Thrombosis and Embolism From Cardiac Chambers and Infected Valves

Transcription

1 768 lacc Vol g, No 6 December B-87B Thrombosis and Embolism From Cardiac Chambers and Infected Valves PHILIP C. ADAMS, BA, MRCP,* MARC COHEN, MD, FACC,* JAMES H. CHESEBRO, MD, FACC,t VALENTIN FUSTER, MD, FACC* New York. New York and Rochester, Minnesota In a number of cardiac conditions (acute myocardial infarction, chronic left ventricular aneurysm, dilated cardiomyopathy, infective endocarditis and atrial fibrillation in the absence of valvular disease), the risk of embolism gives cause for concern. Although anticoagulation with warfarin (Coumadin)-derivatives has been shown to be effective in some of these situations, there is no evidence regarding the role of antiplatelet agents. The common factor in the thromboembolic potential of acute myocardial infarction, chronic left ventricular aneurysm and dilated cardiomyopathy is mural throm bus. This can be detected by two-dimensional echocardi ography and indium-ill platelet scintigraphy. Although of value in elucidating the natural history of mural thrombus, in most cases, management is not substantially aided by these investigations. In patients with extensive myocardial infarction, particularly anterior infarction, moderate intensity anticoagulation started soon after hospital admission reduces the rate of embolism. After 8 to 12 weeks, embolic risk is low so that anticoagulants can usually be discontinued. Patients with chronic left ventricular aneurysm have a low incidence ofembolism; anticoagulation is, therefore, inappropriate. Dilated cardiomyopathy is associated with a high risk of embolism; moderate intensity anticoagulation may be advisable in many such cases. Little information is available regarding the incidence of thromboembolism or the role of antithrombotic therapy in the patient with a diffusely dilated left ventricle due to ischemic heart disease. In native valve infective endocarditis, the risk ofhem- orrhage is high, and the efficacy of conventional anticoagulants unclear; thus, anticoagulation should not be instituted for the cardiac condition as such. However, in prosthetic valve endocarditis, the risk of embolism seems to be very high, and anticoagulant therapy should be continued, but with great care because there is a substantial risk of cerebral hemorrhage. Atrial fibrillation in patients with valvular heart disease is dealt with in a previous review. Patients with nonvalvular atrial fibrillation are at varying risk of embolism, depending on the etiology of the arrhythmia; trials of antithrombotic therapy are needed for the various subsets of patients. In most elderly patients, the etiology is not known, and their stroke risk is high. The risk of embolism in younger patients with idiopathic atrial fibrillation is so low as to make any antithrombotic therapy unnecessary. Patients with atrial fibrillation due to hypertrophic cardiomyopathy are at a high risk of embolism, while for patients with atrial fibrillation due to ischemic heart disease, the risk is uncertain. Anticoagulation is appropriate in some patients around the time ofdirect current cardioversion. In addition, in those conditions that would of themselves merit long-term an ticoagulation, this should be continued after cardioversion. For each patient, the potential benefits of anticoagulation have to be carefully balanced against the risks of bleeding. The intensity of anticoagulant therapy applied should be matched to the patient's clinical condition. (J Am Coil CardioI1986;8:76B-87B) In this presentation, we will discuss intracavitary thrombosis and consequent embolism in patients with acute myocardial infarction, chronic left ventricular aneurysm, dilated car- From the *Division of Cardiology, Department of Medlcme, Mount Sinai School of Medicine of the City University of New York, New York, New York and the tdivision of Cardiology, Mayo Clinic, Rochester, Minnesota, Dr. Adams is a recipient of a BritJsh-American Research Fellowship of the American Heart Association and British Heart Foundation, 1986 by tbe Amencan College of Cardiology diomyopathy. infective endocarditis and atrial fibrillation. Embolism makes an important contribution to the morbidity and mortality associated with each of these conditions, but there is considerable controversy regarding their optimal antithrombotic management because few large randomized Address for reprints: Valentm Fuster, MD, DivisIOn of Cardiology, Mount Sinai Medical Center, One Gustave Levy Place, New York, New York 10029, /86/$350

