Adolf W. Karchmer
The prototypic lesion of infective endocarditis, the vegetation(Fig. 25-1), is a mass of platelets, fibrin, microcolonies of microorganisms, and scant inflammatory cells. Infection most commonly involves heart valves (either native or prosthetic) but may also occur on the low-pressure side of a ventricular septal defect, on the mural endocardium where it is damaged by aberrant jets of blood or foreign bodies, or on intracardiac devices themselves. The analogous process involving arteriovenous shunts, arterioarterial shunts (patent ductus arteriosus), or a coarctation of the aorta is called infective endarteritis.
Vegetations (arrows) due to viridans streptococcal endocarditis involving the mitral valve.
Endocarditis may be classified according to the temporal evolution of disease, the site of infection, the cause of infection, or a predisposing risk factor such as injection drug use. While each classification criterion provides therapeutic and prognostic insight, none is sufficient alone. Acute endocarditis is a hectically febrile illness that rapidly damages cardiac structures, hematogenously seeds extracardiac sites, and, if untreated, progresses to death within weeks. Subacute endocarditis follows an indolent course; causes structural cardiac damage only slowly, if at all; rarely metastasizes; and is gradually progressive unless complicated by a major embolic event or ruptured mycotic aneurysm.
In developed countries, the incidence of endocarditis ranges from 2.6 to 7 cases per 100,000 population per year and has remained relatively stable during recent decades. While congenital heart diseases remain a constant predisposition, predisposing conditions in developed countries have shifted from chronic rheumatic heart disease (which remains a common predisposition in developing countries) to illicit IV drug use, degenerative valve disease, and intracardiac devices. The incidence of endocarditis is notably increased among the elderly. In developed countries, 30–35% of cases of native valve endocarditis (NVE) are associated with health care, and 16–30% of all cases of endocarditis involve prosthetic valves. The risk of prosthesis infection is greatest during the first 6–12 months after valve replacement; gradually declines to a low, stable rate thereafter; and is similar for mechanical and bioprosthetic devices.
Although many species of bacteria and fungi cause sporadic episodes of endocarditis, a few bacterial species cause the majority of cases (Table 25-1). Because of their different portals of entry, the pathogens involved vary somewhat with the clinical types of endocarditis. The oral cavity, skin, and upper respiratory tract are the respective primary portals for the viridans streptococci, staphylococci, and HACEK organisms (Haemophilus, Actinobacillus, Cardiobacterium, Eikenella, and Kingella; Haemophilus aphrophilus and Actinobacillus actinomycetemcomitans have been reclassified into the genus Aggregatibacter). Streptococcus gallolyticus (formerly S. bovis) originates from the gastrointestinal tract, where it is associated with polyps and colonic tumors, and enterococci enter the bloodstream from the genitourinary tract. Health care–associated NVE, commonly caused by Staphylococcus aureus, coagulase-negative staphylococci (CoNS), and enterococci, has a nosocomial onset (55%) or a community onset (45%) in patients who have had extensive contact with the health care system over the preceding 90 days. Endocarditis complicates 6–25% of episodes of catheter-associated S. aureus bacteremia; the higher rates are detected by careful transesophageal echocardiography (TEE) screening (see “Echocardiography,” later).
ORGANISMS CAUSING MAJOR CLINICAL FORMS OF ENDOCARDITIS
Prosthetic valve endocarditis (PVE) arising within 2 months of valve surgery is generally nosocomial, the result of intraoperative contamination of the prosthesis or a bacteremic postoperative complication. This nosocomial origin is reflected in the primary microbial causes: S. aureus, CoNS, facultative gram-negative bacilli, diphtheroids, and fungi. The portals of entry and organisms causing cases beginning >12 months after surgery are similar to those in community-acquired NVE. PVE due to CoNS that presents 2–12 months after surgery often represents delayed-onset nosocomial infection. Regardless of the time of onset after surgery, at least 68–85% of CoNS strains that cause PVE are resistant to methicillin.
Transvenous pacemaker– or implanted defibrillator– associated endocarditis is usually nosocomial. The majority of episodes occur within weeks of implantation or generator change and are caused by S. aureus or CoNS, both of which are commonly resistant to methicillin.
Endocarditis occurring among injection drug users, especially that involving the tricuspid valve, is commonly caused by S. aureus, many strains of which are resistant to methicillin. Left-sided valve infections in addicts have a more varied etiology. In addition to the usual causes of endocarditis, these cases are caused by Pseudomonas aeruginosa and Candida species and sporadically by unusual organisms such as Bacillus, Lactobacillus, and Corynebacterium species. Polymicrobial endocarditis occurs among injection drug users. HIV infection in drug users does not significantly influence the causes of endocarditis.
From 5% to 15% of patients with endocarditis have negative blood cultures; in one-third to one-half of these cases, cultures are negative because of prior antibiotic exposure. The remainder of these patients are infected by fastidious organisms, such as nutritionally variant organisms (now designated Granulicatella and Abiotrophia species), HACEK organisms, Coxiella burnetii, and Bartonella species. Some fastidious organisms occur in characteristic geographic settings (e.g., C. burnetii and Bartonella species in Europe, Brucella species in the Middle East). Tropheryma whipplei causes an indolent, culture-negative, afebrile form of endocarditis.
