David T. Durack MB, DPHIL1
Adolf W. Karchmer MD, FACP2
1Consulting Professor, Duke University School of Medicine, Vice President, Corporate Medical Affairs, Becton Dickinson, Franklin Lakes, New Jersey
2Professor of Medicine, Harvard Medical School, Chief, Division of Infectious Diseases, Beth Israel Deaconess Medical Center
David T. Durack, M.B., D.Phil., F.A.C.P., is an employee of Becton, Dickinson and Company and has received grants for educational activities from the Merieux Foundation.
Adolf W. Karchmer, M.D., has no commercial relationships with manufacturers of products or providers of services discussed in this chapter.
Infective endocarditis is a relatively rare but important disease, presenting in a wide variety of forms and manifestations; all are associated with significant morbidity and mortality. The diagnosis and management of infective endocarditis involves many clinical specialties, including internal medicine, cardiology, infectious diseases, microbiology, radiology, cardiovascular surgery, neurology, nephrology, and dentistry.
Definitions and Terminology
Infective endocarditis is a localized microbial infection of cardiac valves or mural endocardium caused by bacteria or fungi. The primary lesion is a vegetation, which is an infected platelet-fibrin thrombus located inside the heart. Native valve endocarditis is an infection of either normal or abnormal natural heart valves, whereas prosthetic valve endocarditis (PVE) involves implanted artificial valves. Nosocomial endocarditis occurs as a complication of medical treatment. On the basis of its clinical course, endocarditis may be classified as either acute or subacute. The acute form, often called acute bacterial endocarditis (ABE), is caused by invasive pathogens and tends to be rapidly progressive, usually leading to hospital admission less than a week after clinical onset and causing death in less than 4 weeks unless successfully treated. The subacute form, often called subacute bacterial endocarditis (SBE), is usually caused by low-grade pathogens or commensal organisms, with symptoms often present for weeks to months before diagnosis. Nonbacterial thrombotic endocarditis (NBTE) and marantic endocarditis refer to sterile vegetations in the heart, which can develop at the same locations as the vegetations of infective endocarditis. Infective endaortitis and infective endarteritis are analogous conditions that are localized on the endothelial surface of the aorta or large arteries, respectively.
The overall incidence of infective endocarditis ranges from 1.7 to 6.2 cases per 100,000 person-years.1,2,3 Although the overall incidence has remained relatively stable over the past 5 decades, the relative frequency of the disease in particular subgroups has changed, as have many other aspects of its epidemiology. The median age of patients with endocarditis has increased; 50% of patients are older than 55 years.1,4,5
Among the major predisposing conditions, chronic rheumatic heart disease has become relatively uncommon in developed countries, whereas degenerative valvular diseases, such as calcified aortic stenosis, calcified mitral valve annulus, and mitral valve prolapse have become more important.6 Cases of endocarditis associated with parenteral drug abuse and prosthetic cardiac valves have become more common, as has nosocomial infective endocarditis. Infections involving implanted intravascular devices other than valves are increasing in frequency.
Gram-positive cocci, comprising various species of streptococci and enterococci, as well as Staphylococcus aureus, are the leading cause of community-acquired native valve endocarditis1,2,3,6 [see Table 1]. SBE is usually caused by relatively avirulent bacteria, the most common species being streptococci from the normal oral or gastrointestinal flora. These organisms lack sufficient invasiveness to infect normal heart valves or endocardium, but they can infect deformed heart valves and some congenital cardiac lesions. The leading examples are the a-hemolytic streptococci, a heterogeneous collection of species that are loosely grouped together under the term viridans streptococci because they cause incomplete, greenish hemolysis when grown on blood agar.
Table 1 Causes of Native Valve Endocarditis2,3,21,86,87,88
Many of the streptococci that cause endocarditis can be categorized according to Lancefield serogrouping. About 20% are group D (mainly enterococci and Streptococcus bovis), about 15% are group H (S. sanguis and others), and about 15% belong to other serogroups, including B, C, G, and K. About 5% are anaerobic streptococci, and the remaining 40% to 45% are nongroupable viridans streptococci.
Enterococcus faecalis, E. faecium, and the nonenterococcal group D streptococcus S. bovis, all of which originate from the GI and genitourinary tracts, are important causes of SBE. The portal of entry for S. bovis is often a malignant or premalignant lesion in the colon.7Consequently, evaluation for possible colonic lesions is mandated for patients who have developed S. bovis SBE.
Coagulase-negative staphylococci (many of which are species other than S. epidermidis), and fastidious gram-negative HACEK organisms (Haemophilus, Actinobacillus actinomycetemcomitans, Cardiobacterium hominis, Eikenella, Kingella) are other important causes of infective endocarditis. Species that less commonly cause infective endocarditis are Brucella, Legionella, Coxiella burnetii, Chlamydia psittaci, andCorynebacterium.8,9 Bartonella, recognized as a cause of infective endocarditis only in the mid-1990s, may account for up to 3% of cases in some regions.10 B. quintana endocarditis occurs more commonly in the homeless, and endocarditis caused by B. henselae may be associated with cats.10,11,12 Corynebacteria (diphtheroids) usually involve prosthetic heart valves. Endocarditis caused by Coxiella burnetii, a notably indolent infection, occurs in the setting of preexisting valvular disease13 [see 7:XVII Infections Due to Rickettsia, Ehrlichia, and Coxiella].
ABE is commonly caused by primary pathogens, microorganisms that are more invasive than most of the species causing SBE. These invasive organisms can infect both normal and abnormal heart valves and mural endocardium; can rapidly destroy cardiac valves; can more readily invade perivalvular tissues to form abscesses; and can establish metastatic suppurative foci at distant sites in the body such as the brain or spleen. S. aureus is the most frequent and most important cause of ABE; others include S. pneumoniae, group A streptococci,Neisseria gonorrhoeae, Salmonella species, other members of the family Enterobacteriaceae, and Pseudomonas aeruginosa. Salmonella is the most frequent cause of infective aortitis in the elderly; these patients may have a continuous Salmonella bacteremia in the absence of a cardiac murmur.
Nosocomial endocarditis is a complication of nosocomial bacteremias or fungemias, often associated with intravascular and other indwelling devices, genitourinary tract manipulation, or wound infections.14,15,16,17,18 The primary pathogens are S. aureus, enterococci, and coagulase-negative staphylococci, most of which are methicillin-resistant S. epidermidis. In parenteral drug abusers, S. aureus (often methicillin resistant) causes 50% or more of cases of infective endocarditis19 [see Table 1].
Fungal endocarditis can be caused by yeasts or molds.20,21 The leading causative agents are yeasts of the genera Candida and Torulopsis; endocarditis from these yeasts arises as a complication of parenteral drug abuse, of prosthetic valves, or of intravascular devices, such as pacemakers. Endocarditis caused by mycelial or dimorphic fungi is relatively rare but can occur in parenteral drug abusers, patients with prosthetic valves, or immunocompromised patients. Aspergillus species cause most of these cases; the remainder are caused byHistoplasma, Penicillium, and a wide variety of other fungi.
Two host factors strongly predispose to development of infective endocarditis: (1) a damaged or otherwise abnormal endocardial surface and (2) high-velocity, turbulent blood flow across a defective valve or a congenital defect. Both factors favor platelet deposition. When blood is driven from a high-pressure area into a low-pressure sink, the dynamics of pressure gradients and turbulent flow favor deposition of platelets on nearby endothelium, especially if that endothelium is already abnormal as a result of previous disease such as rheumatic fever, congenital abnormalities, degenerative conditions, or infective endocarditis. Subsequently, a form of localized thrombosis can occur, resulting in the formation of a sterile platelet-fibrin aggregate known as nonbacterial thrombotic endocarditis.22 During transient bacteremias, passing microorganisms can adhere to this nidus, which provides them with both nutrition and protection from host defenses, especially leukocytes. The platelet-fibrin layers form an effective physical barrier between the embedded bacteria and leukocytes from the blood. This situation permits luxuriant growth; the organisms often attain vast numbers and form dense colonies within the platelet-fibrin aggregate. The bacteria themselves may promote further thrombosis by elaborating extracellular products that cause platelet aggregation or by eliciting procoagulant tissue factor from the endothelial substrate and local monocytes.23,24 In this way, the newly infected thrombotic lesion grows to form a macroscopic vegetation; these vegetations constitute the prototypical pathologic lesion of infective endocarditis.22
Local host defenses that could inhibit or kill bacteria in the vegetation include leukocytes, antibodies and complement, and platelet-derived bactericidal proteins (thrombocidins).25,26 These antimicrobial host defenses may succeed in curing some early cases of endocardial infection, but apparently they seldom if ever succeed in eradicating the organisms once a vegetation is fully established.
