Endocarditis predominantly affects the valves of the left side of the heart: a large autopsy series revealed mitral involvement in 86%, aortic 55%, tricuspid 20%, and pulmonary 1%. In the developing world rheumatic heart disease is the commonest predisposing factor, but in developed countries over 50% of patients have no known pre-existing cardiac lesion.

Clinical features

Presenting symptoms and signs include those of a bacteraemic illness, tissue destruction (heart valve(s) and adjacent structures), systemic or pulmonary embolism, and phenomena thought to be related to circulating immune complexes, e.g. splinter and conjunctival haemorrhages, Osler’s nodes, Janeway lesions, vasculitic rash, Roth spots, and nephritis. Right-sided infective endocarditis accounts for only 5% of cases overall, is usually associated with intravenous drug abuse or indwelling intravascular devices, and often causes septic pulmonary emboli that can lead to cavitating pulmonary infarcts.

Investigation and diagnosis—blood culture is the most important laboratory investigation in the diagnosis of endocarditis, with prolonged incubation requested in circumstances where endocarditis is strongly suspected. Serological tests can aid in the identification of organisms that are difficult to isolate. Echocardiography should be performed as soon as possible when endocarditis is suspected: its principal role is to detect vegetations, but it is not sufficiently sensitive to allow the clinician to exclude the diagnosis confidently on the basis of a negative result. Diagnosis is based on pathological criteria (demonstration of microorganisms by culture or histological examination, or histological evidence of active endocarditis) or—more usually—a combination of major and minor clinical criteria, with the major clinical criteria relating to (1) positive blood cultures of ‘typical’ or ‘consistent’ organisms, and (2) evidence of endocardial involvement detected on physical examination (new murmur) or by echocardiography.

Causes and management

The causes of endocarditis are viridans streptococci (up to 58%), Staphylococcus aureus (30% of community acquired and 46% of hospital acquired), Streptococcus bovis (up to 12%), enterococcus species (up to 10%), fungal (up to 10%), coagulase-negative staphylococci (5% of native valve endocarditis), the HACEK group of organisms (3%), and others.

Best management is provided by a multidisciplinary team involving cardiologists, microbiologists, infectious disease specialists, and cardiac surgeons. Bactericidal antibiotics are the mainstay of treatment. Recommended empirical therapy for the patient with suspected endocarditis presenting acutely is flucloxacillin (8–12 g/day IV in four to six divided doses) plus gentamicin (1 mg/kg body weight IV 8-hourly, modified according to renal function), and for the patient presenting with a more indolent course is penicillin (7.2 g/day IV in six divided doses) or ampicillin/amoxicillin (2 g IV 6-hourly) plus gentamicin (as above). This should be modified to a definitive antibiotic treatment regimen when the pathogen is known (see text for details). Surgery is required in about 30% of cases during the acute phase and in 20 to 40% of cases thereafter, with the main indications being haemodynamic instability, persistent infection, and annular or aortic abscesses.

Prevention

Until recently, antibiotic prophylaxis in ‘at risk’ patients—meaning any with a wide variety of cardiac lesions undergoing a wide variety of medical/surgical procedures—were accepted as reasonable, but there is no good evidence to support this practice. Recommendations from relevant UK, European and American professional bodies are now much more restrictive. UK (National Institute for Health and Clinical Excellence) guidelines state that antibiotic prophylaxis should only be given to patients at risk if they are undergoing a gastrointestinal or genitourinary procedure at a site where there is suspected infection. Most cardiologists feel that this is too restrictive and prefer European and American guidelines which recommend prophylaxis before dental and non-dental procedures for patients at high risk, including those with prosthetic cardiac valves or other prosthetic material within their hearts, previous infective endocarditis, and some forms of congenital heart disease.

When prophylaxis is recommended for dental and other procedures, regimen typically include amoxicillin (or clindamycin if penicillin-allergic), with the addition of gentamicin if risks are thought to be high, and substitution of vancomycin (or teicoplanin) for amoxicillin if the patient is penicillin-allergic (or has taken more than a single dose of penicillin in the previous month).

Endocarditis in great detail

Historical 

According to Laennec, one of the first descriptions of the cardiac pathological alteration of infective endocarditis came from Lazare Riviere, professor of Medicine at the University of Montpellier. When in 1646 he was consulted by a patient who complained of palpitations, Riviere detected the following symptoms and signs: first, a faint and irregular pulse. A month later, the patient was suffering from respiratory problems and swollen legs; the patient's condition progressively worsened; his difficulty in breathing increased and there was no pulse at the wrist; "when I placed my hand over his heart", Riviere detected a faint, rapid and irregular palpitation. After the patient died, Riviere performed an autopsy and in the left ventricle he found some "small round outgrowths resembling the lungs in texture, the largest of which was about the size of a hazelnut, which blocked the aortic valve". This description seems to indicate that Riviere had some knowledge of the internal structure of the heart and that what he observed in this case struck him as peculiar.

Read more:

Lazare Riviere

In 1769 Morgani made the link between infection (fulminating gonorrhoea) and ‘whitish polypus concretions on the upper part of the aortic valve near its borders’.

The clinical picture of endocarditis was first described by Jean Baptiste Bouillard, in 1835: ‘fever, an irregular pulse, cardiomegaly (by percussion) and a bellows murmur in the heart’. He gave the disease the name ‘endocarditis’, or an inflammation of the inner membrane of the heart and fibrous tissues of the valve, and was the first to use the term ‘vegetations’ for the valvular lesions.

Winge used the term ‘mycoses endocardi’ for the groups of microorganisms that he saw when he examined vegetations under the microscope in 1870. In 1886, Wyssecokowitch cultured Staphylococcus aureus from an endocardial vegetation. Lenthartz, in 1901, was the first to use blood cultures in the diagnosis of endocarditis. ‘Infective endocarditis’ was the term used by Thomas Horder, in 1901, to describe the syndrome consisting of (1) the presence of valvular disease, (2) the occurrence of systemic embolism, and (3) the discovery of microorganisms in the bloodstream.