The endothelium, unless damaged, is resistant to infection by most bacteria and to thrombus formation. Endothelial injury (e.g., at the site of impact of high-velocity blood jets or on the low-pressure side of a cardiac structural lesion) allows either direct infection by virulent organisms or the development of an uninfected platelet-fibrin thrombus—a condition called nonbacterial thrombotic endocarditis (NBTE). The thrombus subsequently serves as a site of bacterial attachment during transient bacteremia. The cardiac conditions most commonly resulting in NBTE are mitral regurgitation, aortic stenosis, aortic regurgitation, ventricular septal defects, and complex congenital heart disease. NBTE also arises as a result of a hypercoagulable state; this phenomenon gives rise to the clinical entity of marantic endocarditis(uninfected vegetations seen in patients with malignancy and chronic diseases) and to bland vegetations complicating systemic lupus erythematosus and the antiphospholipid antibody syndrome.
Organisms that cause endocarditis generally enter the bloodstream from mucosal surfaces, the skin, or sites of focal infection. Except for more virulent bacteria (e.g., S. aureus) that can adhere directly to intact endothelium or exposed subendothelial tissue, microorganisms in the blood adhere at sites of NBTE. If resistant to the bactericidal activity of serum and the microbicidal peptides released locally by platelets, the organisms proliferate and induce platelet deposition and a procoagulant state at the site by eliciting tissue factor from the endothelium or, in the case of S. aureus, from monocytes as well. Fibrin deposition combines with platelet aggregation and microorganism proliferation to generate an infected vegetation. The organisms that commonly cause endocarditis have surface adhesin molecules, collectively called microbial surface components recognizing adhesin matrix molecules (MSCRAMMs), that mediate adherence to NBTE sites or injured endothelium. Fibronectin-binding proteins present on many gram-positive bacteria, clumping factor (a fibrinogen- and fibrin-binding surface protein) on S. aureus, and glucans or FimA (a member of the family of oral mucosal adhesins) on streptococci facilitate adherence. Fibronectin-binding proteins are required for S. aureus invasion of intact endothelium; thus these surface proteins may facilitate infection of previously normal valves. In the absence of host defenses, organisms enmeshed in the growing platelet-fibrin vegetation proliferate to form dense microcolonies. Organisms deep in vegetations are metabolically inactive (nongrowing) and relatively resistant to killing by antimicrobial agents. Proliferating surface organisms are shed into the bloodstream continuously.
The pathophysiologic consequences and clinical manifestations of endocarditis—other than constitutional symptoms, which probably result from cytokine production—arise from damage to intracardiac structures; embolization of vegetation fragments, leading to infection or infarction of remote tissues; hematogenous infection of sites during bacteremia; and tissue injury due to the deposition of circulating immune complexes or immune responses to deposited bacterial antigens.
The clinical syndrome of infective endocarditis is highly variable and spans a continuum between acute and sub-acute presentations. NVE (whether acquired in the community or in association with health care), PVE, and endocarditis due to injection drug use share clinical and laboratory manifestations (Table 25-2). The causative microorganism is primarily responsible for the temporal course of endocarditis. β-Hemolytic streptococci, S. aureus, and pneumococci typically result in an acute course, although S. aureus occasionally causes subacute disease. Endocarditis caused by Staphylococcus lugdunensis (a coagulase-negative species) or by enterococci may present acutely. Subacute endocarditis is typically caused by viridans streptococci, enterococci, CoNS, and the HACEK group. Endocarditis caused by Bartonella species, T. whipplei, or C. burnetii is exceptionally indolent.
CLINICAL AND LABORATORY FEATURES OF INFECTIVE ENDOCARDITIS
The clinical features of endocarditis are nonspecific. However, these symptoms in a febrile patient with valvular abnormalities or a behavior pattern that predisposes to endocarditis (e.g., injection drug use) suggest the diagnosis, as do bacteremia with organisms that frequently cause endocarditis, otherwise-unexplained arterial emboli, and progressive cardiac valvular incompetence. In patients with subacute presentations, fever is typically low grade and rarely exceeds 39.4°C (103°F); in contrast, temperatures of 39.4°–40°C (103°–104°F) are often noted in acute endocarditis. Fever may be blunted or absent in patients who are elderly or severely debilitated or who have marked cardiac or renal failure.
Although heart murmurs are usually indicative of the predisposing cardiac pathology rather than of endocarditis, valvular damage and ruptured chordae may result in new regurgitant murmurs. In acute endocarditis involving a normal valve, murmurs may be absent initially but ultimately are detected in 85% of cases. Congestive heart failure (CHF) develops in 30–40% of patients; it is usually a consequence of valvular dysfunction but occasionally is due to endocarditis-associated myocarditis or an intracardiac fistula. Heart failure due to aortic valve dysfunction progresses more rapidly than does that due to mitral valve dysfunction. Extension of infection beyond valve leaflets into adjacent annular or myocardial tissue results in perivalvular abscesses, which in turn may cause intracardiac fistulae with new murmurs. Abscesses may burrow from the aortic valve annulus through the epicardium, causing pericarditis, or into the upper ventricular septum, where they may interrupt the conduction system, leading to varying degrees of heart block. Perivalvular abscesses arising from the mitral valve rarely interrupt conduction pathways near the atrioventricular node or in the proximal bundle of His. Emboli to a coronary artery occur in 2% of patients and may result in myocardial infarction.