The streptococci that most frequently cause SBE—the viridans streptococci (including S. mutans, S. mitior, and S. sanguis [group H]), andS. bovis (group D)—readily adhere to platelet-fibrin thrombi, owing to the presence of dextran and other “sticky” molecules (adhesins) on their surfaces.27 Streptococcal species that do not produce dextran may also cause endocarditis, but they do so less frequently. Virulent organisms, such as those associated with ABE, may adhere to either normal or abnormal endocardial surfaces. Fibronectin receptors and clumping factor, which occur on the surface of S. aureus, appear to facilitate the adherence of these organisms to cardiac valves.27,28,29Bacterial species that are resistant to platelet-derived microbicidal proteins are more likely to cause endocarditis than susceptible strains.
SUBACUTE BACTERIAL ENDOCARDITIS
The constitutional symptoms of SBE usually begin insidiously and often persist for weeks to months. Fevers, sweats, weakness, myalgias, arthralgias, malaise, anorexia, and easy fatigability are prominent. Fewer than 5% of patients are afebrile, and such patients are often elderly, markedly malnourished, or azotemic. Chills and chilly sensations are common, but frank rigors are unusual.
Because of greater awareness of this disease and more frequent use of blood cultures in the evaluation of febrile illnesses, the diagnosis of SBE is now made earlier than in the past; consequently, many of the classic features of longer-standing SBE (e.g., splenomegaly, clubbing, and Osler nodes) are seldom present. Fever and other nonspecific symptoms in the presence of a predisposing cardiac lesion may be the only clinical manifestations of SBE in some patients. The presenting complaints may arise from organs or sites other than the heart, which may confuse the diagnosis. For example, if the patient has meningitis, cerebral emboli, or glomerulonephritis, the physician's attention may be focused on the central nervous system or kidneys as the primary site of illness. Although it was once taught that the symptoms of endocarditis in the elderly are milder than in young patients, current data suggest that the clinical features are fairly similar across age groups.30
ACUTE BACTERIAL ENDOCARDITIS
The onset of ABE is usually abrupt, and rigors are common.31,32 Fevers reach 39.4° to 40.6° C (102.9° to 105.1° F) and are often remittent. Cutaneous manifestations, particularly petechiae and small peripheral infarcts, may be prominent, especially in ABE from S. aureus. Occasionally, the clinical features in a patient with acute S. aureus endocarditis mimic those of acute meningococcemia, with similar skin lesions, including petechiae, purpura, and focal gangrene; similar hematologic changes, including disseminated intravascular coagulation; and neurologic findings of nuchal rigidity and CSF pleocytosis. Pustular petechiae or purulent purpura strongly suggest S. aureusendocarditis rather than meningococcemia [see Figures 1a, 1b, and 1c]. A Gram stain and culture of material from a skin lesion sometimes can reveal the etiologic agent and thus direct antibiotic therapy.
Figure 1a. Janeway Lesion
Findings on physical examination of patients with endocarditis. Shown here is an erythematous, palpable, nontender lesion at the base of the first finger consistent with a Janeway lesion.
Figure 1b. Conjunctival Petechiae
Findings on physical examination of patients with endocarditis. Shown here is a patient with conjunctival petechiae.
Figure 1c. Pustulonecrotic Septic Embolic Lesions
Findings on physical examination of patients with endocarditis. Shown here are pustulonecrotic septic embolic lesions at the base of the nail of the right great toe and at the medial aspect of the left great toe at the level of the distal interphalangeal joint.
Emboli are common in ABE. Metastatic infections in the spleen, bones (particularly the vertebrae), joints, kidneys, brain, eye (endophthalmitis), and lungs may arise from either septic embolization or sustained bacteremia. These metastatic infections may cause organ-specific symptoms or persistent fever in spite of antimicrobial therapy and may require drainage or surgical intervention. Osler nodes may occur, but less often than in SBE; Janeway lesions occur in 5% to 10% of patients who have S. aureus endocarditis [see Cutaneous Manifestations, below].
The appearance of a new cardiac murmur, particularly one characteristic of valvular insufficiency, strongly suggests valvular destruction and thus helps confirm a diagnosis of ABE. Valvular damage can lead to severe heart failure, necessitating valve replacement surgery. A few patients with ABE have no detectable cardiac murmur.
The cardiac manifestations of infective endocarditis reflect any underlying valvular or congenital lesions, with superimposed findings from endocarditis itself. Murmurs are present in more than 90% of patients. Changes in the intensity of a systolic murmur may be associated with the development of anemia, high fever, or tachycardia and are often of little significance. However, the appearance of a new murmur indicating valvular regurgitation is a key diagnostic event, which also may have implications for surgical intervention. A new aortic diastolic murmur suggests dilatation of the aortic annulus or eversion, rupture, or fenestration of an aortic leaflet. The sudden onset of a loud mitral pansystolic murmur suggests rupture of a chorda tendineae or fenestration of a mitral valve leaflet.
Heart failure resulting from valvular dysfunction may be progressive, becoming severe in some patients. Extension of infection into the annulus may result in cardiac rhythm disturbances, particularly if the infection progresses through the right coronary and noncoronary leaflet portion of the aortic annulus into the membranous septum and the area of the atrioventricular node.33 Occasionally, annular infection extends to the pericardium and causes pericarditis [see Diagnostic Features of Cardiac Complications, below].
Petechiae commonly occur in the conjunctivae [see Figures 1a, 1b, and 1c], in the oropharynx, and on the skin; they are particularly common on the lower extremities. Petechiae may continue to appear for some time despite appropriate antibiotic treatment. Linear subungual “splinter” hemorrhages located in the base or in the middle of the nail bed are a feature of SBE, whereas splinter hemorrhages in the distal nail bed are more often the result of trauma. Osler nodes—tender, purplish subcutaneous nodules that develop in the pulp of the fingers and disappear within several days—occur in about 5% of patients with endocarditis. Osler nodes may be caused by small emboli or may result from an immunologically mediated small-vessel vasculitis. Small, flat, nonpainful erythematous or hemorrhagic areas on the palms or soles, called Janeway lesions, are common in ABE and also may occur in SBE [see Figures 1a, 1b, and 1c].
Myalgias, arthralgias, arthritis, or low back pain occurs in 40% to 50% of patients with SBE; in about half of these patients, such symptoms represent either initial or prominent manifestations of the disease. Painful, warm, red, tender joints may be noted, but joint effusions are rare. Immunologic mechanisms cause synovial inflammation, articular symptoms, and abnormal laboratory test results. Rheumatoid factor is present in up to 50% of patients with endocarditis of more than 6 weeks' duration; antinuclear antibody is found in some patients; and circulating immune complexes occur in 75% to 90% of patients. Clubbing of the fingers is now seen in less than 15% of patients.
Petechial hemorrhages, flame-shaped hemorrhages, Roth spots, and cotton-wool exudates may be seen in the retinas of patients with endocarditis. Roth spots, which are oval white areas surrounded by a zone of bright-red hemorrhage, are noted in 3% to 5% of patients. Such ocular findings are not pathognomonic of endocarditis; they may be observed in patients with other disorders, such as severe anemia or collagen vascular diseases.
Significant arterial emboli occur in 30% to 50% of patients with endocarditis.34,35 Symptoms and signs include stroke; monocular blindness with occlusion of the central retinal artery; acute abdominal pain, ileus, and melena from mesenteric arterial occlusion; and pain and gangrene in the extremities. Emboli to the CNS are common and especially important, because they adversely affect survival rates and often result in permanent disability.35 Coronary emboli, which are found in as many as 50% of endocarditis patients who undergo autopsy, are often asymptomatic but occasionally result in frank myocardial infarction.
Factors associated with increased risk of embolization include vegetations of 10 mm or more in size as seen on echocardiography; vegetations on the mitral valve, particularly the anterior leaflet; vegetations that increase in size despite appropriate antibiotic therapy; and infection by S. aureus.34,35,36 The incidence of arterial emboli decreases about 10-fold during the initial 2 weeks of antimicrobial therapy.34Nevertheless, emboli occasionally occur late, after microbiologic cure has been achieved, and do not necessarily indicate that antimicrobial treatment has failed.
Pulmonary emboli are a common and important complication of right-sided endocarditis, frequently causing pulmonary infarcts or focal pneumonitis. Either may evolve into lung abscess, empyema, or pyopneumothorax, especially when invasive pathogens such as S. aureusare involved.