Epidemiology

Infective endocarditis was universally fatal before the advent of antibiotic therapy. About 200 deaths are recorded each year in the United Kingdom, but this is almost certainly an underestimate. A recent review of papers published between 1993 and 2003 found the mean age of patients varied between 36 and 69 years, the median incidence being 3.6 per 100 000 population per year (range 0.3–22.4), increasing from 5 or less in individuals aged younger than 50 years to 15 or more in those older than 65 years. The median in-hospital mortality rate was 16% (range 11–26%). The incidence is greater in men, in those over 65 years of age and in those with prosthetic heart valves. In intravenous drug users, the incidence of infective endocarditis is estimated as 150 to 200 per 100 000 person years.

Pathogenesis

Normal vascular endothelium is resistant to microbial infection and very few patients potentially at risk actually develop infective endocarditis. Bacteraemia may occur spontaneously during chewing, tooth brushing, and other normal activities. Since low-grade bacteraemia occurs frequently in everyone, a defence mechanism must exist that can eradicate microbes adherent to vegetations. Platelets play a pivotal role in the antimicrobial host-defence mechanism and human platelets have been found to contain at least 10 different bactericidal proteins or ‘thrombocidins’.

Damage to the endothelial surface of the heart or blood vessels induces platelet and fibrin deposition producing a sterile thrombotic vegetation; infective endocarditis is initiated by the binding of microbes, discharged into the general circulation from a peripheral site, to these vegetations. These microbes become rapidly encased in further depositions of platelets and fibrin, and multiply.

The pathogenesis of infective endocarditis involves complex interactions between microbes and the host-defence mechanisms, both circulating and at the site of endothelial damage. An essential step is the activation of the clotting system and the formation of a fibrin clot on the endothelial surface. Experimental evidence suggests that the main pathogens in infective endocarditis (streptococci and staphylococci) can bind to endothelial cells and induce functional changes within these cells causing monocyte adhesion. The endothelial cells respond to local inflammation by expressing β1-integrins which promote the adhesion of pathogens that carry fibronectin-binding proteins on their surface. The combination of damaged endothelial cells, bacteria, and endothelial bound monocytes results in the induction of tissue-factor-dependent procoagulant activity which initiates clot formation. Polymorphonuclear leucocytes which are recruited to the infected endothelial site may be subsequently involved in the disease progression, with the contents of lysosomes released by the activated leucocytes probably causing softening and separation of valve tissue, leading to its destruction.

In endocarditis, the vegetations are found predominately on the left side of the heart (85%). In a large autopsy series of more than 1000 cases reported over 50 years ago, the mitral valve was involved in 86%, the aortic in 55%, the tricuspid in 20%, and the pulmonary valve in only 1%. The predominance of left-sided lesions has led to the belief that the higher pressures and velocities encountered in the left side of the heart and the proximal aorta must impose a greater mechanical stress on the valves and endocardium, which in turn leads to local damage.

Endocarditis is classically associated with ‘jet lesions’, where blood flowing from a high-pressure area through an orifice to an area of lower pressure produces a high-velocity jet. Vegetations are usually found in the lower-pressure area, e.g. on the atrial surface of the mitral valve in mitral regurgitation, or the ventricular surface of the aortic valve in aortic regurgitation. This particular deposition of vegetations has been explained on the basis of the Venturi effect.

Once a vegetation is established, it determines the subsequent clinical picture by four basic processes: bacteraemia, local tissue destruction, embolization, and the formation of circulating immune complexes.

Clinical features

Early reports of infective endocarditis described a low-grade, febrile illness caused by viridans streptococci from the mouth in a patient with chronic rheumatic heart disease. Night sweats, anorexia, and weight loss were followed by the development of splinter haemorrhages and Osler nodes, finger clubbing, and splenomegaly. The infection progressed relentlessly with increasing cachexia, and the patient died from cardiac failure or a major embolic episode. The term ‘subacute bacterial endocarditis’ was used to describe this illness. ‘Acute or malignant endocarditis’ described an aggressive form of the disease, usually caused by Staph. aureus or other virulent bacteria.

During the past 50 years, there has been a striking change in the pattern of endocarditis. The proportion of patients in developed countries with endocarditis who have no known pre-existing cardiac lesion has risen to over 50%. This change is related both to the decline in rheumatic heart disease and to the increase in extracardiac predisposing factors, including intravenous drug abuse, haemodialysis, and the use of intravascular devices. Prosthetic heart valves are an important predisposing factor and cardiac surgery for complex congenital lesions has increased the lifespan of patients who would previously have died prematurely. Antibiotic-resistant organisms have emerged. The longevity of the populations of developed countries has resulted in an increasing age of patients with infective endocarditis, with mean age rising from under 40 years, before 1940, to 60 to 70 years today.

Features of a bacteraemic illness

Discharge of the infecting agent into the circulation produces constant bacteraemia which may present as pyrexia, rigors, malaise, anorexia, headache, confusion, arthralgia, and anaemia. Some cases of endocarditis may present without fever, particularly in older people.

Features of tissue destruction

Endocarditis initially affects valve cusps, leaflets, or chordae tendineae. Tissue destruction results in valvular incompetence, cusp perforation, or rupture of the chordae, producing an appropriate cardiac murmur that may change in character during the course of the illness: 80% of patients present with a murmur, and 15 to 20% develop one during their hospital stay. Large vegetations rarely obstruct a native valve, but mechanical obstruction of prosthetic valves is more common and clinically more difficult to detect.

As the infective process progresses, it may extend beyond the valve into the paravalvular structures. Aortic root abscess is a serious complication: extension through the aortic wall into other tissues or cavities can create a fistula or pseudoaneurysm. Particular problems can include sinus of Valsalva aneurysm and involvement of the coronary ostia, and septal abscesses can lead to progressive conduction defects evidenced by prolongation of the PR interval on the ECG and eventually complete heart block.