The classic nonsuppurative peripheral manifestations of subacute endocarditis are related to the duration of infection and, with early diagnosis and treatment, have become infrequent. In contrast, septic embolization mimicking some of these lesions (subungual hemorrhage, Osler’s nodes) is common in patients with acute S. aureus endocarditis (Fig. 25-2). Musculoskeletal pain usually remits promptly with treatment but must be distinguished from focal metastatic infections (e.g., spondylodiscitis), which may complicate 10–15% of cases. Hematogenously seeded focal infection is most often clinically evident in the skin, spleen, kidneys, skeletal system, and meninges. Arterial emboli are clinically apparent in up to 50% of patients. Endocarditis caused by S. aureus, vegetations >10 mm in diameter (as measured by echocardiography), and infection involving the mitral valve are independently associated with an increased risk of embolization. Emboli occurring late, during, or after effective therapy do not in themselves constitute evidence of failed antimicrobial treatment. Cerebrovascular emboli presenting as strokes or occasionally as encephalopathy complicate 15–35% of cases of endocarditis. One-half of these events precede the diagnosis of endocarditis. The frequency of stroke is 8 per 1000 patient-days during the week prior to diagnosis; the figure falls to 4.8 and 1.7 per 1000 patient-days during the first and second weeks of effective antimicrobial therapy, respectively. This decline exceeds that which can be attributed to change in vegetation size. Only 3% of strokes occur after 1 week of effective therapy. Other neurologic complications include aseptic or purulent meningitis, intracranial hemorrhage due to hemorrhagic infarcts or ruptured mycotic aneurysms, and seizures. (Mycotic aneurysms are focal dilations of arteries occurring at points in the artery wall that have been weakened by infection in the vasa vasorum or where septic emboli have lodged.) Microabscesses in brain and meninges occur commonly in S. aureusendocarditis; surgically drainable intracerebral abscesses are infrequent.
Septic emboli with hemorrhage and infarction due to acute Staphylococcus aureus endocarditis. (Used with permission of L. Baden.)
Immune complex deposition on the glomerular basement membrane causes diffuse hypocomplementemic glomerulonephritis and renal dysfunction, which typically improve with effective antimicrobial therapy. Embolic renal infarcts cause flank pain and hematuria but rarely cause renal dysfunction.
Manifestations of specific predisposing conditions
Almost 50% of endocarditis cases associated with injection drug use are limited to the tricuspid valve and present with fever but with faint or no murmur. In 75% of cases, septic emboli cause cough, pleuritic chest pain, nodular pulmonary infiltrates, or occasionally pyopneumothorax. Infection of the aortic or mitral valves on the left side of the heart presents with the typical clinical features of endocarditis.
Health care–associated endocarditis has typical manifestations if it is not associated with a retained intracardiac device or masked by the symptoms of concurrent comorbid illness. Transvenous pacemaker– or implanted defibrillator–associated endocarditis may be associated with obvious or cryptic generator pocket infection and results in fever, minimal murmur, and pulmonary symptoms due to septic emboli.
Late-onset PVE presents with typical clinical features. In cases arising within 60 days of valve surgery (early onset), typical symptoms may be obscured by comorbidity associated with recent surgery. In both early-onset and more delayed presentations, paravalvular infection is common and often results in partial valve dehiscence, regurgitant murmurs, CHF, or disruption of the conduction system.
The Duke criteria
The diagnosis of infective endocarditis is established with certainty only when vegetations are examined histologically and microbiologically. Nevertheless, a highly sensitive and specific diagnostic schema—known as the Duke criteria—has been developed on the basis of clinical, laboratory, and echocardiographic findings (Table 25-3). Documentation of two major criteria, of one major criterion and three minor criteria, or of five minor criteria allows a clinical diagnosis of definite endocarditis. The diagnosis of endocarditis is rejected if an alternative diagnosis is established, if symptoms resolve and do not recur with ≤4 days of antibiotic therapy, or if surgery or autopsy after ≤4 days of antimicrobial therapy yields no histologic evidence of endocarditis. Illnesses not classified as definite endocarditis or rejected as such are considered cases of possible infective endocarditis when either one major criterion and one minor criterion or three minor criteria are fulfilled. Requiring the identification of clinical features of endocarditis for classification as possible infective endocarditis increases the specificity of the schema without significantly reducing its sensitivity.
THE DUKE CRITERIA FOR THE CLINICAL DIAGNOSIS OF INFECTIVE ENDOCARDITISa
The roles of bacteremia and echocardiographic findings in the diagnosis of endocarditis are emphasized in the Duke criteria. The requirement for multiple positive blood cultures over time is consistent with the continuous low-density bacteremia characteristic of endocarditis. Among patients with untreated endocarditis who ultimately have a positive blood culture, 95% of all blood cultures are positive. The diagnostic criteria attach significance to the species of organism isolated from blood cultures. To fulfill a major criterion, the isolation of an organism that causes both endocarditis and bacteremia in the absence of endocarditis (e.g., S. aureus, enterococci) must take place repeatedly (i.e., persistent bacteremia) and in the absence of a primary focus of infection. Organisms that rarely cause endocarditis but commonly contaminate blood cultures (e.g., diphtheroids, CoNS) must be isolated repeatedly if their isolation is to serve as a major criterion.
Isolation of the causative microorganism from blood cultures is critical for diagnosis, determination of antimicrobial susceptibility, and planning of treatment. In the absence of prior antibiotic therapy, three 2-bottle blood culture sets, separated from one another by at least 1 h, should be obtained from different venipuncture sites over 24 h. If the cultures remain negative after 48–72 h, two or three additional blood culture sets should be obtained, and the laboratory should be consulted for advice regarding optimal culture techniques. Pending culture results, empirical antimicrobial therapy should be withheld initially from hemodynamically stable patients with suspected subacute endocarditis, especially those who have received antibiotics within the preceding 2 weeks; thus, if necessary, additional blood culture sets can be obtained without the confounding effect of empirical treatment. Patients with acute endocarditis or with deteriorating hemodynamics who may require urgent surgery should be treated empirically immediately after three sets of blood cultures are obtained over several hours.