Splenomegaly occurs in 15% to 30% of patients with endocarditis. Splenic infarcts occur in up to 40% of patients; they may occur with or without splenomegaly. Splenic infarcts may cause sharp left upper quadrant pain, but they are commonly asymptomatic. Splenic abscesses, which develop in about 5% of patients, may manifest as left shoulder, left upper quadrant, or pleuritic left chest pain; fever that persists during antibiotic therapy; or a relapse of bacteremia. Splenic lesions are best imaged by computed tomography, magnetic resonance techniques, and, to a lesser degree, ultrasonography; however, the differentiation of abscesses from infarcts by imaging is difficult.
Microscopic hematuria is observed in about 50% of infective endocarditis patients. Embolic renal infarction may cause flank pain and hematuria, but it rarely causes renal failure. Some degree of acute renal failure occurs in up to one third of patients and is associated with a worse prognosis,37 but it usually resolves completely if therapy for endocarditis is successful. Diffuse membranoproliferative glomerulonephritis results from immune complex glomerulonephritis, with deposition of IgG, IgM, and complement in a granular or nodular fashion in the glomerular basement membrane or on the basement membrane in subepithelial or subendothelial locations. Bacterial antigen may be identified in these glomerular deposits. In such patients, serum complement levels usually are reduced. Diffuse membranoproliferative glomerulonephritis may cause renal failure, which generally resolves after treatment of the infection. Focal embolic glomerulonephritis, originally thought to result from small bacterial emboli, is now recognized as an anatomic variation of immune complex disease. It occasionally causes renal failure.
Mycotic aneurysms are arterial aneurysms that develop in association with infections, especially infective endocarditis. Mycotic aneurysms occur in 2% to 8% of infective endocarditis patients and can form in any artery; they are particularly important when they involve cerebral arteries38,39 [see Neurologic Manifestations, below]. They may become symptomatic early, during the active phase of endocarditis, or late, after valvular infection has been eradicated.40 The clinical manifestations, which arise from enlargement or rupture of the aneurysm, include headache, pain, a pulsatile mass, persistent fever despite appropriate antibiotic treatment, focal signs from pressure on adjacent structures, the sudden development of an expanding hematoma, or signs of major blood loss. Small aneurysms (less than 5 mm in size) may resolve during antibiotic therapy. For larger aneurysms, to prevent possible rupture, therapeutic embolization, clipping, or excision is usually indicated when such interventions are feasible without undue risk.
Neurologic complications develop in 25% to 40% of patients with endocarditis, causing major morbidity and increased mortality.41,42,43Altered mental status at presentation is associated with higher mortality at 6 months.44 Cerebral embolism may be an initial manifestation of endocarditis. Strokes caused by cerebral emboli, commonly to the middle cerebral artery or one of its branches, are the most frequent major neurologic complication, occurring in about 15% of patients. Some patients have multiple small embolic infarcts, which may manifest as an altered level of consciousness, seizures, fluctuating focal neurologic signs, or a combination of these symptoms. An intracerebral hemorrhage in a patient with infective endocarditis can be secondary to an embolic stroke or to rupture of a mycotic aneurysm.42 Cerebral mycotic aneurysms occur in 1% to 5% of patients with infective endocarditis and most commonly affect the distal branches of the middle cerebral arteries.39 Multiple aneurysms occur in some patients. In patients with symptomatic intracranial aneurysms, the overall mortality is high—above 50%. Patients with mycotic aneurysms may present with headache, focal signs, or, if the aneurysm ruptures, manifestations of acute intracerebral or subarachnoid hemorrhage. A slowly leaking aneurysm may cause mild meningeal irritation. In such cases, the cerebrospinal fluid, although sterile, may contain erythrocytes, leukocytes, and an increased concentration of protein.
In the search for an intracerebral aneurysm, contrast-enhanced CT may provide localizing information by detecting intracerebral bleeding. Magnetic resonance angiography is insufficiently sensitive to reliably detect aneurysms that are 5 mm or less; therefore, cerebral angiography is the optimal diagnostic test.45,46 Cerebral angiography for the detection of aneurysms has been advised for patients with focal neurologic signs, especially in the setting of ABE, and for patients with persistent unexplained headache or meningeal irritation, particularly if anticoagulant therapy is planned.
Treatment of a mycotic aneurysm that has been detected by angiography depends on its location and surgical accessibility, the presence or absence of hemorrhage, and changes in size that may occur during antimicrobial therapy. Resolution or healing of mycotic aneurysms during treatment of endocarditis has been demonstrated by angiography.47 However, a leaking aneurysm, an aneurysm that is large or progressively enlarging, or one that persists after antibiotic therapy should be removed surgically, provided it is accessible. These complex cases should be managed with the input from specialists in infectious disease, neurology, radiology, and neurosurgery.
Brain abscesses are uncommon in SBE. In patients with acute S. aureus endocarditis, septic emboli may give rise to multiple intracerebral foci of inflammation or to small abscesses.
Toxic encephalopathy and seizures, which are usually triggered by emboli or strokes, also may complicate active endocarditis. A CSF pleocytosis with polymorphonuclear leukocytes predominating is observed in some patients with ABE caused by pyogenic organisms, especially S. aureus. Patients with SBE may have findings of aseptic inflammation in the CSF.
ENDOCARDITIS ASSOCIATED WITH PARENTERAL DRUG ABUSE
The annual incidence of endocarditis in injecting drug users (IDUs) is 0.2% to 2.0%. At the time of their initial attack of endocarditis, 70% to 80% of IDUs have no history or findings of preexisting valvular heart disease. In IDUs, the tricuspid valve is infected more frequently (55%) than the aortic valve (35%) or mitral valve (30%). Multiple episodes of endocarditis are common in IDUs.
Although many of the manifestations of endocarditis in IDUs are similar to those of ABE in non-IDUs, there are some differences, because of the high frequency of tricuspid valve involvement, the spectrum of infecting organisms, and occasional pulmonic valve infection. High fevers, chills, rigors, malaise, cough, and, especially, pleuritic chest pain are common presenting complaints in right-sided endocarditis in IDUs. Septic pulmonary emboli occur in about 75% of cases, particularly in patients with S. aureus infection, and cause sputum production, hemoptysis, and initial radiologic findings that may suggest pneumonia. Cavitation of embolic pulmonary lesions is quite common. Significant cardiac murmurs are heard in most patients at some time during their illness but may not be present initially. The murmur of tricuspid regurgitation, a short ejection systolic murmur that is louder on inspiration, may be difficult to detect. Hemodynamically, significant tricuspid insufficiency is manifested by V waves in the jugular vein and a pulsating liver.
Because S. aureus and other pyogenic bacterial species are the predominant causes of infective endocarditis in IDUs, metastatic infections are a frequent complication. Neurologic manifestations and peripheral emboli are common; the latter may occlude major vessels and require surgical management.
PROSTHETIC VALVE ENDOCARDITIS
The incidence of PVE is 1% to 2% at 1 year and approximately 0.5% per year thereafter, resulting in a cumulative incidence of 4% to 5% during the first 5 years after valve implantation.49,50,51 Infection may be introduced at the time of valve placement or from transient bacteremia at any time thereafter. The overall risks of infection are similar for mechanical and porcine bioprosthetic valves and for aortic and mitral valve prostheses.49,50,51 The leading cause of PVE during the first year after surgery is methicillin-resistant coagulase-negative staphylococci, predominantly S. epidermidis [see Table 2]. Coagulase-negative staphylococci continue to cause cases of PVE that occur a year or more after surgery; however, these staphylococci are often species other than S. epidermidis, and only 20% to 30% are methicillin resistant. S. aureus, streptococci, enterococci, and fastidious gram-negative coccobacilli, the leading organisms associated with native valve endocarditis, cause about three quarters of PVE cases after 1 year following valve replacement.49,50,51 Coagulase-negative staphylococci are responsible for at least 35% of cases of PVE and, therefore, should not be dismissed as contaminants (which they usually are in other settings) if isolated from the blood of a patient who has a prosthetic valve.