Paravalvular abscess is more common in native aortic valve endocarditis than in mitral valve infection. Infection of a mechanical valve involves the sewing ring and may lead to valve dehiscence. In the case of a mechanical aortic valve, where infection is often localized to the junction between the sewing ring and the aortic annulus, a large false aneurysm may develop in this area. Free wall myocardial abscesses may rupture and cause sudden death.

Features of systemic or pulmonary emboli

Fragments of an infected vegetation may be dislodged into the systemic or pulmonary circulation, producing emboli in 20 to 40% of cases (up to 50% reported in autopsy series). These may lodge in any part of the circulation and present as a cerebrovascular accident, arterial occlusion of a limb, myocardial infarction, sudden unilateral blindness, or infarction of the spleen or a kidney. Septic embolism from the left side of the heart may result in the formation of a cerebral abscess. In right heart endocarditis, recurrent septic pulmonary emboli may be misinterpreted as ‘pneumonia’.

Mycotic aneurysms arise from embolism of the vasa vasorum that weakens the arterial wall: these have been reported in almost 3% of clinical cases but are found in up to 15% at autopsy. In the cerebral circulation, such aneurysm may produce subarachnoid haemorrhage or intracerebral haemorrhage. The popliteal artery is a common site for mycotic aneurysms.

Emboli are characteristic of Staph. aureus infections and large emboli are a feature in HACEK (see below) and fungal endocarditis. They usually occur before or within the first few days after starting antimicrobial therapy. Anterior mitral valve-leaflet vegetations are more likely to embolize than aortic valve vegetations, especially if they are highly mobile. Vegetation size does not predict systemic embolization, but large vegetations (>10 mm) are associated with a poor outcome overall.

After an embolic complication, recurrent episodes are likely to follow, especially if vegetations persist on echocardiography. In more than 50% of cases, such recurrence occurs within 30 days of the first episode. This is not reduced by treatment with anticoagulants such as warfarin or antiplatelet therapy such as aspirin: both may increase the risk of bleeding and should be avoided unless they are essential.

Features of circulating immune complexes

The infected vegetation acts as an antigen that triggers an immune response. Chronic antigenaemia stimulates generalized hypergammagloblinaemia such that after several weeks of infection a variety of autoantibodies can be detected. Immune complex deposition probably causes many of the extracardiac manifestations of infective endocarditis, but these classical signs are relatively uncommon and are often absent in individual patients.

• ◆ Splinter haemorrhages—these are found in the nail bed of the fingers and less commonly the toes, and are linear in form. They occur in 5 to 15% of cases
• ◆ Conjunctival haemorrhages
• ◆ Osler’s nodes—these transient painful erythematous nodules are found at the ends of fingers and toes and the thenar and hypothenar eminences. They occur in 5 to 10% of cases, and may be due to minute infected emboli rather than immune complex deposition
• ◆ Janeway lesions—irregular painless erythematous macules found in roughly the same distribution as Osler’s nodes; they tend to blanch with pressure
• ◆ Vasculitic rash—immunoglobulin and complement deposits are found in the walls of skin capillaries. Vasculitis may account for some of the neurological findings in infective endocarditis
• ◆ Roth spots—boat-shaped haemorrhages in the retina are often called Roth spots, but true Roth spots are white retinal exudates that may be surrounded by haemorrhage. They consist of perivascular collections of lymphocytes and occur in 5 to 10% of cases.
   
• ◆ Splenomegaly—clinical splenomegaly is less common than was reported in earlier literature. CT scanning of the abdomen shows the spleen to be enlarged in at least 50% of cases and often demonstrates splenic infarcts. Splenic abscesses sometimes occur and splenic rupture can be fatal
• ◆ Nephritis—immune complexes can cause glomerulonephritis, manifest as proteinuria, haematuria, and decline in renal function, with immunoglobulin and complement deposition within glomeruli on renal biopsy. Key investigations are simply dipstick testing of the urine (with microscopy if more than 1+ positive for blood and/or protein) and measurement of serum creatinine
• ◆ Arthralgia—the joint manifestations of infective endocarditis may result from immune complex deposition in the synovial membrane 

Other features

Up to 30% of patients with endocarditis present with neurological symptoms: these are most common in staphylococcal infection, in which one-third present with the clinical features of meningitis. Headaches, confusion, and toxic psychosis can be present as well as encephalomyelitis. It is not certain whether some of these neurological manifestations result from repeated small emboli or from a vasculitic process within the cerebral circulation as a consequence of immune complex deposition. The cerebrospinal fluid can show an increase in white cells, but is usually sterile on culture, although very occasionally it may be positive for staphylococcal infection.

Although immune-mediated glomerulonephritis has been regarded as the typical renal lesion of infective endocarditis, this assumption was based on small series predating modern treatment regimens. More recent work indicates that the commonest renal histological finding is infarction, usually septic. Circulatory compromise can rarely cause severe renal impairment as a result of renal cortical necrosis.

Finger clubbing is one of the classical features of infective endocarditis, usually seen after 1 or 2 months of the illness. It is seldom seen now, but when present is still a useful sign because it rarely occurs in conditions with which infective endocarditis can be confused.

Specific types or circumstances of endocarditis

Prosthetic valve endocarditis

Patients with prosthetic heart valves have a small, but constant, risk of infective endocarditis, estimated at 0.2 to 1.4 events per 100 patient years. The incidence of prosthetic valve endocarditis is about 3% in the first postoperative year, with the highest risk during the first 3 months. Prosthetic valve endocarditis is five times more common in the aortic area than the mitral area and may involve mechanical, xenograft, and homograft valves.

Prosthetic valve endocarditis has been classified as early or late according to its temporal relationship to surgery. Early prosthetic valve endocarditis usually occurs within 60 days of open heart surgery and accounts for 30% of cases. It is caused either by contamination of the prosthetic valve at implantation or by perioperative bacteraemia from intravenous catheters, arterial lines, urethral catheters, or endotracheal tubes. The commonest organisms are coagulase-negative staphylococci.