2 lacc Vol. 8, No 6 December B-87B 1NTRACARDIAC THROMBOSIS AND EMBOLISM 77B studies have addressed this issue, Our analysis aims to describe the impact of thromboembolism and define the potential risks and benefits of anticoagulation in this group of conditions, We offer guidelines for therapy on the basis of our own clinical practice and the recommendations of the American College of Chest Physicians (ACCP)-National Heart, Lung, and Blood Institute Committee (1), Mural Thrombi Intraventricular mural thrombi are common in patients with acute myocardial infarction, chronic left ventricular aneurysm and dilated cardiomyopathy, The improvements in imaging of thrombi by both two-dimensional echocardiography and scintigraphy with indium-labeled platelets has led to increased interest in the detection of mural thrombi. with several reviews (2-7) appearing overthe last few years, Underlying pathology. Two major mechanisms contribute to intracavitary thrombus formation in patients with acute myocardial infarction, chronic left ventricular aneurysm and dilated cardiomyopathy, namely, abnormalities of the endocardium and stasis, In the first 2 days after myocardial infarction, leukocytic infiltration lifts off the endocardium (8), exposing subendothelial tissue. Consequently, the circulating blood is exposed to a thrombogenic surface, and thrombi become common by 4 to 5 days after infarction (9), If a left ventricular aneurysm develops, endocardial abnormalities persist and may be fibroelastic or thrombotic (10), The latter is more likely to be the source of emboli, Endocardial abnormalities may also be seen in patients with cardiomyopathy (11). Thrombi developing in patients with myocardial infarction are always adherent to regions of akinesia or dyskinesia (12), with stasis of adjacent blood. Stasis of intracavitary blood within some left ventricular aneurysms can be detected by the demonstration of swirling smoke-like echoes by two-dimensional echocardiography (13,14). Stasis in the atria in patients with atrial fibrillation predisposes to thrombus formation in a similar way. In patients with myocardial infarction, a third factor, the systemic reaction to the event, with fever, leukocytosis and increased plasma concentrations of acute phase proteins, may also contribute to the tendency to form thrombi (15), Imaging techniques for the detection of mural thrombus. Angiography may detect left ventricular thrombus. However, it is not suited for repeated observations and is insensitive, detecting only about 30% of pathologically confirmed mural thrombi (16,17). In addition, angiography in patients with left ventricular aneurysm may fail to detect thrombus because the apex may not be well filled during left ventriculography, and the thrombus often conforms to the outline of the rest of the ventricle, Computed tomography and nuclear magnetic resonance imaging have been used to detect mural thrombus and are promising techniques. Two-dimensional echocardiography has been applied to the detection of mural thrombi in the three left ventricular conditions under consideration (18,19) (Fig. 1). Adequate images are obtained in most patients (19), and the technique is suitable for repeated observations. Correct interpretation of intracavitary echoes as thrombus is clearly essential. Thrombi are almost always associated with an abnormality of wall motion (12), commonly at the apex. The margins of the thrombus are distinct, with occasionally free motion of protruding parts of the mass. The texture of the mass is usually different from that of the underlying myocardium. The mass may vary in appearance in serial examinations. In some cases, the center of the thrombus is lucent (19,20), a feature seen shortly before the thrombus becomes undetectable, and thought to be due to spontaneous lysis (19). Sources of difficulty may include apical trabeculation in patients with left ventricular hypertrophy, false chordae and tangential imaging of normal left ventricular myocardium (21). Nearfield artifact is a particularproblem when imaging from the apex, especially in the presence of increased myocardial echogenicity (20), Despite these problems, studies (18,20,22) of the performance of echocardiography suggest a sensitivity and specificity of 80 to 90% or greater. Scintigraphy after the administration of indium-ill-labeled platelets, unlike two-dimensional echocardiography, provides pathophysiologic data by detecting continuing deposition of platelets on the thrombus (18). Images have to be obtained soon after injection and later after time has elapsed to allow platelets to accumulate on the thrombus, typically 3 to 4 days. Only those thrombi actively incorporating platelets are imaged. Because of this, the technique has a sensitivity of only about 70%, although false positive studies are unusual (6,23). It is not known whether continuing platelet incorporation into the thrombus, as detected by indium-labeled platelet scintigraphy, is a marker for greater embolic potential when compared with the echocardiographic detection of thrombus that does not take up platelets actively, Both two-dimensional echocardiography and indium-labeled platelet scintigraphy are reliable and efficient techniques and have furthered our understanding of the frequency and significance of mural thrombi. However, their value to the clinician in the management of the individual patient is not well defined. We believe that the results of these tests have only limited value in therapeutic decision making. Mural Thrombosis and Thromboembolism in Specific Clinical Settings Acute Myocardial Infarction Myocardial infarction predisposes to both mural thrombus formation and subsequent embolism. Only a small proportion of patients with myocardial infarction and mural