Serologic tests can be used to implicate causally some organisms that are difficult to recover by blood culture: Brucella, Bartonella, Legionella, Chlamydophila psittaci, and C. burnetii. Pathogens can also be identified in vegetations by culture, microscopic examination with special stains (i.e., the periodic acid–Schiff stain for T. whipplei), or direct fluorescence antibody techniques and by the use of polymerase chain reaction (PCR) to recover unique microbial DNA or 16S rRNA that, when sequenced, allows identification of organisms.
Echocardiography allows anatomic confirmation of infective endocarditis, sizing of vegetations, detection of intracardiac complications, and assessment of cardiac function (Fig. 25-3). Transthoracic echocardiography (TTE) is noninvasive and exceptionally specific; however, it cannot image vegetations <2 mm in diameter, and in 20% of patients it is technically inadequate because of emphysema or body habitus. TTE detects vegetations in only 65% of patients with definite clinical endocarditis. Moreover, TTE is not adequate for evaluating prosthetic valves or detecting intracardiac complications. TEE is safe and detects vegetations in >90% of patients with definite endocarditis; nevertheless, initial studies may be false-negative in 6–18% of endocarditis patients. When endocarditis is likely, a negative TEE result does not exclude the diagnosis but rather warrants repetition of the study in 7–10 days. TEE is the optimal method for the diagnosis of PVE or the detection of myocardial abscess, valve perforation, or intracardiac fistulae.
Imaging of a mitral valve infected with Staphylococcus aureus by low-esophageal four-chamber-view transesophageal echocardiography (TEE). A. Two-dimensional echocardiogram showing a large vegetation with an adjacent echo-lucent abscess cavity. B. Color-flow Doppler image showing severe mitral regurgitation through both the abscess-fistula and the central valve orifice. A, abscess; A-F, abscess-fistula; L, valve leaflets; LA, left atrium; LV, left ventricle; MR, mitral central valve regurgitation; RV, right ventricle; veg, vegetation. (With permission of Andrew Burger, MD)
Experts favor echocardiographic evaluation of all patients with a clinical diagnosis of endocarditis; however, the test should not be used to screen patients with a low probability of endocarditis (e.g., patients with unexplained fever). An American Heart Association approach to the use of echocardiography for evaluation of patients with suspected endocarditis is illustrated in Fig. 25-4.
The diagnostic use of transesophageal and transtracheal echocardiography (TEE and TTE, respectively). †High initial patient risk for endocarditis as listed in Table 25-8 or evidence of intracardiac complications (new regurgitant murmur, new electrocardiographic conduction changes, or congestive heart failure). *High-risk echocardiographic features include large vegetations, valve insufficiency, paravalvular infection, or ventricular dysfunction. Rx indicates initiation of antibiotic therapy. (Reproduced with permission from Diagnosis and Management of Infective Endocarditis and Its Complications. Circulation 98:2936, 1998. © 1998 American Heart Association.)
HIGH-RISK CARDIAC LESIONS FOR WHICH ENDOCARDITIS PROPHYLAXIS IS ADVISED BEFORE DENTAL PROCEDURES
Many laboratory studies that are not diagnostic—i.e., complete blood count, creatinine determination, liver function tests, chest radiography, and electrocardiography—are nevertheless important in the management of patients with endocarditis. The erythrocyte sedimentation rate, C-reactive protein level, and circulating immune complex titer are commonly increased in endocarditis (Table 25-2). Cardiac catheterization is useful primarily to assess coronary artery patency in older individuals who are to undergo surgery for endocarditis.
TREATMENT Infective Endocarditis
ANTIMICROBIAL THERAPY It is difficult to eradicate bacteria from the vegetation because local host defenses are deficient and because the largely non-growing, metabolically inactive bacteria are less easily killed by antibiotics. To cure endocarditis, all bacteria in the vegetation must be killed; therefore, therapy must be bactericidal and prolonged. Antibiotics are generally given parenterally to achieve serum concentrations that, through passive diffusion, lead to effective concentrations in the depths of the vegetation. To select effective therapy requires knowledge of the susceptibility of the causative microorganisms. The decision to initiate treatment empirically must balance the need to establish a microbiologic diagnosis against the potential progression of disease or the need for urgent surgery (see “Blood Cultures,” earlier). Simultaneous infection at other sites (such as meningitis), allergies, end-organ dysfunction, interactions with concomitant medications, and risks of adverse events must be considered in the selection of therapy.
Although given for several weeks longer, the regimens recommended for the treatment of endocarditis involving prosthetic valves (except for staphylococcal infections) are similar to those used to treat NVE (Table 25-4). Recommended doses and durations of therapy should be adhered to unless alterations are required by end-organ dysfunction or adverse events.
ANTIBIOTIC TREATMENT FOR INFECTIVE ENDOCARDITIS CAUSED BY COMMON ORGANISMSa
Streptococci Optimal therapy for streptococcal endocarditis is based on the minimal inhibitory concentration (MIC) of penicillin for the causative isolate (Table 25-4). The 2-week penicillin/gentamicin or ceftriaxone/gentamicin regimens should not be used to treat complicated NVE or PVE. The regimen recommended for relatively penicillin-resistant streptococci is advocated for treatment of group B, C, or G streptococcal endocarditis. Nutritionally variant organisms (Granulicatella or Abiotrophia species) and Gemella morbillorum are treated with the regimen for moderately penicillin-resistant streptococci, as is PVE caused by these organisms or by streptococci with a penicillin MIC of >0.1 µg/mL (Table 25-4).