Late prosthetic valve endocarditis accounts for 70% of cases and usually occurs 60 days or more after surgery. The pathogens are those seen in native valve endocarditis, with a preponderance of viridans streptococci and staphylococci, but with a higher incidence of other organisms. Some patients with late prosthetic valve endocarditis will have acquired the infection at the time of surgery, but a bacteraemia is usually the principal cause.

Bacteraemia in a patient with a prosthetic valve must always be taken seriously, but it may not always be the result of endocarditis. The clinical picture of prosthetic valve endocarditis is typically fever, malaise, and weakness, with the more classical signs usually absent. The condition is often insidious and difficult to diagnose clinically. A new murmur may appear, and heart failure and embolic phenomena cause high mortality (20–50%). Infection in a mechanical valve is located in the sewing ring, from which the infection can spread into the host tissues producing annular/myocardial abscesses, paravalvular leak, and prosthetic dehiscence as described above. Infection of a tissue valve usually involves the valve leaflets, resulting in destruction or perforation and valvular incompetence. Vegetations may cause obstruction with all forms of prosthetic valve.

The diagnosis of prosthetic valve endocarditis requires a high index of clinical suspicion, blood cultures, and transoesophageal echocardiography, which is far superior to the transthoracic approach for finding vegetations and identifying periprosthetic spread of infection. Vegetations are more difficult to identify in patients with mechanical valves than those with bioprostheses.

Right-sided endocarditis

Right-sided infective endocarditis accounts for only 5% of cases overall, but centres that treat large numbers of intravenous drug users will have a higher incidence. The clinical picture differs significantly from left-sided disease. It is usually associated with intravenous drug addiction or indwelling intravascular devices, including pacemakers, central venous lines of all types, and septal occluder devices. Staph. aureus is the commonest pathogen and the tricuspid valve is more commonly affected (80%) than the pulmonary. Fever is almost always present and a cardiac murmur is found in 80% of cases. There may be septic pulmonary emboli and the resultant pulmonary infarcts may cavitate. Symptoms include cough, haemoptysis, and pleuritic chest pain; a chest radiograph shows pulmonary infiltrates often misdiagnosed as ‘patches of pneumonia’. Renal involvement has been described in over one-half of cases, most commonly abscess formation or diffuse pyelonephritis. Myocarditis is more common in right-sided involvement than left. Peripheral stigmata, splenomegaly, and central nervous system involvement are rare (no more than 5% of cases). Death is most commonly due to sepsis, rarely to heart failure.

Endocarditis in intravenous drug users

Endocarditis is a serious complication of intravenous drug abuse. The right side of the heart is affected most commonly, but the left may also be involved in a substantial number of patients (37%), and both right and left side in a few (7%). On the left side, mitral and aortic valves are equally infected. A history of previous heart disease is only found in some 25% of cases.

Staph. aureus is responsible for 40% of all cases. Gram-negative bacilli are next most frequent, Pseudomonas aeruginosa and Serratia marcescens accounting for most of these. Candida can cause endocarditis in intravenous drug users, and polymicrobial endocarditis accounts for 5% of cases.

The skin is the commonest site from which pathogens enter the bloodstream via needles. Gram-negative bacilli are rarely recovered from needles or the drug itself, and it has been suggested that these organisms come from tap water, sinks, or lavatory pans.

The clinical picture of intravenous drug use-associated endocarditis depends on which side of the heart is affected. Right-sided disease is described above; left-sided disease behaves like that seen in nondrug cases, with a high incidence of heart failure, arterial embolism, central nervous system involvement, and peripheral stigmata.

The overall mortality depends on when the patient presents: it is high if they present late, reflecting among other things the difficulty in dealing with addicts because of their poor compliance and reluctance to discontinue their drug habit. The principles of management are similar to those in patients who are not drug abusers. The duration of intravenous antibiotics should be at least 4 weeks, but this is frequently impossible to do in practice, and in right-sided endocarditis simple removal of the valve without replacement appears to be the best strategy.

Endocarditis in children

Endocarditis does occur in children but is rare, especially in the first decade of life. In the older literature, tetralogy of Fallot was the commonest cardiac problem associated with infective endocarditis. Complex cyanotic disease, congenital heart disease corrected with prosthetic material, and small ventricular septal defects make up the bulk of cases now.

Diagnosis of infective endocarditis

Laboratory methods

This is the most important laboratory investigation in the diagnosis of endocarditis. See table 1 below:

Isolation of the pathogen enables an effective antibiotic treatment regimen to be devised. Optimal technique is necessary to avoid false-positive cases due to contaminating organisms from the skin. Each set of blood cultures should be taken from a separate venepuncture and at least 10 ml of blood be injected into each culture bottle. Blood cultures should be taken before antibiotics are given; if they have already been given, cultures should still be done and, if possible, the giving of further antibiotics delayed for a few days. However, previous antibiotics may render the blood sterile for some time, and the chances of recovering the pathogen, particularly when it is a viridans streptococcus, are very low. Much mystique has been attached to the number and timing of blood cultures in cases of suspected endocarditis. What is known is that the bacteraemia is usually constant, and that whenever the blood is obtained for culture, and however many sets are taken, then in most cases all bottles will grow the pathogen. There are, of course, rare exceptions when only a few bottles taken are positive, and this is one reason why it is conventional to take two or three sets. Another reason for several cultures is to assess the relevance of the common skin contaminants, particularly the coagulase-negative staphylococci but also corynebacterium, that can cause endocarditis.

In infective endocarditis, an elevated ESR and C-reactive protein are almost invariable, and these inflammatory markers are used most commonly to monitor the activity of the disease. A normochromic normocytic anaemia is often present and a polymorphonuclear leucocytosis is found in most cases. Hypergammaglobulinaemia and a low serum complement may be present, together with a false-positive rheumatoid factor. Circulating immune complexes may be detected.