3 788 lacc Vol ~. No () December B-87B thrombus will develop clinical embolism, although most patients with clinical arterial embolism have or have had mural thrombus. We, therefore, consider both mural thrombus formation, detected pathologically and by the imaging techniques discussed, and embolism detected clinically. Natural history and risk factors. Autopsy studies (24,25) suggest that about 40% (range 17 to 66) of patients dying after myocardial infarction have mural thrombus at the time of death. This figure may overestimate the actual frequency in life because it includes patients with larger infarcts (those who died) and patients with infarcts of various ages. Mural thrombus detection. Echocardiography frequently reveals mural thrombus in patients with acute myocardial Figure l. Mural thrombus detected by echocardiography. Upper panel, Thrombus (T) protruding into the left ventricular cavity in this apical four chamber view. Lower panel, Multilobular. pedunculated thrombus (TH), in a patient with a dyskinetic apex due to recent infarction, again in an apical four chamber view. (Lower panel reprinted with permission from Goldman ME, Friedman S. Introduction to Cross-sectional Echocardiography. an AV Program. Medical Education Programs Ltd, 1984.) infarction (Fig. I). Most echocardiographic studies show an incidence rate of 30 to 40% in patients with anterior infarction, and up to 5% in patients with inferior myocardial infarction. The thrombi are seen most frequently on the third to sixth days after the infarction. Asinger et a1. (12) reported the results of echocardiography in 70 patients with a first infarction studied prospectively, in whom echocardiograms were performed at intervals of approximately 3 days. Despite the adminstration of low dose anticoagulants, 12 of 35 patients with anterior infarction developed mural thrombus compared with none of 35 patients with inferior infarction. Thrombi were first detected between I and 11 (mean 5) days after infarction. A similar study (26) of 54 patients with anterior infarction revealed mural thrombus in 17 (31 %) in the absence of anticoagulation. More detail of the natural history was provided by a study (27) of 96 patients treated with heparin 10,000 IU/day subcutaneously, with additional warfarin for patients with anterior infarction. These patients were imaged within 24 hours of infarction and at 3 days, 4 days and 4 months. A total of 18 patients developed thrombus. Two thrombi were present within 24 hours of admission, a further 3 by 2 days and a further 10 by 3 days. Three additional patients had developed thrombus by the 4 month examination. In all three studies, thrombus was much more frequent in patients with a large anterior wall infarct, particularly those with apical dyskinesia. However, mural thrombus has been detected in 30% of patients with inferior infarction in one series (28). Platelet scintigraphy detects a similar number of mural thrombi. For example, 48% of patients with anterior infarction studied in the first week after infarction had a positive scan, despite the use of prophylactic subcutaneous heparin (29). Embolism. Mural thrombosis, as just described, predisposes to embolism, most emboli arising from mural thrombi. In the 1930s, some 10% of the deaths of patients with myocardial infarction were attributed to emboli (30). Furthermore, autopsy studies (24,25) reveal emboli in 25 to 50% of cases of myocardial infarction, most commonly to the viscera. These were likely to be clinically important in only 5 to 8% of the population studied. In the British Medical Research Council study (31) of anticoagulation in acute myocardial infarction, 3.4% of the control group (no anticoagulation) developed clinical systemic embolism. Embolic stroke occurs in some 2 to 3% of patients within 4 weeks of myocardial infarction, 90% of these strokes occurring in patients with anterior infarction (32,33). The risk of embolism peaks late in the first week after infarction, and then decreases rapidly so that after about 6 to 8 weeks, it is very low (0.35/100 patient-years) (34). Although riskfactorsfor embolism and mural thrombosis are similar, clinical prediction of embolism is clearly most important. Emboli occur most frequently in patients with echocardiographic evidence of mural thrombus (12,26,35).