Enterococci Enterococci are resistant to oxacillin, nafcillin, and the cephalosporins and are only inhibited—not killed—by penicillin, ampicillin, teicoplanin (not available in the United States), and vancomycin. To kill enterococci requires the synergistic interaction of a cell wall–active antibiotic (penicillin, ampicillin, vancomycin, or teicoplanin) that is effective at achievable serum concentrations and an aminoglycoside (gentamicin or streptomycin) to which the isolate does not exhibit high-level resistance. An isolate’s resistance to cell wall–active agents or its ability to replicate in the presence of gentamicin at ≥500 µg/mL or streptomycin at 1000–2000 µg/mL—a phenomenon called high-level aminoglycoside resistance—indicates that the ineffective antimicrobial agent cannot participate in the interaction to produce killing. High-level resistance to gentamicin predicts that tobramycin, netilmicin, amikacin, and kanamycin also will be ineffective. In fact, even when enterococci are not highly resistant to gentamicin, it is difficult to predict the ability of these other aminoglycosides to participate in synergistic killing; consequently, they should not in general be used to treat enterococcal endocarditis. High concentrations of ampicillin plus ceftriaxone or cefotaxime, by expanded binding of penicillin-binding proteins, kill E. faecalis in vitro and in animal models of endocarditis.
Enterococci causing endocarditis must be tested for high-level resistance to streptomycin and gentamicin, β-lactamase production, and susceptibility to penicillin and ampicillin (MIC, <8 µg/mL) and to vancomycin (MIC, ≤4 µg/mL). If the isolate produces β-lactamase, ampicillin/sulbactam or vancomycin can be used as the cell wall–active component; if the penicillin/ampicillin MIC is ≥8 µg/mL, vancomycin can be considered; and if the vancomycin MIC is ≥8 µg/mL, penicillin or ampicillin can be considered. In the absence of high-level resistance, gentamicin or streptomycin should be used as the aminoglycoside (Table 25-4). If there is high-level resistance to both these drugs, no aminoglycoside should be given; instead, an 8- to 12-week course of a single cell wall–active agent—or, for E. faecalis, high doses of ampicillin combined with ceftriaxone or cefotaxime—is suggested. If this alternative therapy fails or the isolate is resistant to all of the commonly used agents, surgical treatment is advised. The role of newer agents potentially active against multidrug-resistant enterococci (quinupristin/dalfopristin [E. faeciumonly], linezolid, and daptomycin) in the treatment of endocarditis has not been established. Although the dose of gentamicin used to achieve bactericidal synergy in treating enterococcal endocarditis is smaller than that used in standard therapy, nephrotoxicity is not uncommon during treatment for 4–6 weeks. Regimens in which the aminoglycoside component is discontinued at 2–3 weeks because of toxicity have been curative. Thus, discontinuation of the aminoglycoside is recommended when nephrotoxicity develops in patients who have responded satisfactorily to therapy. Alternatively, the ampicillin-ceftriaxone regimen can be used to treat E. faecalis endocarditis if nephrotoxicity develops or is exceptionally threatening.
Staphylococci The regimens used to treat staphylococcal endocarditis (Table 25-4) are based not on coagulase production but rather on the presence or absence of a prosthetic valve or foreign device, the native valve(s) involved, and the susceptibility of the isolate to penicillin, methicillin, and vancomycin. All staphylococci are considered penicillin resistant until shown not to produce penicillinase. Similarly, methicillin resistance has become so prevalent among staphylococci that therapy should be initiated with a regimen for methicillin-resistant organisms and subsequently revised if the strain proves to be susceptible to methicillin. The addition of 3–5 days of gentamicin (if the isolate is susceptible) to a β-lactam antibiotic to enhance therapy for native mitral or aortic valve endocarditis has been optional. While the addition of gentamicin minimally hastens eradication of bacteremia, it does not improve survival rates, and even abbreviated gentamicin therapy may be associated with nephrotoxicity and thus is not recommended. Gentamicin generally is not added to the vancomycin regimen in this setting.
For treatment of endocarditis caused by methicillin-resistant S. aureus (MRSA), vancomycin dosing to achieve trough concentrations of 15–20 µg/mL is recommended, with the recognition that this regimen may be associated with nephrotoxicity. Although resistance to vancomycin among staphylococci is rare, reduced vancomycin susceptibility among MRSA strains is increasingly encountered. Isolates with a vancomycin MIC of 4–16 µg/mL have intermediate susceptibility and are referred to as vancomycin-intermediate S. aureus (VISA). Isolates with an MIC of 2 µg/mL may harbor subpopulations with higher MICs. These isolates, called hetero-resistant VISA (hVISA), are not detectable by routine susceptibility testing. Because of the pharmacokinetics/pharmacodynamics of vancomycin, killing of MRSA with a vancomycin MIC of 2 µg/mL is unpredictable even with aggressive vancomycin dosing. Although not approved by the U.S. Food and Drug Administration (FDA), daptomycin (6 mg/kg [or, as some experts prefer, 8–10 mg/kg] IV once daily) has been recommended as an alternative to vancomycin, particularly for endocarditis caused by VISA, hVISA, and isolates with a vancomycin MIC of 2 µg/mL. These isolates should be tested to document daptomycin susceptibility. Treatment of endocarditis in which bacteremia persists despite this therapy is beyond the scope of this chapter and requires consultation with an infectious disease specialist. The efficacy of linezolid for left-sided MRSA endocarditis has not been established.