Table 2 Etiology of Prosthetic Valve Endocarditis74,89,90
Infection of prosthetic valves is often associated with invasion of perivalvular tissues—resulting in valve-ring abscessses and valvular dysfunction—and occasionally with myocardial abscesses.41,49,52,53 Necrosis of the annulus from invasive infection can cause partial dehiscence of the prosthesis, resulting in hemodynamically significant paravalvular regurgitation. These pathologic changes can occur with either porcine or mechanical prostheses, particularly when the valves are in the aortic position or when infection occurs during the first postoperative year.51,54 Occasionally, vegetations may partially obstruct the valve orifice or restrict valve movement, causing functional stenosis. Such changes are more likely with prostheses in the mitral position. When infection is restricted to the leaflets of a porcine bioprosthetic valve, leaflet destruction, obstructing vegetations, and the delayed onset of leaflet stiffness may cause clinically significant valvular dysfunction.54
The dominant clinical feature of PVE that occurs during the first 60 days after surgery (early PVE) is fever, whether or not there is a regurgitant murmur associated with the prosthetic valve. Prosthetic valve dysfunction with resulting heart failure is seen in some patients. Petechiae occur in about half of patients with early PVE, but Roth spots, Osler nodes, and Janeway lesions are not common. Emboli are common in early PVE; emboli that occlude large peripheral arteries suggest fungal endocarditis. Because blood cultures are often negative in patients with fungal endocarditis, this diagnosis must often be made on the basis of histologic examination, culture, and sometimes molecular testing of surgical or autopsy specimens or vegetation recovered at embolectomy.
The clinical features of PVE that occurs more than 60 days after surgery (late PVE) are similar to those of acute or subacute endocarditis on native valves, depending on the infecting organism.
Although the diagnosis of endocarditis may be readily evident in patients with the classic syndrome of fever, a murmur associated with valvular dysfunction, typical peripheral signs, and bacteremia, the diagnosis is less evident in most patients.55 A diagnostic approach known as the Duke criteria has been designed by the Duke Endocarditis Service56 [see Tables 3 and 4]. With these criteria, the diagnosis of endocarditis can be established definitively by either pathologic or clinical criteria. The pathologic criteria include direct evidence gleaned from surgery or autopsy, and the clinical criteria are derived from microbiologic data (culture, serologic or molecular testing), echocardiography, physical examination, and other laboratory findings [see Table 4]. In retrospective evaluations of pathologically proven cases of native valve endocarditis and PVE, these clinical criteria have been found to be both sensitive56,57,58 and highly specific.57,59,60Misdiagnoses rarely resulted when these clinical criteria were used to reassess previously pathologically confirmed cases. The specificity and negative predictive values of these criteria have been reported as 99% and 92%, respectively.59,60 Use of these clinical criteria very rarely results in the rejection of cases considered to be endocarditis on independent expert evaluation.61
Table 3 Duke Criteria for the Diagnosis of Infective Endocarditis43
Table 4 Definitions of Terms Used in Duke Criteria for the Diagnosis of Infective Endocarditis42
The Duke criteria utilize echocardiography for defining the anatomic features of endocarditis. Both the sensitivity and specificity of echocardiography are high when experienced echocardiographers apply specific criteria. In patients with pathologically proven native valve endocarditis, vegetations can be detected in 60% to 75% by use of transthoracic echocardiography (TTE) and in 87% to 94% with transesophageal echocardiography (TEE); the specificity of both techniques is high.62,63,64 TEE is notably superior to TTE in the evaluation of patients with suspected PVE65 [see Prosthetic Valve Endocarditis, above]. Notwithstanding the increased sensitivity of TEE compared with TTE, if the decision is made to treat all cases identified by the Duke criteria as definite endocarditis and most cases identified as possible endocarditis, then evaluation by TEE rarely alters the treatment decision. Occasional exceptions to this observation are PVE cases that were missed by TTE. Echocardiography is the preferred technique for identification of perivalvular infection and other intracardiac complications of endocarditis. For detection of abscesses, a transesophageal study is significantly more sensitive (76% to 87%) than transthoracic imaging (18% to 28%), with equal specificity.53,65,66
All patients in whom endocarditis is seriously suspected should undergo echocardiography. However, echocardiography should not be regarded as a general screening test for patients with a low prior probability of endocarditis (e.g., most patients with acute febrile illnesses). Given its high sensitivity, a negative transesophageal echocardiogram is good evidence against endocarditis in patients at low or intermediate risk of this infection. Conversely, if the prior probability of endocarditis is high, a negative echocardiogram does not fully exclude the diagnosis; the false negative rate for transesophageal echocardiography is 6% to 13%.67 A repeat transesophageal study in patients with endocarditis reduces the false negative results to approximately 5%.67 Finally, echocardiography cannot reliably distinguish an infected vegetation from a sterilized vegetation, nonbacterial thrombotic vegetations of marantic endocarditis, or intracardiac thrombi, as seen in the antiphospholipid antibody syndrome.
Transthoracic echocardiography has limited usefulness in the diagnosis of PVE because the prosthesis itself produces echoes that often obscure vegetations and abscesses. Transesophageal two-dimensional and Doppler echocardiography more effectively assess prosthetic valves and perivalvular tissues, especially when a mitral valve prosthesis is present.68,69 In PVE, the transesophageal technique identified vegetations in 82% of patients, compared with a 36% identification rate with transthoracic echocardiography.53 Similarly, detection of paravalvular abscesses in patients with PVE is markedly increased by use of transesophageal rather than transthoracic echocardiography.68
For patients suspected of having endocarditis, identification of the microbial etiologic agent and determination of its antimicrobial susceptibility are of paramount importance for both diagnosis and treatment. In most patients with SBE, blood cultures drawn before initiation of antibiotic therapy will all be positive, reflecting the sustained bacteremia associated with an infected endothelial surface. To optimize the value of blood cultures, best-practice guidelines recommend careful antiseptic skin preparation to minimize contamination of the specimen with skin flora: cleaning, followed by application of 70% isopropyl alcohol, which is allowed to dry, followed by application of an iodophor or chlorhexidine, which is also allowed to dry.70 Blood should be drawn using three separate venipunctures, at three different sites, that are taken over several hours if the patient presents with a subacute syndrome or that are taken over a few minutes if acute endocarditis is suspected. The ideal volume for each specimen is 12 to 20 ml, which should be divided equally into two culture bottles, yielding six bottles in all. Obtaining a large volume of blood for culture maximizes the yield,70 but smaller volumes must be accepted from infants and small children. Use of both aerobic and anaerobic broth media also helps maximize yield. If the patient has recently received any antibiotic treatment, use of a medium that contains an antibiotic removal device such as resin will increase the yield by 5% to 15%.70 Blood cultures should be incubated for 5 days, because modern improved culture media will yield growth for the vast majority of relevant microorganisms within this period. This is true even for most of the so-called fastidious or slow-growing bacteria such as the HACEK group and the nutritionally variant streptococci (Abiotrophia species).
Blood cultures are negative in 5% to 20% of patients with infective endocarditis9,71 [see Tables 1 and 2]. In about half of these cases, blood cultures are negative as a consequence of previous antimicrobial therapy, even though the vegetation is still infected. Recovery of the causative organism from the blood of these previously treated patients may often be accomplished by repeating blood cultures several days after antibiotics have been discontinued, but some patients remain persistently blood culture negative. Blood cultures may also be negative in some cases of right-sided endocarditis caused by relatively noninvasive organisms. Currently, however, most cases of right-sided endocarditis occur in IDUs and are caused by pyogenic bacteria, such as S. aureus and P. aeruginosa, which are readily isolated from the blood. Culture of bone marrow or arterial blood does not provide materially more information than can be obtained from culture of venous blood.
Several additional factors can thwart isolation of the infecting agent from blood of patients with infective endocarditis. Bartonella species may require more than 5 days for isolation, and certain mycelial fungi such as H. capsulatum and Aspergillus species are difficult to isolate even with special techniques such as lysis centrifugation. Finally, isolation of C. burnetii and C. psittaci requires techniques beyond the capabilities of most laboratories.13,72,73 Endocarditis caused by Bartonella species, Legionella species, Brucella species, C. psittaci, and C. burnetii can be presumptively diagnosed with serologic tests. These tests should be performed in the evaluation of apparently culture-negative endocarditis, particularly when negative cultures cannot be attributed to prior antimicrobial therapy.73 Direct cultures on special media, histopathologic examination with special stains, and molecular techniques to recover DNA or 16S ribosomal RNA all can be used to determine etiology from examination of vegetations that have been removed from valves or, having embolized, from peripheral arteries.73,74,75,76
Bacteremia in Patients with Prosthetic Heart Valves
As in native valve endocarditis, the diagnosis of PVE is usually based on history, physical examination, echocardiography, and the detection of bacteremia. However, only a subgroup of patients with a prosthetic valve who experience bacteremia subsequently develop PVE. Of bacteremic patients with a prosthetic valve whose initial positive blood culture is not sentinel evidence of active PVE itself, only 15% to 20% will develop PVE caused by that blood culture isolate.17
Transient bacteremia with gram-negative bacilli from extracardiac sources usually does not result in colonization of prosthetic valves; however, recrudescent or sustained bacteremia that occurs after the extracardiac focus of infection has been eradicated suggests PVE, as does persistent gram-negative bacillary bacteremia with no identifiable extracardiac source, even if a new regurgitant murmur or other signs of endocarditis are lacking. Persistent or high-grade coagulase-negative staphylococcal bacteremia in a patient with a prosthetic valve strongly suggests PVE. Similarly, PVE is likely when coagulase-negative staphylococci that have been isolated sporadically from multiple blood cultures are shown by molecular techniques to belong to a single clone. Blood cultures are negative in about 6% of PVE cases. Negative cultures are usually the result of previous antibiotic therapy or reflect unique characteristics of the infecting organisms.