Serological tests aid in the identification of organisms that are difficult to isolate, including bartonella, coxiella (Q fever), chlamydia, mycoplasma, legionella, brucella, and fungi. Candida antibodies are of no diagnostic value.

Echocardiography

In suspected cases of endocarditis, echocardiography should be performed as soon as possible and interpreted by an experienced cardiologist. Its principal role is to detect vegetations, but it is not sufficiently sensitive to allow the clinician to exclude the diagnosis confidently on the basis of a negative result. The sensitivity depends on the size of the vegetations and the time course of the disease: it can resolve vegetations as small as 1 to 2 mm, but it is more difficult with prosthetic than native valves and more difficult with mechanical than biological prostheses.

Vegetations appear as thick, ragged, nonuniform echoes oscillating on or around a cardiac valve or in the path of a regurgitant jet. They do not usually restrict leaflet mobility and they exhibit valve-dependent motion. On native valves, vegetations are usually attached to the ventricular side of the aortic valve and the atrial side of the mitral and tricuspid valves.

Two-dimensional echocardiography should be employed initially in all cases of suspected endocarditis. Transoesophageal echocardiography has improved the rate of diagnosis of infective endocarditis over that of transthoracic echocardiography, particularly in the presence of a prosthetic valve. It has also made it easier to recognize many complications of prosthetic valve endocarditis, such as abscesses, fistulas, and paravalvular leak. In addition to vegetations, echocardiography may demonstrate indirect signs of valvular integrity, such as excessive systolic expansion of the left atrium in mitral incompetence or fluttering of the anterior leaflet of the mitral valve in aortic incompetence. Ventricular size and contractility are both easily assessed.

The diagnosis of right-sided endocarditis has been greatly facilitated by echocardiography, particularly transoesophageal echocardiography. Vegetations tend to be larger on the right side and can be demonstrated in 80 to 100% of cases.

Vegetations need to be differentiated from other conditions which produce echo-density on cardiac valves, including calcification, myxomatous degeneration, and atrial myxoma.

Examination of the heart valve and other tissues

Histology

Histology remains the gold standard for explanted valves. When valve replacement is undertaken, valvular tissue (including vegetation) should be examined histologically and cultured for the presence of microorganisms, which may allow postoperative antibiotics to be tailored accordingly. However, the isolation of microorganisms by valvular culture in infective endocarditis is infrequent: only 15% in one large series, with staphylococci being most common. Fastidious and rare microorganisms have been demonstrated on heart valves by various staining techniques.

Nucleic acid-based techniques

There are now several papers in the literature describing the application of polymerase chain reaction techniques to samples obtained from heart valves, vegetations, and embolic tissue in patients with suspected endocarditis. The intention is to allow identification of the infecting microorganism when blood cultures are negative due to prior antibiotic therapy, or the causative organism is fastidious or cannot be cultured. This is not yet routine practice because of the following issues. Firstly, bacterial DNA is present within heart valves for many months and possibly years following successful treatment. Secondly, contamination of samples with any bacterial DNA leads to false-positive results. Thirdly, false-negative results can occur due to polymerase chain reaction inhibitory factors present within blood and other bodily fluid.

Criteria for the diagnosis of infective endocarditis

In 1994, Durack and his colleagues introduced criteria for the diagnosis of infective endocarditis that have been accepted as the ‘Duke criteria’, which categorize patients into definite, possible, and rejected groups. Although these criteria have been shown to be superior to previous diagnostic tools, they have limitations: in particular, there is a possibility of misclassification when blood cultures remain negative or echocardiography is inconclusive. Negative blood cultures occur in 5 to 31% of cases of infective endocarditis, commonly due to prior antibiotic therapy, also because of fastidious and atypical microorganisms. Transthoracic echocardiography visualizes vegetations in only about 50% of cases: transoesophageal echocardiography has a higher sensitivity for detecting signs of endocarditis on both native and prosthetic valves, but will only be diagnostic in 50 to 94% of cases. These issues mean that the number of patients who may be incorrectly diagnosed as having possible infective endocarditis, as opposed to definite, could be as high as 24%.

Modification of the Duke criteria to increase their sensitivity has been suggested by several authors (see table below). Positive serology for typical microorganisms and the use of polymerase chain reaction techniques have been suggested as major criteria, and the following additional minor criteria have been proposed—the presence of newly diagnosed clubbing, splenomegaly, splinter haemorrhages and petechiae, microscopic haematuria, a high erythrocyte sedimentation rate or a high CRP, also the presence of central nonfeeding lines and peripheral lines. See table 2 below:

Microbiology

Although almost any microorganism can cause infective endocarditis, particularly when this involves a prosthetic valve, certain species do so much more commonly than others. The predominant species involved in the infection have not changed significantly in their incidence in the past three decades. Overall, viridans streptococci and staphylococci account for about two-thirds of cases. However, endocarditis cannot be considered as a microbiologically homogeneous entity as the incidence of any specific organism depends (1) on the patient, whether an intravenous drug user or not; (2) on the valve, whether native or prosthetic—and if native, whether previously abnormal or not, and if prosthetic whether mechanical or a bioprosthesis, and whether the infection was acquired early or late; and (3) where (and how) the infection was acquired, whether in the community or (as increasingly these days) in hospital, usually via an infected intravascular device.

The more common species encountered will be considered individually.

Streptococci

The genus Streptococcus includes species of differing virulence and pathogenicity as well as differing normal habitat in humans. The genus has undergone numerous taxonomic revisions over the past decade or more, and the previous dependence on haemolytic activity on blood agar and serological reactions has been superseded, in many cases, by molecular and chemotaxonomic approaches. Examples of such taxonomic change include the assignment of the faecal streptococci to the genus Enterococcus, and of Streptococcus morbillorum to Gemella morbillorum, and of the nutritionally dependent streptococci previously known as Streptococcus adjacens and Streptococcus defectivus to the genus Abiotrophia. There are many other examples, but taxonomic change is of limited interest to clinicians and has no bearing on the management of infection.