Methicillin-susceptible S. aureus endocarditis that is uncomplicated and limited to the tricuspid or pulmonic valve—a condition occurring almost exclusively in injection drug users—can often be treated with a 2-week course that combines oxacillin or nafcillin (but not vancomycin) with gentamicin. Patients with prolonged fever (≥5 days) during therapy or multiple septic pulmonary emboli should receive standard therapy. Right-sided endocarditis caused by MRSA is treated for 4 weeks with a standard vancomycin regimen or with daptomycin (6 mg/kg as a single daily dose).
Staphylococcal PVE is treated for 6–8 weeks with a multidrug regimen. Rifampin is an essential component because it kills staphylococci that are adherent to foreign material in a biofilm. Two other agents (selected on the basis of susceptibility testing) are combined with rifampin to prevent in vivo emergence of resistance. Because many staphylococci (particularly MRSA and S. epidermidis) are resistant to gentamicin, susceptibility to gentamicin or an alternative agent should be established before rifampin treatment is begun. If the isolate is resistant to gentamicin, then another aminoglycoside, a fluoroquinolone (chosen on the basis of susceptibility), or another active agent should be substituted for gentamicin.
Other Organisms In the absence of meningitis, endocarditis caused by S. pneumoniae with a penicillin MIC of ≤1 μg/mL can be treated with IV penicillin (4 million units every 4 h), ceftriaxone (2 g/d as a single dose), or cefotaxime (at a comparable dosage). Infection caused by pneumococcal strains with a penicillin MIC of ≤2 µg/mL should be treated with vancomycin. Until the strain’s susceptibility to penicillin is established, therapy should consist of vancomycin plus ceftriaxone, especially if concurrent meningitis is suspected. P. aeruginosa endocarditis is treated with an antipseudomonal penicillin (ticarcillin or piperacillin) and high doses of tobramycin (8 mg/kg per day in three divided doses). Endocarditis caused by Enterobacteriaceae is treated with a potent β-lactam antibiotic plus an aminoglycoside. Corynebacterial endocarditis is treated with penicillin plus an aminoglycoside (if the organism is susceptible to the aminoglycoside) or with vancomycin, which is highly bactericidal for most strains. Therapy for Candida endocarditis consists of amphotericin B plus flucytosine and early surgery; long-term (if not indefinite) suppression with an oral azole is advised. Caspofungin treatment of Candida endocarditis has been effective in sporadic cases; nevertheless, the role of echinocandins in this setting has not been established.
Empirical Therapy In the design and execution of therapy without culture data (i.e., before culture results are known or when cultures are negative), clinical clues (e.g., site of infection, patient’s predispositions), as well as epidemiologic clues to etiology must be considered. Thus, empirical therapy for acute endocarditis in an injection drug user should cover MRSA and gram-negative bacilli. Treatment with vancomycin plus gentamicin, initiated immediately after blood is obtained for cultures, covers these as well as many other potential causes. Similarly, treatment of health care–associated endocarditis must cover MRSA. In the treatment of culture-negative episodes, marantic endocarditis must be excluded and fastidious organisms sought by serologic testing. In the absence of prior antibiotic therapy, it is unlikely that S. aureus, CoNS, or enterococcal infection will present with negative blood cultures; thus, in this situation, recommended empirical therapy targets not these organisms but rather nutritionally variant organisms, the HACEK group, and Bartonella species. Pending the availability of diagnostic data, blood culture–negative subacute NVE is treated either with ampicillin-sulbactam (12 g every 24 h) or with ceftriaxone plus gentamicin; doxycycline (100 mg twice daily) is added for Bartonella coverage. Vancomycin, gentamicin, cefepime, and rifampin should be used if prosthetic valves in place for ≤1 year are involved. Empirical therapy for infected prosthetic valves in place for >1 year is similar to that for culture-negative PVE. If negative cultures have been confounded by prior antibiotic administration, broader empirical therapy may be indicated, with particular attention to pathogens likely to be inhibited by the specific prior therapy.
Outpatient Antimicrobial Therapy Fully compliant patients who have sterile blood cultures, no fever, and no clinical or echocardiographic findings that suggest an impending complication may complete therapy as outpatients. Careful follow-up and a stable home setting are necessary, as are predictable IV access and use of antimicrobial agents that are stable in solution.
Monitoring Antimicrobial Therapy The serum bactericidal titer—the highest dilution of the patient’s serum during therapy that kills 99.9% of the standard inoculum of the infecting organism—is no longer recommended for assessment of standard regimens. However, in the treatment of endocarditis caused by unusual organisms, this measurement may provide a patient-specific assessment of in vivo antibiotic effect. Serum concentrations of aminoglycosides and vancomycin should be monitored.
Antibiotic toxicities, including allergic reactions, occur in 25–40% of patients and commonly arise during the third week of therapy. Blood tests to detect renal, hepatic, and hematologic toxicity should be performed periodically.