ADJUNCTIVE LABORATORY TESTS
Results of many laboratory tests are likely to be abnormal in patients with endocarditis because of the systemic impact of the infection and because of injury to various organs. Such tests include the complete blood count, urinalysis, blood urea nitrogen level, creatinine concentration, rheumatoid factor, quantitative immunoglobulins, complement levels, erythrocyte sedimentation rate (ESR), and C-reactive protein level. These adjunctive tests may offer clues to the diagnosis and may be useful for monitoring the progress of treatment for endocarditis, but they usually are not specific and are not critical in establishing a diagnosis; hence, they are not among the Duke diagnostic criteria.
DIAGNOSTIC FEATURES OF CARDIAC COMPLICATIONS
Heart failure is the most frequent cardiac complication of infective endocarditis and may result from a variety of factors. Preexisting valvular disease can be worsened by the effects of infective endocarditis, which can include tears, perforations, and obstruction of valves and rupture of chordae tendineae. These complications can also affect previously normal valves. Damage to the aortic valve by infection can cause rapidly progressive and severe hemodynamic impairment, more so than comparable damage to the mitral valve. Coronary artery embolism can cause silent or overt myocardial infarction, which can contribute to heart failure. A mycotic aneurysm of a sinus of Valsalva or an aortic annulus abscess may rupture through the membranous septum into the right atrium or ventricle [see Figure 2]. Flow through the resulting fistula causes a sudden rise in the jugular venous pressure and a continuous or to-and-fro murmur and thrill along the left sternal border. The lungs remain relatively clear. Valvular damage and subsequent dysfunction, as well as intracardiac fistula formation, can be accurately defined with two-dimensional and Doppler echocardiography from a transthoracic or transesophageal approach.
Figure 2. Anatomy of the Aorta
Anatomic relations between the noncoronary cusp and left cusp of the aortic valve, interventricular septum, membranous septum, tricuspid valve, and mitral valve are shown schematically at the level of the aortic root.
The development of new conduction abnormalities may signal the extension of infection into the septum, affecting its conduction tissues.33,52 PR interval prolongation, left bundle branch block, or right bundle branch block with left anterior hemiblock suggests the extension of infection from the aortic valve. The proximity of the weakest area of the aortic valve annulus to the membranous septum and the conduction system accounts for the development of these conduction abnormalities [see Figure 3]. Similarly, extension of infection from the mitral annulus, which is close to the bundle of His and to the atrioventricular node, may also produce conduction defects, but such extension occurs less frequently than extension from the aortic valve. In the absence of digitalis toxicity or a recent inferior myocardial infarction, the development of nonparoxysmal junctional tachycardia, a Wenckebach block, or complete heart block with a narrow QRS complex serves as a clue to the spread of infection from the mitral annulus into the AV node and proximal bundle of His. Ventricular premature beats in patients with endocarditis who do not have electrolyte abnormalities or digitalis toxicity may reflect myocarditis, myocardial abscesses, or coronary arterial emboli.
Figure 3. Anatomy of Cardiac Valves
The close relation of the three cardiac valves and the cardiac conduction system, as seen in this superior schematic view, accounts for the appearance of conduction defects in endocarditis. (AV—atrioventricular)
The presence of an annular abscess may be suggested by the recent onset of aortic regurgitation, persistent fever during appropriate antimicrobial therapy, and the development of pericarditis.77,78 Pericarditis caused by extension of a valve ring abscess into the epicardium is more often hemorrhagic or fibrinous than purulent. Occasionally, pericarditis in the course of infective endocarditis is the result of transmural myocardial infarction secondary to coronary emboli. Transesophageal echocardiography is the most sensitive and preferred noninvasive test for detecting valve-ring abscesses in both native valve endocarditis and PVE.
The possibility of infective endocarditis should be considered in any patient with a heart murmur and fever. The physician must be particularly alert for atypical cases in which the clinical findings reflect complications of endocarditis affecting organs other than the heart.
Infective endocarditis can cause fever of undetermined origin (FUO), so the differential diagnosis includes the many other infections that may cause FUO. These include tuberculosis, salmonellosis, and various intra-abdominal and genitourinary infections [see 7:XXIV Hyperthermia, Fever, and Fever of Undetermined Origin].
A variety of noninfectious illnesses can mimic infective endocarditis, including immune-mediated diseases and rheumatologic conditions such as juvenile rheumatoid arthritis and polymyalgia rheumatica. Acute rheumatic fever can cause fever, cardiac murmurs, and heart failure, but acute rheumatic fever can be distinguished from infective endocarditis on clinical grounds and by negative blood cultures, raised anti-streptolysin O antibody titer, and response to salicylates [see 7:I Infections Due to Gram-Positive Cocci]. Marantic endocarditis, which can give rise to multiple embolic episodes and fever, is usually associated with an underlying neoplasm or chronic wasting disease. In polyarteritis nodosa, the presence of fever, anemia, and renal involvement may suggest SBE, and the findings on biopsy of a lesion in a large artery may even resemble those of a mycotic aneurysm. In both systemic lupus erythematosus and antiphospholipid antibody syndrome, the manifestations of fever, nonbacterial thrombotic vegetations, systemic emboli, and spontaneous thrombotic events can simulate infective endocarditis.
A cardiac myxoma, usually in the left atrium, may mimic infective endocarditis in both clinical and laboratory features, including low-grade fever, weight loss, arthralgias, cutaneous lesions, clubbing of the fingers, emboli to major arteries, and auscultatory findings suggesting mitral stenosis and regurgitation. Cardiac myxoma syndrome may further simulate infective endocarditis by giving rise to cerebral aneurysms at the sites of myxomatous emboli. Negative blood cultures and echocardiography can help establish the correct diagnosis.
Neoplasms may mimic infective endocarditis by inducing marantic endocarditis or by their hemodynamic effects. For example, richly vascular tumors may be associated with fever, anemia, and hyperdynamic circulation with flow murmurs. A left renal tumor mass may be mistaken for splenomegaly. Carcinoid tumors occasionally mimic endocarditis when they produce endocardial and valvular fibrosis that lead to tricuspid insufficiency and pulmonary stenosis.
Two major modalities are used to treat endocarditis: (1) antibiotic therapy and (2) surgical debridement of vegetations and infected perivalvular tissue, with valve repair or replacement as needed.
Effective antimicrobial treatment requires identification of the etiologic agent and determination of its antimicrobial susceptibility. Therefore, in the evaluation of a patient with subacute or indolent disease, it is usually best to delay antibiotic therapy until the results of blood cultures are obtained. If recently administered antibiotics have rendered the initial cultures negative, this delay provides an opportunity to obtain additional blood cultures after the antibiotics and their effects have dissipated. However, if the infection is fulminant or if there is valvular dysfunction that may require urgent surgical intervention, empirical antibiotic therapy must be initiated promptly after blood culture specimens have been obtained. Bactericidal antibiotics are used parenterally in high doses. With the exception of PVE caused by staphylococci, antimicrobial therapy for PVE caused by a specific organism utilizes the same drugs recommended for native valve endocarditis. However, therapy is usually administered over a longer period, typically 6 weeks. Patients must be evaluated frequently to assess the efficacy of antimicrobial therapy and the development of complications of therapy or infection.