Viridans streptococci

For many years, it has been conventional to refer to a group of streptococci that produce greening (α-haemolysis) on blood agar as viridans streptococci; indeed, many still refer (inaccurately) to a microbe ‘Streptococcus viridans’. Although most of these streptococci are virtually specific to the normal oropharyngeal flora and are rarely encountered at other sites, some are not found in the oropharynx at all, e.g. Strep. bovis, and others are found at many sites including the oropharynx, e.g. the milleri group of streptococci. The viridans streptococci are the commonest cause of community-acquired native valve endocarditis and community-acquired late-onset prosthetic endocarditis. The commonest species of the viridans streptococci specific to the oropharynx are Strep. sanguis, Strep. oralis, and Strep. mutans, but there are others. Dextran formation may be a virulence factor in these streptococci. Contrary to popular belief, they do not require a dental extraction to enter the bloodstream and cause frequent bacteraemias after chewing, tooth brushing, etc. They are organisms of low virulence and thus usually only infect previously abnormal heart valves. Whereas Strep. oralis and Strep. sanguis are occasionally isolated from blood cultures of patients who do not have endocarditis, the isolation of Strep. mutans from the blood is virtually synonymous with endocarditis.

Streptococcus bovis

This streptococcus, which may appear ‘viridans’ on blood agar, is part of the normal intestinal flora, but may initially be mistaken for an oral streptococcus. In common with the enterococci, it bears the Lancefield group D antigen and thus can also be mistaken for Enterococcus faecalis, though it is sensitive to penicillin whereas the latter is resistant. There is a significant association between Strep. bovis bacteraemia (and hence endocarditis) and colonic pathology, and any patient with Strep. bovis endocarditis thus warrants appropriate investigation. Strep. bovis endocarditis is much less common than that caused by oral streptococci.

Pyogenic streptococci

These organisms, often referred to as β-haemolytic streptococci, cause endocarditis less frequently than the viridans streptococci, but are more aggressive microbes and likely to affect (and often rapidly destroy) a previously normal valve. The commonest pyogenic streptococcus to cause endocarditis is the Lancefield group B β-haemolytic streptococcus (GBS), sometimes referred to as Strep. agalactiae. This organism is found as normal flora in the genital and gastrointestinal tracts. As with Staph. aureus, any patient with community acquired group B β-haemolytic streptococcus bacteraemia should be assumed to have infection in bone, joint, or on a heart valve until proved otherwise. Groups C and G β-haemolytic streptococci occasionally cause endocarditis, and group A even more rarely.

The milleri group of streptococci are best regarded as pyogenic streptococci: they form part of the normal flora of all mucous membranes and occasionally cause endocarditis, though much more often cause abscesses at many different sites. The milleri group consists of three species, Strep. constellatus, Strep. intermedius, and Strep. anginosus. Interestingly, these streptococci can bear the Lancefield antigens A, C, G, or F (or none); all group F streptococci are milleri, but not all milleri are group F.

Streptococcus pneumoniae (pneumococcus)

Pneumococcal endocarditis accounted for about 10% of cases of endocarditis in the preantibiotic era, but is now rarely seen, although it is sometimes diagnosed at autopsy of patients with fatal pneumococcal infection. The pneumococcus is a virulent pathogen and attacks normal heart valves. Patients with endocarditis generally have pneumonia and sometimes meningitis.

Enterococci

Enterococci form part of the normal gastrointestinal flora. They are more virulent than viridans streptococci and more resistant to antibiotics. The past decade has seen an increase in enterococcal endocarditis, particularly in older people, but this infection is still much less common than that caused by viridans streptococci. Whilst there are many species of enterococci, those causing endocarditis are usually E. faecalis and occasionally E. faecium. Most cases are community acquired, but the infection can sometimes be acquired in hospital as a result of urological instrumentation. Any patient admitted from the community with E. faecalis in the blood should be investigated for endocarditis.

Staphylococci

Staphylococci now account for about one-third of cases of community-acquired endocarditis and are the commonest cause of hospital-acquired endocarditis. Most of these staphylococci are Staph. aureus, but an increasing proportion are now coagulase-negative staphylococci. All staphylococci are skin organisms and patients become infected from their own skin flora, or in the case of methicillin-resistant Staph. aureus (MRSA) from that of others by cross-infection.

Staph. aureus is an important and aggressive pathogen in community-acquired native valve endocarditis. Sometimes a trivial skin lesion can be identified as the source of the organism, but there is often no obvious lesion. Staph. aureus, and increasingly now MRSA, is the commonest cause of hospital-acquired endocarditis. Prosthetic valves can become infected with Staph. aureus, both early as result of sternal wound sepsis and late as with native valves. Staph. aureus is the commonest pathogen causing endocarditis in intravenous drug users.

Although still regarded by many as pathogens of prosthetic rather than native valves, coagulase-negative staphylococci also cause native valve infection. This has become more common, or certainly more commonly recognized, in the last two decades. The infecting species is most often Staph. epidermidis, but in many reports the designation ‘Staph. epidermidis’ tends to be used for any unspeciated coagulase-negative staphylococcus. Many other species have been reported in native valve endocarditis, including Staph. lugdunensis, Staph. simulans, Staph. warneri, Staph. capitis, Staph. caprae, and Staph. sciuri. As in community-acquired Staph. aureus endocarditis, there is sometimes a presumptive predisposing skin lesion. Most patients have a pre-existing cardiac abnormality. Many of these staphylococci can be as virulent as Staph. aureus and share some of the same virulence factors.

Other organisms

A wide variety of organisms account for the few cases of endocarditis that are not caused by streptococci, staphylococci, or enterococci: only a few warrant a specific mention here.