Blood cultures should be repeated daily until sterile, rechecked if there is recrudescent fever, and performed again 4–6 weeks after therapy to document cure. Blood cultures become sterile within 2 days after the start of appropriate therapy when infection is caused by viridans streptococci, enterococci, or HACEK organisms. In S. aureus endocarditis, β-lactam therapy results in sterile cultures in 3–5 days, whereas with MRSA endocarditis positive cultures may persist for 7–9 days with vancomycin treatment. MRSA bacteremia persisting despite an adequate dosage of vancomycin may indicate infection due to a strain with reduced vancomycin susceptibility and therefore may point to a need for alternative therapy. When fever persists for 7 days despite appropriate antibiotic therapy, patients should be evaluated for paravalvular abscess, extracardiac abscesses (spleen, kidney), or complications (embolic events). Recrudescent fever raises the question of these complications but also of drug reactions or complications of hospitalization. Vegetations become smaller with effective therapy; however, 3 months after cure, 50% are unchanged and 25% are slightly larger.
SURGICAL TREATMENT Intracardiac and central nervous system complications of endocarditis are important causes of morbidity and death. In some cases, effective treatment for these complications requires surgery. The indications for cardiac surgical treatment of endocarditis (Table 25-5) have been derived from observational studies and expert opinion. The strength of individual indications vary; thus, the risks and benefits as well as the timing of surgery must be individualized (Table 25-6). From 25% to 40% of patients with left-sided endocarditis undergo cardiac surgery during active infection, with slightly higher surgery rates with PVE than with NVE. Clinical events resulting from intracardiac complications, which are most reliably detected by TEE, justify most surgery. In the absence of randomized trials to evaluate a survival benefit for surgical intervention, the effect of surgery has been assessed in studies comparing populations of medically and surgically treated patients matched for the necessity of surgery (indication), with adjustments for predictors of death (comorbidity) and time of the surgical intervention. Although study results vary, surgery for currently advised indications appears to convey a significant survival benefit (27–55%) that becomes apparent only with follow-up for ≥6 months after the intervention. During the initial weeks after surgery, mortality risk is actually increased (disease-plus surgery-related mortality). With less demanding surgical indications, this combined mortality risk may erode potential long-term benefits. Benefit is greatest for NVE complicated by heart failure or myocardial abscess and is less clear for PVE; this difference may reflect sample size in the relevant studies.
INDICATIONS FOR CARDIAC SURGICAL INTERVENTION IN PATIENTS WITH ENDOCARDITIS
Congestive Heart Failure Moderate to severe refractory CHF caused by new or worsening valve dysfunction is the major indication for cardiac surgical treatment of endocarditis. At 6 months of follow-up, patients with left-sided endocarditis and moderate to severe heart failure due to valve dysfunction who are treated only medically have a 50% mortality rate; the figure is 15% among matched patients who undergo surgery. The survival benefit with surgery is seen in both NVE and PVE. Surgery can relieve functional stenosis due to large vegetations or restore competence to damaged regurgitant valves by repair or replacement.
Perivalvular Infection This complication, which is most common with aortic valve infection, occurs in 10–15% of native valve and 45–60% of prosthetic valve infections. It is suggested by persistent unexplained fever during appropriate therapy, new electrocardiographic conduction disturbances, and pericarditis. TEE with color Doppler is the test of choice to detect perivalvular abscesses (sensitivity, ≥85%). For optimal outcome, surgery is required, especially when fever persists, fistulae develop, prostheses are dehisced and unstable, and invasive infection relapses after appropriate treatment. Cardiac rhythm must be monitored since high-grade heart block may require insertion of a pacemaker.
Uncontrolled Infection Continued positive blood cultures or otherwise-unexplained persistent fevers (in patients with either blood culture–positive or –negative endocarditis) despite optimal antibiotic therapy may reflect uncontrolled infection and may warrant surgery. Surgical treatment is also advised for endocarditis caused by organisms for which experience indicates that effective antimicrobial therapy is lacking (e.g., yeasts, fungi, P. aeruginosa, other highly resistant gram-negative bacilli, Brucella species, and probably C. burnetii).
S. aureus Endocarditis The mortality rate for S. aureus PVE exceeds 50% with medical treatment but is reduced to 25% with surgical treatment. In patients with intracardiac complications associated with S. aureus PVE, surgical treatment reduces the mortality rate twentyfold. Surgical treatment should be considered for patients with S. aureus native aortic or mitral valve infection who have TTE-demonstrable vegetations and remain septic during the initial week of therapy. Isolated tricuspid valve endocarditis, even with persistent fever, rarely requires surgery.
Prevention of Systemic Emboli Death and persisting morbidity due to emboli are largely limited to patients suffering occlusion of cerebral or coronary arteries. Echocardiographic determination of vegetation size and anatomy, although predictive of patients at high risk of systemic emboli, does not identify those patients in whom the benefits of surgery to prevent emboli clearly exceed the risks of the surgical procedure. Net benefits from surgery to prevent emboli are most likely when other surgical benefits can be achieved simultaneously—e.g., repair of a moderately dysfunctional valve or debridement of a paravalvular abscess. Only 3.5% of patients undergo surgery solely to prevent systemic emboli. Valve repair avoiding insertion of a prosthesis makes the benefit-to-risk ratio of surgery to address vegetations more favorable.
Timing of Cardiac Surgery In general, when indications for surgical treatment of infective endocarditis are identified, surgery should not be delayed simply to permit additional antibiotic therapy, since this course of action increases the risk of death (Table 25-6). After 14 days of recommended antibiotic therapy, excised valves are culture-negative in 99% and 50% of patients with streptococcal and S. aureusendocarditis, respectively. Recrudescent endocarditis on a new implanted prosthetic valve follows surgery for active NVE and PVE in 2% and 6–15% of patients, respectively. These frequencies do not justify the risk of adverse outcome with delayed surgery, particularly in patients with severe heart failure, valve dysfunction, and staphylococcal infections. Delay is justified only when infection is controlled and CHF is resolved with medical therapy.