Viridans streptococci, once mostly penicillin sensitive, have demonstrated increasing resistance to penicillin and the cephalosporins over the past 20 years.79 Accordingly, in the planning of endocarditis treatment, all streptococci must be evaluated for susceptibility to penicillin by determining the minimum inhibitory concentration (MIC). Various regimens provide effective treatment for endocarditis caused by those streptococci that are fully penicillin sensitive (MIC < 0.2 µg/ml) [see Table 5].80 One regimen employs parenteral penicillin alone in high doses for 4 weeks. A second regimen utilizes the synergism achieved against most strains of nonenterococcal streptococci by the combination of penicillin and gentamicin. This synergism allows effective treatment with only 2 weeks of combination therapy.80,81 The short-duration regimen should be considered only for selected cases of native valve endocarditis with favorable prognostic features: streptococcal infections that are not complicated by hypotension, renal failure, thrombocytopenia, mycotic aneurysms, or heart failure caused by valvular dysfunction. The etiologic species should be highly susceptible to penicillin (MIC < 0.2 µg/ml) and should not be a nutritionally variant strain—that is, dependent upon pyridoxal or cysteine for growth. PVE should not be treated with the 2-week regimen. Ceftriaxone, given in a single daily intravenous dose for 4 weeks, is now often recommended for treatment of endocarditis caused by penicillin-sensitive streptococci because this regimen is easily adapted for outpatient treatment and has little toxicity.49,80 Although short-duration combination therapy for penicillin-susceptible streptococcal endocarditis using single daily doses of both ceftriaxone and an aminoglycoside has been successful, experience with this regimen is limited, and the regimen is not recommended for general use. S. bovisis highly penicillin sensitive and can be treated with regimens recommended for other penicillin-sensitive streptococci. When combination penicillin-gentamicin therapy is used to treat endocarditis caused by nonenterococcal streptococci, particularly when short-duration therapy is planned, the streptococcus should be screened for high-level resistance to gentamicin. Although rare in these organisms, high-level resistance would preclude bactericidal synergy and thus indicate the need for an alternative regimen.
Table 5 Antimicrobial Therapy for Endocarditis in Adults
Endocarditis caused by relatively penicillin-resistant (MIC = 0.2 to 0.5 µg/ml) viridans or other nonenterococcal streptococci is treated with a higher dose of penicillin G, combined with gentamicin. If the strain is even more resistant to penicillin (MIC > 0.5 µg/ml), the infection is treated with one of the standard regimens for enterococcal endocarditis [see Table 5]. Nutritionally variant streptococci (previously called S. adjacens or S. defectivus; now named Abiotrophia species) are often relatively penicillin resistant, so endocarditis caused by these organisms should be treated with the standard regimen for enterococcal endocarditis.49,80,82 Endocarditis caused by pneumococci or group A streptococci is treated with intravenous penicillin G in a dosage of 20 million units daily for 4 weeks.49,80 Pneumococci that are found to be the cause of endocarditis must be tested for susceptibility to penicillin. Vancomycin is the preferred treatment for endocarditis caused by penicillin-resistant strains with an MIC greater than 1.0 µg/ml. When pneumococcal endocarditis is complicated by concurrent meningitis, the treatment regimen must ensure adequate penetration of antibiotic into cerebrospinal fluid.83 The b-hemolytic streptococci belonging to groups B, C, and G have slightly reduced susceptibility to penicillin; for endocarditis, they should be treated as if they were relatively penicillin resistant, with MICs of 0.2 to 0.5 µg/ml.80,84
Enterococci are relatively resistant to penicillin, ampicillin, and vancomycin and are fully resistant to cephalosporins. Antibacterial synergism is essential for optimal antimicrobial treatment of enterococcal endocarditis. To achieve this, the enterococcus must simultaneously be exposed to a cell wall-active antibiotic such as penicillin, ampicillin, or vancomycin, at a concentration at or above the organism's MIC, and to an aminoglycoside that will exert a lethal effect.49,80,85 The ability of some enterococci to grow in the presence of gentamicin at concentrations of 500 µg/ml or higher indicates high-level aminoglycoside resistance; against such organisms, combination therapy will fail to exert a lethal effect regardless of the cell wall-active antimicrobial agent employed. High-level resistance to gentamicin is the consequence of aminoglycoside-modifying enzymes.
Previously, synergistic bactericidal therapy could be reliably anticipated when gentamicin was combined with penicillin, ampicillin, or vancomycin. This provided regimens for treatment of enterococcal endocarditis49,80 [see Table 5].
Currently, antimicrobial resistance in enterococci presents a complex problem that must be carefully considered in the selection of therapy for enterococcal endocarditis.85 The causative strain must be screened for high-level resistance to gentamicin. Use of gentamicin in combination therapy in the face of high-level resistance exposes the patient to potential toxicity and is without therapeutic benefit. Furthermore, resistance to cell wall-active agents has become increasingly prevalent in enterococci. Intrinsic resistance to penicillin and ampicillin (MIC ≥ 32 µg/ml) is prevalent in E. faecium. Penicillin and ampicillin resistance caused by β-lactamase production, which is not detectable with MIC tests but requires screening with the chromogenic cephalosporin nitrocefin, is occasionally seen in E. faecalis. Finally, vancomycin resistance (MIC ≥ 16 µg/ml) is being encountered increasingly in E. faecalis and E. faecium. If an enterococcus is resistant to a cell wall-active agent, that agent cannot participate in the synergistic killing of the strain. Vancomycin is a suitable cell wall-active agent for combination therapy when organisms have intrinsic or β-lactamase-mediated resistance to penicillin or ampicillin; ampicillin-sulbactam is suitable when enterococci have resistance that is mediated by β-lactamase. Enterococci that are resistant to vancomycin may be susceptible to penicillin and ampicillin but more often are resistant to these antibiotics as well. For a few of these enterococci, teicoplanin, a glycopeptide antibiotic that has not been approved for use in the United States but is available elsewhere, remains an effective cell wall-active antimicrobial.80,85 Endocarditis caused by vancomycin-resistant enterococci that are not susceptible to ampicillin requires special, individualized regimens, chosen with the assistance of an infectious disease consultant. All enterococci that cause endocarditis must be tested for susceptibility to antimicrobials that have therapeutic potential. Thereafter, a bactericidal synergistic combination of a cell wall-active agent (e.g., penicillin, ampicillin, vancomycin, ampicillin-sulbactam, or teicoplanin) plus gentamicin can be selected. Single daily dosing of gentamicin should not be used in treating enterococcal endocarditis. If a synergistic combination is not possible because of high-level resistance to gentamicin, prolonged therapy (8 to 12 weeks) with high doses of an effective cell wall-active agent should be administered. To prevent or minimize possible toxicity of gentamicin or vancomycin, it is important to monitor serum levels, renal function, and otologic symptoms every 3 to 5 days throughout therapy.49,80
Bactericidal synergistic therapy for enterococcal endocarditis is associated with cure rates approaching 85%. Treatment with an effective cell wall-active agent alone results in cure rates of 40% to 50% at best.80,86 If bactericidal synergistic therapy is not available, patients with enterococcal endocarditis in whom treatment with a cell wall-active agent alone has failed should undergo excision of the infected valve while suppressive antibiotic therapy is continued.49,80,85,86
The treatment of choice for S. aureus native valve endocarditis is a penicillinase-resistant penicillin such as nafcillin or oxacillin49,80 [seeTable 5]. Vancomycin provides an alternative treatment for patients who are hypersensitive to penicillins. Vancomycin, although active against both methicillin-sensitive S. aureus (MSSA) and MRSA, is not the first choice, because it is somewhat less effective than the penicillinase-resistant penicillins.87 However, vancomycin should be added initially because of the increasing frequency of MRSA strains, not only in nosocomial infections but also in community-acquired infections. If the blood culture yields MSSA, the vancomycin should be discontinued; if MRSA is yielded, then vancomycin alone is continued80 [see Table 5]. Penicillin G, 20 to 30 million units daily, should be substituted for the few strains of S. aureus that prove to be penicillin sensitive, but only after susceptibility has been confirmed by tube dilution testing and the isolate has been demonstrated not to produce penicillinase. Gentamicin is synergistic with penicillinase-resistant penicillins against S. aureus in vitro, and combination therapy may reduce the duration of bacteremia and fever in patients. However, the in vivo benefit is marginal; combination therapy does not improve overall cure rates. If the option to add gentamicin is chosen, it should be stopped after about 3 days to minimize the risk of nephrotoxicity.
Rare strains of S. aureus that are partially or even fully resistant to glycopeptides such as vancomycin have been detected. In the unlikely event of endocarditis being caused by such a strain, alternative drugs to be considered include linezolid, quinupristin-dalfopristin, or teicoplanin and daptomycin.
Recommended regimens for treatment of endocarditis in IDUs are usually the same as regimens for treatment in other patients. However, in IDUs who have MSSA endocarditis restricted to the tricuspid valve, without complicating metastatic infection, successful treatment has been accomplished with a penicillinase-resistant penicillin plus an aminoglycoside (e.g., nafcillin plus gentamicin or tobramycin) given for only 2 weeks.88 Such patients who do not respond promptly and completely to this short-term regimen should receive longer courses of treatment.