HACEK group

These are fastidious, slow-growing species that are oropharyngeal commensals and have a predilection for heart valves such that their presence in blood cultures is virtually synonymous with this infection. The group consists of Haemophilus aphrophilus/paraphrophilus, Actinobacillus actinomycetemcomitans, Cardiobacterium hominis, Eikenella corrodens, and Kingella kingae. A. actinomycetmecomitans, in particular, seems more likely to infect prosthetic than native valves. The large vegetations thought to be characteristic of HACEK organisms in native valve infection may be the result of diagnostic delay and prolonged illness rather than any inherent property of the microbes.

Organisms that cannot be cultured by routine techniques

Endocarditis is a rare (and late) sequel of acute Coxiella burnetii (Q fever) infection, mostly in middle-aged men with pre-existing valve disease. The reservoir of the organism is usually sheep or cattle, but the source and mode of transmission in many cases is unknown. The diagnosis is usually made serologically, although C. burnetii can be recovered from the blood and excised valves by special techniques. The disease is almost certainly underdiagnosed, with some cases labelled ‘culture negative’ endocarditis.

Bartonella quintana endocarditis was first recognized in 1995 in homeless, alcoholic patients; B.henselae infection may be associated with cat or cat-flea contact, and other species of bartonella have also been described as causing endocarditis. Bartonella infection is usually diagnosed by serology, although these bacteria can be recovered from the blood and excised valves by special culture techniques and their presence detected by polymerase chain reaction. False-positive serology for chlamydia has been reported with bartonella infections, but chlamydia species, particularly Chlamydia psittaci, can also cause endocarditis (very rarely). It is possible that some cases attributed to chlamydia in the past on the basis of serology may have been caused by bartonella.

Fungi

Fungal endocarditis is very rare and more likely to occur on prosthetic than native valves, except in intravenous drug users. Most infections are acquired in hospital, with infection at intravascular access sites and broad-spectrum antibiotics predisposing to candida infections. Candida species, usually Candida albicans, are the commonest fungi, but Aspergillus spp. and more exotic genera have also been reported. Blood cultures are only likely to be positive with candida, and often only intermittently; for other fungi, the diagnosis must be made by serology and culture of the fungus from the excised valve or detection on valve histology.

Blood culture negative endocarditis

The possibility that an illness is not due to endocarditis should always be entertained when blood cultures are repeated negative. However, in 5 to 31% of definite cases of endocarditis the blood cultures will be negative. The commonest explanation for this is previous administration of antibiotics. In a few cases the pathogen will be recovered from another site, including the excised valve, excised emboli, or—specifically in right-sided endocarditis—respiratory specimens. Other causes of negative blood cultures are infection with organisms that cannot be grown by conventional blood culture methods, and infections that are diagnosed by serology such as C. burnetii, bartonella, and chlamydia.

Treatment

Initial therapy

The treatment of infective endocarditis ideally should be undertaken by a multidisciplinary team involving cardiologists, microbiologists, infectious disease specialists, and cardiac surgeons. Where possible, patients should be treated in cardiac centres that undertake cardiac surgery. Bactericidal antibiotics are the mainstay of treatment. The choice and duration of treatment depend upon the type of microorganism and its susceptibility profile, whether infection involves a native or prosthetic valve, and whether the patient is allergic to any antimicrobials.

In those patients who have been ill for many weeks, antibiotic treatment can be deferred until the blood cultures are positive and the pathogen known. Antibiotic treatment should be started immediately after taking blood cultures in patients who are acutely ill, using a broad-spectrum combination that can be adjusted when the pathogen is known. However, endocarditis is often not suspected initially in many patients who are acutely ill with native valve infection: there may be no obvious signs of this and antibiotics are started for ‘septicaemia’. When methicillin-resistant staphylococci (whether Staph. aureus or coagulase-negative staphylococci) are likely pathogens, vancomycin or teicoplanin are an essential component of any combination. If empirical therapy is indicated the choice of antimicrobial agent should be dictated by the type of presentation, whether or not there is an intracardiac prosthesis in place, and on the expected causative organism as indicated from the clinical picture. See table 3 below:

Definitive therapy

There are various national guidelines for the treatment of specific organisms. It is important to realize that these are consensus based because there are no randomized controlled trials to show the efficacy of any particular regimen. It is conventional to estimate the minimum inhibitory concentration (MIC) of the antibiotic for the pathogen, but in practice routine disc sensitivity tests are quite satisfactory in many cases. Although it is widely believed that prosthetic endocarditis requires a longer duration of antibiotic treatment than native valve infection, there are few data to support this.

Recommendations for the treatment of the commonest causative organisms are taken from guidelines published by the British Society for Antimicrobial Chemotherapy. See tables 4 and 5 below:

HACEK endocarditis

If amoxicillin-sensitive, 2 g amoxicillin/ampicillin should be administered intravenously 4 to 6-hourly, plus gentamicin 1 mg/kg body weight according to renal function (for the first 2 weeks only) and with regular monitoring of drug levels. If amoxicillin-resistant, ceftriaxone 1 to 2 g (maximum 4 g) should be administered intravenously once daily, plus gentamicin as above.

Other uncommon causes of endocarditis

Treatments for uncommon culture-negative causes of endocarditis are shown in the table 6 below:

Fungal endocarditis

For candida, amphotericin B 1 mg/kg per day and flucytosine 100 mg/kg should be administered in four divided doses according to renal function (first choice), or fluconazole 400 mg 12-hourly orally (second choice), or caspofungin 70 mg as a loading dose, followed by 50 mg once daily (70 mg per day if weight >80 kg) (first choice if intolerance or resistance precludes other options).

For aspergillus, voriconazole 6 mg/kg 12-hourly for two doses (loading) should be administered, then 4 mg/kg 12-hourly intravenously, or 400 mg 12-hourly for 24 h followed by 200 mg 12-hourly orally, or amphotericin B 1 mg/kg per day according to renal function.