4 JACC Vol. 8, No 6 December 1986:76B-87B 79B Mural thrombus is detected most commonly in patients with a large anterior myocardial infarction and poor hemodynamic status. Whether the detection of a mural thrombus in such patients suggests that they are at greater risk of embolismthan otherwise similarpatientsin whomsuch thrombi are not detected is at present unknown. However, if a thrombus is detected, certain features of the thrombus at echocardiography are associated with a higher rate of embolism. Protruding thrombi, particularly if they are mobile (13), are more likely to embolize than are flat thrombi (36,37). Detection of high risk factors. Other features than those detected by echocardiography may also be helpful in defining a group of patients at high risk of embolism. Site of infarction is important, with emboli being more frequent, as mentioned, in patients with anterior infarction. However, in one series (38), 30% of strokes were in patients with inferior infarctions. Patients with a large infarction are at higher risk of embolism. In one study (38), nine cases of stroke occurred in 480 patients receiving no anticoagulants. All the strokes occurred in patients with a peak creatine kinase concentration of more than nine times the upper limit of normal. Venous thromboembolic phenomena were also confined to this group of patients. Similar findings were reported by Johannessen et al. (19), who found that patients developing stroke had a larger heart and higher peak enzyme concentration than did those without stroke. Another study (33) failed to confirm these findings, although up to 25% of the patients with stroke in that series may have had a nonembolic stroke, and no information is given about therapy. Atrial fibrillation was, however, more common in patients who developed stroke (33). Plasma fibrinogen concentration, which peaks 4 to 6 days after infarction, is also strongly associated with thromboembolism. In one study (15) of 120 patients with infarction, all 10 thromboembolic events recorded (including 4 pulmonary emboli) occurred in those patients with a fibrinogen concentration greaterthan 650 mg/dl, with 9 occurring when the concentration was greater than 800 mg/dl. In that study (15), fibrinogen concentration was the best predictor of embolism in a multivariate model. It is clear from the foregoing discussion that patients at high risk of embolism are those with a large infarction, particularly if anterior, whether or not they have mural thrombus. Studies of preventive therapy. The frequency both of mural thrombus and embolism may be reduced by anticoagulant therapy, although the two may not be reduced to the same extent. To be effective in the prevention of mural thrombus, anticoagulation has to be given in adequate dose and as early as possible in the course of the condition before the thrombus has had time to become established. Autopsy studies (39,40) reveal a reduction in frequency of mural thrombus of 50 to 70% and an approximate halving in the rate of systemic embolism with high dose subcutaneous anticoagulation given soon after arrival in hospital. Mural thrombus formation and subsequent embolism are not greatly reduced by low dose or inappropriately timed anticoagulation. This is revealed by the echocardiographic studies mentioned before, in which patients received anticoagulants, in low doses (that is, heparin subcutaneously 5,000 units twice daily) (12) or the addition of coumadin to initial low dose heparin only after mural thrombus was detected (27). However, heparin (to maintain the activated partial thromboplastin time at 70 to 140 seconds) and warfarin (to maintain the Thrombotest at 5 to 10%; prothrombin time of 1.4 to 2.2 times control) prevented the development ofmural thrombus in a single-blind study (41), this treatment being started within 12 hours of the onset of symptoms of infarction. Some studies (12,42) suggest that resolution of mural thrombi is more rapid and complete if anticoagulants are given. However, another study (43) showed no increase in the late resolution of mural thrombus with anticoagulants, despite a definite reduction in the incidence ofembolic events. The differences between the results of the studies on acute infarction may relate to differences in the groups of patients studied, in particular, the time at which the echocardiogram is performed and the time at which the patient enters the hospital after the onset of symptoms. In summary, the evidence suggests that full anticoagulation when given within 12 hours of the onset of pain moderately reduces the incidence of mural thrombus. Three large studies (31,44,45) tested whether short-term anticoagulation can reduce mortality in acute myocardial infarction and examined the prevention of embolic events by this therapy. Anticoagulation in these studies consisted of full dose heparin, either intravenously (31) or