Endocarditis caused by coagulase-negative staphylococci is usually engrafted on a prosthetic valve. More than 80% of the coagulase-negative staphylococci that cause PVE within 1 year after valve implantation are resistant to methicillin and to all β-lactam antibiotics, including cephalosporins.49,80 Only when the onset of this form of PVE occurs more than a year after surgery does the frequency of β-lactam resistance in the coagulase-negative staphylococci fall below 30%.89,90 Moreover, the β-lactam resistance may not be detected by standard antimicrobial susceptibility testing, particularly when automated instruments or microsystems are used. Most of these coagulase-negative staphylococci are sensitive to vancomycin and rifampin, but plasmid-mediated resistance to gentamicin is prevalent. For PVE caused by methicillin-resistant organisms that are susceptible to rifampin and gentamicin, the most effective treatment combines vancomycin with rifampin and gentamicin [see Table 5].49,80,90 If the infecting strain is resistant to gentamicin, either a fluoroquinolone or another aminoglycoside to which the strain is susceptible should be substituted for gentamicin. If the staphylococci are susceptible to methicillin, a penicillinase-resistant penicillin should be used in lieu of vancomycin [see Table 5]. Antimicrobial therapy for PVE caused by S. aureusshould utilize the same regimens as those used for coagulase-negative staphylococcal PVE [see Table 5].
Rigorous susceptibility testing is required, but the coagulase-negative staphylococci that cause native valve endocarditis are usually sensitive to β-lactam antibiotics unless the infection has been acquired nosocomially. Native valve endocarditis caused by coagulase-negative staphylococci can be treated with the same regimens that are employed to treat native valve endocarditis caused by S. aureus.49,80
Some HACEK organisms that cause endocarditis have been found to produce β-lactamase, which makes those organisms resistant to ampicillin. Accordingly, endocarditis caused by HACEK organisms should be treated with a third-generation cephalosporin for 4 weeks [seeTable 5].49,80
The combination of penicillin and gentamicin acts synergistically in vitro to kill Corynebacterium species (diphtheroids) that are susceptible to gentamicin (MIC ≤ 4 µg/ml). Synergism is not achieved against strains that are resistant to gentamicin. For PVE caused by gentamicin-sensitive strains, therapy with intravenous penicillin G (20 million units daily) plus gentamicin (1 mg/kg I.M. or I.V. every 8 hours) is recommended. Vancomycin, which is bactericidal against diphtheroids, is recommended as the initial treatment of diphtheroid PVE, with subsequent treatment determined on the basis of antibiotic-susceptibility test results. Vancomycin is also the drug of choice if the infecting strain is resistant to gentamicin or if the patient is allergic to penicillin.
Endocarditis caused by gram-negative bacilli should be treated with a potent β-lactam antibiotic, such as a third-generation cephalosporin, with or without an aminoglycoside, depending on the susceptibility of, and the published experience with, the specific gram-negative bacillus involved. Effective treatment of endocarditis caused by P. aeruginosa requires the synergistic combination of ticarcillin or piperacillin (either drug at a dosage of 3 g I.V. every 4 hours) and tobramycin (2.7 mg/kg I.M. or I.V. every 8 hours; 8 mg/kg daily) for at least 6 weeks. Because pseudomonal endocarditis is difficult to cure with antibiotics alone and is often complicated by perivalvular abscesses, aggressive early surgical intervention is recommended.91
Endocarditis caused by yeasts is more difficult to cure than most bacterial infections. The best results are achieved by a combination of antifungal chemotherapy and valve debridement or replacement.21,92 Amphotericin B is the drug of choice for initial treatment for most patients. Synergism in vitro between flucytosine and amphotericin B has been reported with some strains of Candida, and their combined use for this life-threatening infection seems warranted. Serum levels of flucytosine should be monitored to reduce toxicity. The drug, which is cleared by the kidney, should not be given to patients with renal failure, because excessive serum concentrations can induce severe bone marrow depression. Long-term (possibly lifelong) suppressive therapy with oral fluconazole may be effective in patients with Candidaendocarditis if they have no intracardiac complications. Similarly, long-term suppressive therapy has been advocated for patients withCandida endocarditis who have responded to combined antifungal and surgical treatment. Some yeast infections can be suppressed (and a few even cured) by treatment with fluconazole or itraconazole. With optimal therapy, the overall survival rate for patients with yeast endocarditis is about 50%.
Dimorphic and mycelial fungi
Endocarditis caused by dimorphic and mycelial fungi is usually fatal despite therapy, with overall survival of less than 10%.21 Aggressive treatment with amphotericin B and valvular surgery will save a few patients. If the fungus is susceptible to flucytosine, combination treatment with this agent plus amphotericin B may be tried.
When blood cultures are negative or culture results are not available, the selection of empirical therapy requires careful consideration of any clinical clues that might suggest the identity of the causative organism. Appropriate special cultures and serologic studies should be pursued in an attempt to identify microorganisms that are difficult to culture.70,71
In the absence of information suggesting a probable cause of native valve SBE, the same treatment as that used for enterococcal endocarditis (i.e., ampicillin plus gentamicin) is advised. If the course is acute and S. aureus is an etiologic consideration, the addition of vancomycin to this regimen is recommended. Because blood cultures rarely remain negative in patients with ABE, such empirical therapy can usually be revised later on the basis of the culture results. The spectrum of microorganisms causing PVE differs from that implicated in native valve endocarditis, and culture-negative PVE should be treated with a combination of vancomycin, gentamicin, and a third-generation cephalosporin. The third-generation cephalosporin is directed against the HACEK organisms, which are important causes of late PVE and can be difficult to isolate from blood cultures. Before embarking on therapy for culture-negative endocarditis, the physician should carefully consider the possibility that the patient does not have endocarditis, but rather has one of the conditions that mimics endocarditis.
Monitoring Clinical Response
A decrease in fever is usually evident within 1 to 7 days after the start of antibiotic therapy.93 Fever that persists during the second week of appropriate antimicrobial therapy suggests the possibility of uncontrolled infection, which could be from antibiotic resistance, intracardiac abscess, metastatic foci of infection, or nosocomial complications of therapy. Thorough evaluation is indicated.78,93 Petechiae and embolic phenomena may occur for several weeks after initiation of recommended treatment; such findings do not necessarily indicate that therapy is ineffective, particularly if other signs indicate that the patient is getting better.
If the patient does not become afebrile within a few days after antibiotic therapy is initiated, repeat blood cultures should be obtained. If the patient is afebrile and otherwise responding well, routine follow-up blood cultures during treatment are usually negative and are not recommended. Similarly, follow-up cultures after the end of antibiotic treatment are unnecessary in most patients, unless nonstandard antimicrobial regimens have been used or there is reason to suspect relapse, such as recurrent fever. Although the ESR slowly returns to normal in the months after effective treatment, this test is not useful for monitoring therapy or as a test of cure. C-reactive protein levels fall within 1 to 2 weeks after starting effective therapy; some clinicians use this test to provide reassurance that treatment is working.94The serum bactericidal test, in which dilutions of the patient's serum are tested for bactericidal activity against the organism causing the endocarditis, is no longer recommended for assessment of therapy in patients receiving standard antimicrobial regimens.80
Operative intervention to debride infected perivalvular tissue or to replace or reconstruct a dysfunctioning valve is important in the management of complicated infective endocarditis that involves either a native or a prosthetic valve.49,51,80,91,95,96 Overall, surgery is indicated in 25% to 40% of patients with infective endocarditis. Several observations have prompted earlier and more frequent surgical intervention in active endocarditis: (1) the mortality of patients undergoing valve surgery early during active endocarditis is not greater than that for patients treated medically or operated upon later (after microbial cure), and in many series is lower91,97; (2) the risk that recurrent endocarditis from the same infecting organism will develop on a prosthesis newly implanted for treatment of endocarditis is very low97; and (3) some intracardiac complications of endocarditis can be corrected only with surgery. For patients with PVE, early aggressive surgical treatment is an essential element of therapy in about 45%.
The currently accepted indications for surgical treatment of active native or prosthetic valve endocarditis have been developed through a retrospective analysis of many cases. Moderate to severe heart failure from valvular dysfunction is the most widely accepted indication for valve replacement and accounts for the majority of decisions to operate. Cardiac surgery for these indications has been convincingly shown to improve survival. Bulky vegetations that obstruct the valve orifice may also produce heart failure and are a compelling indication for surgery. Surgical intervention is also indicated when there is clinical evidence of perivalvular invasion and abscess formation and when infection remains uncontrolled for more than 1 to 3 weeks despite maximal antimicrobial therapy. Fungal endocarditis, especially in patients with intracardiac complications, is commonly treated surgically.92 Similarly, PVE that occurs within a year after valve implantation and is caused by coagulase-negative staphylococci frequently results in perivalvular invasion. Regardless of the causative agent, if there is evidence that PVE is complicated by perivalvular extension of the infection, particularly if the valve is unstable, the PVE should be managed surgically. Because left-sided native valve endocarditis and, in particular, PVE caused by S. aureus are frequently invasive, early surgery should be considered for patients with these infections who do not show prompt and sustained improvement during antibiotic therapy.51,91,96,97,98,99 Endocarditis caused by P. aeruginosa or other gram-negative bacilli that has not responded after 7 to 10 days of maximal antibiotic therapy should be treated surgically. Although patients with native valve endocarditis who experience relapses after appropriate antibiotic therapy often can be cured by repeat courses of antibiotics, surgical treatment improves the outcome in patients with PVE who experience relapse.
