is no growth, and the diagnosis is still likely, two more cultures should be taken. If the patient has already been administered antibiotics, three more cultures can be taken over the following week. For patients with acute bacterial endocarditis that have evidence of sepsis or appear to have rapid destruction of valve tissue, three blood cultures should be taken over a 3-hour time span and empirical antibiotic therapy should be started. Cultures should be incubated for at least three weeks and Gram stains should be made at intervals, even if no growth is apparent.

Leukocytosis with a left shift is present in approximately 80% of dogs with IE. An increase in the absolute number of monocytes is seen in approximately 90% and is usually > 1500 cells/ml. Anemia is also common (60%) and is usually normocytic and normochromic. Severe azotemia occurs in about 10% of cases. If glomerulonephritis is present, proteinuria is present and the patient may be or may become hypoalbuminemic. Hypergammaglobulinemia can occur secondary to the chronic immune stimulation.

Ideally, antibiotic therapy is chosen based on culture of the offending organism from blood and on identifying an antibiotic to which the organism is sensitive. Every patient suspected of having IE should have its blood cultured before antibiotic therapy is instituted. The time that it takes to collect the samples for blood culture does not alter the course of the disease. Antibiotics selected to treat a patient with IE must be bactericidal. The bacteria are growing slowly within the vegetations and the vegetations prevent leukocytes from phagocytizing cells. Consequently, a bacteriostatic agent will not successfully sterilize the vegetation. Optimally, the serum concentrations of the antibiotics should be in the high end of therapeutic range for weeks. In humans, antibiotics are administered intravenously every 4-8 hours for 4-6 weeks. Presumably, this type of aggressive approach would also be preferred in dogs and cats. This approach, however, is often not economically feasible. Teaching an owner to administer antibiotics via intramuscular or subcutaneous injection may be an alternate approach. Usually, we try to hospitalize a patient for 1-2 weeks to administer antibiotics parenterally. If the disease appears to be controlled after that period the patient is usually discharged and the antibiotics are administered parenterally or orally at home.

Basing antibiotic therapy on blood culture results is often not feasible, either because the blood culture is negative or because the blood culture does not identify the organism early enough in a patient with acute bacterial endocarditis. A combination of ampicillin (a broad spectrum antibiotic) and nafcillin or oxacillin (for beta-lactamase-producing bacteria) and gentamicin or amikacin (for gram negative bacteria) should be administered to these patients. Special care must be taken with gentamicin or amikacin as both are nephrotoxic. Usually, therapy with these agents should be limited to 1-2 weeks, if possible. Patients with pre-existing renal disease or renal failure must be approached with extreme caution. Furosemide potentiates the renal toxicity of the aminoglycosides.

Fluoroquinolones (e.g., enrofloxacin) may be an alternative antibiotic class to the aminoglycosides for treating confirmed or suspected gram negative IE. They are bactericidal, concentrate well in heart valves and myocardium, and are effective against E. coli and Pseudomonas aeruginosa in experimental models of IE. At least seven fluoroquinolones have been studied in experimental models of IE. They have been found effective against staphylococcal IE, enterococcal IE and gram-negative IE. Three quinolones have been studied in experimental models of gram-negative bacillary IE. In all cases, the quinolones were equivalent to or superior to comparison drugs, including gentamicin and amikacin. In humans with prosthetic valve endocarditis, the fluoroquinolones may resolve bacteremia and so reduce fever and clinical signs. They, however, do not sterilize the vegetation. The only comparable situation in veterinary medicine would be endocarditis associated with a pacemaker lead.

The treatment for Bartonella species is aminoglycoside antibiotics. Ciprofloxacin may also be effective as it is reported to be more effective than aminoglycosides for B. henselae (cat-scratch fever).

Congestive heart failure is one of the most common complications of IE and is the most common cause of death. Aortic valve endocarditis often produces severe heart failure that is difficult to manage. Standard heart failure therapy with furosemide, an angiotensin converting enzyme inhibitor, and digoxin is warranted. In addition, a more potent arteriolar dilator, such as hydralazine, can be very beneficial in a patient that has acute, severe pulmonary edema or that is refractory to the other drugs. Hydralazine reduces peripheral vascular resistance and so reduces the amount of regurgitation in both aortic and mitral regurgitation. This results in decreased diastolic intracardiac pressures and a rapid reduction in pulmonary edema formation. Care must be taken not to produce profound hypotension when administering hydralazine along with an angiotensin converting enzyme inhibitor. Blood pressure monitoring is crucial to prevent this complication.

Corticosteroids are administered with some frequency to dogs with IE, presumably either to control the fever or to treat signs mistaken for an immune mediated disease. Corticosteroid administration exacerbates the clinical signs and worsens the prognosis in patients with IE. The patient may feel better for 24-48 hours but then usually deteriorates ("crashes and burns"). Patients that have occult IE will usually develop clinical signs rapidly following corticosteroid administration. For these reasons, corticosteroid administration is contraindicated in a patient with IE.

The prognosis for dogs and cats with active IE is poor. In one study, of 45 dogs proven to have IE either by fulfillment of clinical diagnostic criteria or by necropsy, only 20% survived. In a second study of 45 dogs, the survival rate was identical (20%). Dogs with large aortic valve lesions commonly develop severe left heart failure and die because of it. In one study, all dogs with aortic valve endocarditis died. Dogs with severe mitral valve endocarditis may follow a similar course but generally have a better prognosis, depending on the severity of regurgitation. No definitive treatment for valve destruction (i.e., prosthetic valve replacement) exists in veterinary medicine. Consequently, once severe regurgitation is produced heart failure and death are ultimately the expected result. Other patients die of embolic complications, especially renal infarction and failure. Other patients die of sepsis or infective complications.