subcutaneously (44,45), followed by oral anticoagulants to prolong the prothrombin time to up to 1.5 times control. In the Medical Research Council (MRC) study (31), for example, the incidence of systemic emboli during the hospital admission was reduced by treatment from 3.4% (24 of 715 patients) in the control group to 1.3% (9 of 712 patients) in the treated group. Overall, the three studies (31,44,45) suggest a 24 to 75% reduction in relative risk with this moderate intensity of oral anticoagulation. Mural thrombosis and systemic embolism may not be affected by therapy in the same way. Most likely no treatment will prevent the development of minor mural thrombus due to exposure ofdamaged tissue. However, further growth of thrombus with the potential for clinically significant embolism is inhibited by anticoagulants, the rate of embolism being reduced in the three large trials (31,44,45). Although the evidence that less intense anticoagulation reduces mural thrombus formation is contradictory, the evidence is convincing that full dose intravenous heparin followed by oral vitamin K antagonists protects against embolism. The risks of anticoagulation in patients with no clear contraindications were accurately defined by the three large

5 80B JACC Vol X. No 6 December IYX6 76B-X7B studies of acute myocardial infarction (46). In the Veterans Administration study (44), there were 13 major bleeding episodes (2.6% of 500 patients) in the treated group and 6 (1.2% of499) in the untreated group (the anticoagulant agent was heparin, adjusted according to the clotting time, together with coumadin until the prothrombin time was greater than 1.3 times control, then the heparin was discontinued). In the MRC study (31), clinically obvious hemorrhage was recorded in 36 (5%) o 112 treated patients and in 9 (1.3%) of 715 untreated patients. Most of these bleeding episodes were minor, and no deaths occurred. The risks of shortterm anticoagulation in this group of patients are therefore low. The demonstration of mural thrombus on an echocardiogram is unlikely to materially alter the decision about anticoagulation. Anticoagulation has been shown to be effective if given on admission to hospital, not at other times. A delay until an echocardiogram is at hand may miss the time in which therapy can be most valuable. The prevention of the development of a mural thrombus is the most important goal of therapy in the acute phase. In one study (19), warfarin was added to prophylactic heparin only after a mural thrombus was demonstrated by echocardiography. Five strokes occurred in the series of 90 patients, and three were in patients who had been given warfarin after mural thrombus had been detected (19). Because the majority of mural thrombi develop only after 3 days after infarction, echocardiography in the first 2 days would fail to detect the majority of the patients who would develop mural thrombus (27). Echocardiography later in the hospital course, for example, at 10 days (47), might detect thrombi only after a significant proportion of the emboli had occurred (33). Multiple echocardiographic examinations are impractical, except as a research tool. Indeed, as discussed, clinical features, available in all patients, usefully define a high risk group. Suggested management. All patients with a large infarction should receive anticoagulants on admission, with intravenous heparin in full doses. We favor treating patients who have an anterior infarction or a very extensive inferior infarction. Although not specifically addressed in this presentation, the potential for venous thromboembolism is taken into account when deciding to initiate anticoagulant therapy in a patient with myocardial infarction. Some risk factors are common for both arterial and venous thromboembolism, namely, high peak enzyme concentration, heart failure and atrial fibrillation. Oral anticoagulation ofmoderate intensity (prothrombin time of 1.2 to 1.5 times control, international normalized ratio of 1.6 to 2.5) should be started 2 to 3 days before heparin is to be discontinued, and continued for 8 to 12 weeks. Anticoagulation at this intensity is recommended because it prevented embolism in the MRC and Veterans Administration studies (31,44). The prevention of venousthrombosis in patients with acute myocardial infarction is dealt with elsewhere in this symposium (48,49). Chronic Left Ventricular Aneurysm Natural history, epidemiology and risk factors. Variation in the definition of this condition makes valid comparison of various series difficult, particularly because different techniques may be used to define the abnormality (autopsy, surgery, angiography, echocardiography). However, there is agreement that mural thrombus within a left ventricular aneurysm is common. Surgical series (2) have reported an overall frequency of 47%, with up to 95% in one series (50). As most of the