Monitoring of treatment

Serum bactericidal titres against the infecting organism are no longer recommended. There was always great variation in the monitoring methods used for these tests and in the interpretation of their results. At best, they could only predict bacteriological not clinical cure, and bacteriological failure is very rare. The most useful laboratory test for monitoring the response to treatment (which is usually obvious clinically) is serial estimation of C-reactive protein; this is of much more use than the ESR, which is much slower to fall.

Relapse of endocarditis (if it occurs) usually occurs within 2 months of cessation of treatment. The relapse rate is lowest for patients with native valve endocarditis caused by penicillin-sensitive viridans streptococci. Relapse rate in prosthetic valve endocarditis is between 10 and 15%.

Prevention and prophylaxis

Antibiotic prophylaxis in ‘at risk’ patients is generally accepted as reasonable. This is largely based on indirect data from in vitro studies, experimental animal models, and studies of clinical bacteraemia, but there are many uncertainties about its value, and data confirming its clinical effectiveness are lacking. However, many authorities continue to recommend antibiotic prophylaxis to cover certain procedures associated with predictable and significant bacteraemia in patients known to be at high risk, but accept that this might fail, even with the recommended regimens, and that significant adverse reactions to antibiotics are important, even if relatively uncommon.

The most controversial area for the use of prophylactic antibiotics concerns dental treatment. A working party of the British Society of Antimicrobial Chemotherapy first recommended that the practice of giving all patients with cardiac abnormalities antibiotics before dental treatment should be stopped, except for those who have a history of healed infective endocarditis, prosthetic heart valves, or surgically constructed conduits. Many other groups vigorously opposed this recommendation, not least because some cases of endocarditis that involve dental procedures have resulted in litigation, and in most of these legal cases, endocarditis was judged to be caused by dental manipulations on the basis of the dental procedure, cardiac pathology, infecting microorganism, and the temporal link between the onset of endocardial infection and the dental manipulation.

In 2007 the American Heart Association revised its guidelines limiting the use of antibiotic prophylaxis to the highest risk patients who were undergoing the highest risk procedures; in the case of dental treatment these were manipulation of gingival tissue, the periapical region of teeth or perforation of the oral mucosa. See table below:

The National Institute for Health and Clinical Excellence (NICE) has developed guidelines for adoption by the National Health Service in England, Wales and Northern Ireland. Based on its findings that (1) there is no consistent association between having an interventional procedure and infective endocarditis, (2) that the clinical effectiveness of antibiotic prophylaxis is not proven, (3) that the risk of antibiotic associated adverse effects exceeds the benefits, and (4) that prophylaxis is not cost effective, NICE concluded that antibiotic prophylaxis should not be given to any at risk patients undergoing an interventional procedure. NICE made one exception; namely in patients undergoing a gastrointestinal or genitourinary procedure where there is suspected pre-existing infection, who should receive an antibiotic that covers endocarditis causative organisms.

Most recently the ACC/AHA Task Force on Practice Guidelines has downgraded from Class 1 (mandatory) to Class 2 (reasonable practice) the recommendation for antibiotic prophylaxis for high risk patients. Not surprisingly these departures from established practice have met with mixed reactions; the dental profession in the UK has welcomed the NICE proposals, but many British cardiologists and cardiovascular surgeons have opposed them. A sensible approach would appear to be to allow individual doctors to do what they feel is best for their patients and to be encouraged to discuss their reasons for taking a particular stance on antibiotic prophylaxis with those patients. Patients themselves should be taught the importance of good oral hygiene and to recognize symptoms that might indicate infective endocarditis and when to seek expert help. Suitable prophylactic antibiotic regimens are described in the table below:

Surgical treatment of infective endocarditis

Surgery will be required in about 30% of cases during the acute phase (first 4 months) of endocarditis, and in 20 to 40% of cases thereafter. Since surgery may be required at any time during an episode of endocarditis, it is essential to involve a cardiac surgeon in the overall management from the outset, which in practice means transferring the patient to a centre with cardiac surgery whenever possible. Surgery for endocarditis carries a risk of 10 to 25% mortality, with up to 25% of patients developing a paravalvular leak requiring a further operation. The main predictive factors for mortality associated with surgery are prosthetic valve endocarditis, infections due to staphylococci or candida, perioperative shock, or late referral.

The timing of surgery is all-important and demands experience and clinical judgement. The main indications are haemodynamic instability, persistent infection, and annular or aortic abscesses. In such cases surgery should never be delayed, even if only hours or days of antibiotic treatment have been given. The primary goals of the surgeon are to remove all infected material and to reconstruct the heart and/or restore valvular function at the lowest operative risk. An understanding of the surgical anatomy of infective endocarditis is a precondition for surgical success, which means the involvement of an experienced surgical team. Wherever possible, surgeons now strive to preserve the native valve, either by removal of the vegetation(s) or by valve repair. In prosthetic valve endocarditis, removal of all foreign material is mandatory. Actuarial survival figures indicate a 75% survival at 5 years and a 61% survival at 10 years after cardiac surgery for infective endocarditis.

There are several unresolved issues with regards to the surgical treatment of endocarditis. First, the use of surgery when embolization has taken place remains controversial. Recurrent emboli, persistent vegetation, after a major systemic embolus, and vegetation size (>10 mm), have all been put forward as indications, but there are no controlled trials to support a firm recommendation. Secondly, the optimal timing of surgery in patients who have had a cerebrovascular accident, either as a result of an embolic stroke or from haemorrhage due to a ruptured mycotic aneurysm: as a general rule, delay of at least 1 week is suggested if haemorrhage is detected by CT scanning, but surgery can be undertaken within 72 h if there is no haemorrhage present. Thirdly, the duration of antibiotic treatment postoperatively: if the excised valve is sterile, it is doubtful whether further antibiotics are of any benefit, but if the pathogen is isolated from the excised valve a continuance of 2 weeks of antibiotics seems reasonable, although if debridement is incomplete whatever antibiotics are given may fail.