TIMING OF CARDIAC SURGICAL INTERVENTION IN PATIENTS WITH ENDOCARDITIS
Among patients who have experienced a neurologic complication of endocarditis, further neurologic deterioration can occur as a consequence of cardiac surgery. The risk of neurologic deterioration is related to the type of neurologic complication and the interval between the complication and surgery. Whenever feasible, cardiac surgery should be delayed for 2–3 weeks after a nonhemorrhagic embolic infarction and for 4 weeks after a cerebral hemorrhage. A ruptured mycotic aneurysm should be treated before cardiac surgery.
Antibiotic Therapy after Cardiac Surgery
Bacteria visible in Gram-stained preparations of excised valves do not necessarily indicate a failure of antibiotic therapy. Organisms have been detected on Gram’s stain—or their DNA has been detected by PCR—in excised valves from 45% of patients who have successfully completed the recommended therapy for endocarditis. In only 7% of these patients are the organisms, most of which are unusual and antibiotic resistant, cultured from the valve. Despite the detection of organisms or their DNA, relapse of endocarditis after surgery is uncommon. Thus, when valve cultures are negative in uncomplicated NVE caused by susceptible organisms, the duration of preoperative plus postoperative treatment should equal the total duration of recommended therapy, with ~2 weeks of treatment administered after surgery. For endocarditis complicated by paravalvular abscess, partially treated PVE, or cases with culture-positive valves, a full course of therapy should be given postoperatively.
Extracardiac Complications Splenic abscess develops in 3–5% of patients with endocarditis. Effective therapy requires either image-guided percutaneous drainage or splenectomy. Mycotic aneurysms occur in 2–15% of endocarditis patients; one-half of these cases involve the cerebral arteries and present as headaches, focal neurologic symptoms, or hemorrhage. Cerebral aneurysms should be monitored by angiography. Some will resolve with effective antimicrobial therapy, but those that persist, enlarge, or leak should be treated surgically if possible. Extracerebral aneurysms present as local pain, a mass, local ischemia, or bleeding; these aneurysms are treated by resection.
Older age, severe comorbid conditions and diabetes, delayed diagnosis, involvement of prosthetic valves or the aortic valve, an invasive (S. aureus) or antibiotic-resistant (P. aeruginosa, yeast) pathogen, intracardiac and major neurologic complications, and an association with health care adversely affect outcome. Death and poor outcome often are related not to failure of antibiotic therapy but rather to the interactions of comorbidities and endocarditis-related end-organ complications. Overall survival rates for patients with NVE caused by viridans streptococci, HACEK organisms, or enterococci (susceptible to synergistic therapy) are 85–90%. For S. aureus NVE in patients who do not inject drugs, survival rates are 55–70%, whereas 85–90% of injection drug users survive this infection. PVE beginning within 2 months of valve replacement results in mortality rates of 40–50%, whereas rates are only 10–20% in later-onset cases.
In the past, in an effort to prevent endocarditis (long a goal in clinical practice), expert committees have supported systemic antibiotic administration prior to many bacteremia-inducing procedures. In the absence of human trials, a reappraisal of the indirect evidence for antibiotic prophylaxis for endocarditis by the American Heart Association has culminated in guidelines that reverse prior recommendations and restrict prophylactic antibiotic use. At best, the benefit of antibiotic prophylaxis is minimal. Most endocarditis cases do not follow a procedure. In case-control studies, dental treatments—widely considered as predisposing to endocarditis—occur no more frequently before endocarditis than in matched controls. Furthermore, the frequency and magnitude of bacteremia associated with dental procedures and routine daily activities (e.g., tooth brushing and flossing) are similar. Because dental procedures are infrequent, exposure of cardiac structures to bacteremic oral-cavity organisms is notably greater from routine daily activities than from dental care. The relation of gastrointestinal and genitourinary procedures to subsequent endocarditis is more tenuous than that of dental procedures. In addition, cost-effectiveness and cost-benefit estimates suggest that antibiotic prophylaxis represents a poor use of resources.
Studies in animal models suggest that antibiotic prophylaxis may be effective. Thus it is possible that rare cases of endocarditis are prevented. Weighing the potential benefits, potential adverse events, and costs associated with antibiotic prophylaxis, the American Heart Association and the European Society of Cardiology now recommend prophylactic antibiotics (Table 25-7) only for those patients at highest risk for severe morbidity or death from endocarditis (Table 25-8). Maintaining good dental hygiene is essential. Prophylaxis is recommended only when there is manipulation of gingival tissue or the periapical region of the teeth or perforation of the oral mucosa (including surgery on the respiratory tract). Prophylaxis is not advised for patients undergoing gastrointestinal or genitourinary tract procedures. High-risk patients should be treated before or when they undergo procedures on an infected genitourinary tract or on infected skin and soft tissue. The British Society for Antimicrobial Chemotherapy continues to recommend prophylaxis for at-risk patients undergoing selected gastrointestinal and genitourinary procedures. In contrast, the National Institute for Health and Clinical Excellence in the United Kingdom found no convincing evidence that antibiotic prophylaxis was cost-effective and advised discontinuation of the practice (see www.nice.org.uk/guidance/CG64).
ANTIBIOTIC REGIMENS FOR PROPHYLAXIS OF ENDOCARDITIS IN ADULTS WITH HIGH-RISK CARDIAC LESIONSa, b