Arterial embolization, particularly to the brain, is another factor to be considered when deciding whether to operate. Emboli do not constitute an absolute indication for surgery, because the likelihood of recurrent embolization in any individual patient is highly unpredictable. Furthermore, the incidence of embolization falls rapidly after 1 week of effective antibiotic therapy. Choosing whether to operate therefore requires a consideration of relative risks and benefits in the light of multiple clinical, echocardiographic, and microbiologic observations, including embolization. For example, occurrence of an embolus early in the course, with a residual highly mobile vegetation of more than 10 mm in diameter, would be a relative indication for surgical intervention even if the patient did not have heart failure [seeClinical Presentations, Embolic Phenomena, above].34,100,101,102 Surgery may be beneficial if the vegetations are unusually large and mobile.102 In surgically treated patients with endocarditis, operative mortality is proportional to the degree of preoperative hemodynamic impairment. If surgery is indicated, it should generally be performed promptly. Delaying surgery to administer additional antibiotic therapy in the presence of uncontrolled infection or deteriorating hemodynamic status may result in a less favorable outcome.97 On the other hand, delaying surgery (when hemodynamic status permits) may be desirable in patients who have experienced a neurologic complication.43,103 In patients who have sustained a cerebral infarction, the exacerbation rate for cerebral complications is reduced to 10% if surgery is delayed for at least 15 days and is reduced to below 3% if the delay is 28 days. If cerebral hemorrhage has occurred, cardiac surgery should ideally be delayed for at least 14 days—preferably, 28 days or more.43 Exclusion of a complicating cerebral mycotic aneurysm should be considered in this setting [see Clinical Presentations, Neurologic Manifestations, above].
Although implantation of a prosthetic valve in a patient who is actively abusing intravenous drugs may be lifesaving, endocarditis is likely to recur if the drug abuse continues, and it is more difficult to cure PVE than it is to cure native valve endocarditis. To avoid these problems, tricuspid or pulmonary valvulectomy without valve replacement has been employed in the treatment of right-sided endocarditis. However, in IDUs with isolated right-sided S. aureus endocarditis, surgery should be postponed because most of these cases can be cured medically.19
Anticoagulant therapy in a patient with endocarditis carries the potential risk of causing or worsening hemorrhage in the brain or other sites. However, the benefits of anticoagulation probably outweigh the risks if a strong indication exists, such as atrial fibrillation, cardiomyopathy, mural thrombus, or deep vein thrombosis. Anticoagulant therapy may be carefully administered to patients with endocarditis when it is so indicated.104 In patients with prosthetic valves who require long-term warfarin therapy, such therapy should be continued during treatment of endocarditis unless there are specific contraindications. The prothrombin time should be carefully maintained in the lower therapeutic range, at an international normalized ratio (INR) of 2.0 to 3.0. Anticoagulation should be reversed immediately in the event of CNS complications, especially intracranial hemorrhage. Anticoagulation should be interrupted for 1 to 2 weeks after an acute embolic stroke. Heparin therapy should be avoided during active endocarditis if at all possible.
There is no evidence in humans that antithrombotic therapy can prevent arterial embolization of vegetations or otherwise improve the response of endocarditis to standard treatment regimens.
Antibiotics can prevent endocarditis in animal models if given before experimentally induced bacteremias, and may do so in humans if given before medical, dental, and surgical procedures that are known to cause bacteremia. However, this has not been proved. Current guidelines support the administration of prophylactic antibiotic to patients who are at higher risk for endocarditis than the general population when they undergo procedures likely to lead to bacteremia with organisms that commonly cause endocarditis.105,106,107 The appropriate selection of patients for this intervention remains controversial. Such selections should be made on the basis of an assessment of the combined risk of the underlying cardiac condition, the likelihood that the procedure will cause endocarditis (which is always low), and the potential morbidity and mortality86 [see Tables 6, 7 and 8]. Patients who are at the highest risk for endocarditis are those with a history of previous infective endocarditis, prosthetic valves, cyanotic congenital heart disease, or surgically constructed shunts or conduits. Prophylaxis is merited in all such cases.
Table 6 Cardiac Conditions and Endocarditis Prophylaxis83
Table 7 Procedures and Endocarditis Prophylaxis83
Table 8 Prophylactic Antibiotics Recommended for Adults Undergoing Oral Cavity, Respiratory Tract, or Esophageal Procedures83
Table 9 Prophylactic Antibiotics Recommended for Adults Undergoing Gastrointestinal,*Biliary, or Genitourinary Procedures83
Prophylaxis is recommended for patients at high risk for endocarditis who undergo procedures that involve the oral cavity, the respiratory tract, or the esophagus and are likely to cause streptococcal bacteremia. Prophylaxis is also indicated for patients at high risk for endocarditis who undergo procedures that involve the genitourinary tract, the GI tract distal to the stomach, or the biliary tract and are likely to cause bacteremia with enterococci [see Table 7]. For patients who are at unusually high risk for endocarditis, some physicians elect to use prophylaxis even for procedures carrying low likelihood of bacteremia (for which prophylaxis would generally not be recommended).
Antibiotic regimens recommended for prophylaxis in high-risk patients have been designed to cover streptococci or enterococci and are assigned in accordance with the anticipated bacteremic organism [see Tables 8 and 9]. When urinary tract manipulation is anticipated, an evaluation for urinary tract infection should be undertaken, and infection, if confirmed, should be treated before the manipulation. If procedures must be performed on infected tissues, prophylaxis is directed against the likely pathogen causing the infection. Such prophylaxis would include use of an antistaphylococcal penicillin or a first-generation cephalosporin when the infection involves skin and adjacent soft tissue, joints, or bone.
If a patient who is at risk for endocarditis is receiving or has recently received one of the agents recommended for prophylaxis, it is prudent to choose for prophylaxis an antibiotic from a different class, rather than increase the dosage of the current regimen. The dosage of penicillin used to prevent recurrences of acute rheumatic fever is insufficient to prevent bacterial endocarditis. Moreover, patients taking oral penicillins or cephalosporins may harbor relatively penicillin-resistant strains of bacteria in their oral cavity. Prophylaxis with clarithromycin or clindamycin is suggested when relatively penicillin-resistant streptococci are anticipated.
Most cases of endocarditis are not related to defined events or procedures. It is likely that these infections arise from bacteremias associated with minor unrecognized infections or ordinary activities such as vigorous chewing in patients with gingival disease or use of oral irrigating jets. In patients with vulnerable cardiac lesions, maintaining optimal dental health is essential if the risk of endocarditis is to be minimized. Dental evaluation and treatment, including extractions and necessary restorative work, should be performed under appropriate antibiotic prophylaxis several weeks before the insertion of prosthetic heart valves.
Overall survival for patients with infective endocarditis is about 75% to 80%.2,44,108 This rate is little better than that in the 1950s and needs to be improved.108 The outlook depends largely on the promptness of diagnosis and initiation of therapy, the nature of the infecting organism, the presence of comorbid conditions, and the development of cardiac and neurologic complications.44,50 Survival is above 90% when the infecting organism is a viridans-type streptococcus or S. bovis but is only about 50% in non-IDUs infected with S. aureus. Although outcomes are worse in cases of early PVE, survival rates for patients with late PVE appear to be similar to those for patients with native valve infection.5,6,89 With the incorporation of earlier surgical intervention for patients with endocarditis, cardiac failure is less strongly associated with increased mortality than it once was. However, major CNS complications and uncontrolled infection, especially abscess formation, continue to be associated with increased mortality.41 The short-term outlook is relatively good for IDUs with right-sided endocarditis.19 Endocarditis caused by gram-negative bacilli other than HACEK organisms or caused by yeasts is difficult to cure despite optimal antimicrobial therapy and aggressive surgical intervention. Endocarditis caused by mycelial fungi is usually fatal.21
Figures 2 and 3 Carol Donner.
Editors: Dale, David C.; Federman, Daniel D.