surgical data are retrospective, these figures may be underestimated. Necropsy data (again almost entirely retrospective) (2) suggest a very similar (49%) frequency of mural thrombus. The thrombus in most cases persists from the acute phase of the infarction, persistence being favored by the stasis within the aneurysm. Despite the frequent persistence of thrombus, the incidence of clinical systemic embolism gradually decreases after acute infarction to a low level of 0.35 events per 100 patient-years (34) after about 3 months. The prevalence of embolism at necropsy in cases of chronic aneurysm is higher (up to almost 40%) (2). In one study (17) of 100 surgical cases of chronic aneurysm, five patients had had clinical emboli, of which four were within 3 months of infarction; the date of the fifth patient's infarction was unknown. Another series (16) of 58 patients described 2 patients with embolism of which one event occurred despite anticoagulation. In a study in which follow-up and event rate were carefully documented, Lapeyre et al. (34) described a low incidence ofemboli in a medically managed population with aneurysm. One event occurred in 288 patient-years of follow-up in 69 patients without anticoagulation (34). Suggested management. There are no clear data to show that anticoagulants reduce the rate of embolism in the setting of chronic left ventricular aneurysm (more than 3 months after infarction). However, even if there were, because the rate of embolism is so low in medically managed patients, the benefits of anticoagulation are unlikely to outweigh the risks (32,34). If a stroke does occur in a patient with an aneurysm, it may be due to associated cerebrovascular disease rather than to an embolus arising from the heart. The efficacy of anticoagulants in the prevention of recurrent embolism is not known, but most patients who have had an embolus would receive anticoagulants. Occasionally a left ventricular thrombus is discovered at echocardiography or angiography, most often in a patient within 6 to 8 weeks of infarction, with features that suggest that a high risk of embolism is likely. Some investigators (51) have successfully used urokinase to lyse such a thrombus. This approach was successful in the 16 patients in whom it was attempted, with no embolism in association

6 lacc Vol. 8. No.6 December B-87B 81B with the treatment. However, embolism was seen in another group of patients undergoing lysis of thrombus in the left side of the heart, those with thrombosed left-sided prosthetic valves (52). The clinical outcome in these cases was good, despite the embolism. We, therefore, believe that further experience is necessary before lysis of intracardiac thrombus can be considered standard therapy. These observations suggest that the use of thrombolytic therapy for patients with acute infarction will reduce the rate of mural thrombus formation (53). Surgery (thrombectomy or aneurysmectomy) has been used rarely in patients with large, freely mobile thrombi or if recurrent embolism has failed to respond to anticoagulation. Dilated Cardiomyopathy Natural history, epidemiology and risk factors. Autopsy studies reveal a high incidence of mural thrombi in patients with dilated cardiomyopathy. For example, Roberts and Ferrans (II) found intracardiac thrombus in 53% of 60 patients with dilated cardiomyopathy (excluding those with coronary disease). Thrombi were found in the left ventricle in 27 cases, in the right ventricle in 15, in the right atrium in 12 and in the left atrium in 5. In 19 cases, thrombi were found in more than one chamber. The endocardium under the thrombus was commonly abnormal. This suggests that in addition to the slow flow seen in these cases, endocardial abnormalities predispose to thrombus formation. Indeed, the preponderance of left ventricular over left atrial thrombus supports this view because there is a greater degree of stasis in the atrium, particularly in the appendage. Clinical studies (54) of patients with dilated cardiomyopathy suggest that the cumulative incidence of clinical embolism, secondary in most cases to left ventricular thrombus, is 18% in patients not taking anticoagulants. Because two-thirds of the patients in this Mayo Clinic study died within 2 years of diagnosis, the rate of embolism was high. Similar results (15 emboli occurring in 106 patients followed up to 2 to 22 years) were obtained by Hatle et ai. (55). There is also a high risk of pulmonary embolism in these patients, almost 5% in this series (55). Of 100 patients studied prospectively in another series (56), II had systemic emboli, while the same number had pulmonary emboli. Pulmonary embolism, which may contribute to the clinical condition despite being undiagnosed during life, is frequent in these patients; thus, the 7% mortality rate due to pulmonary embolism in this series probably reflects a higher undiagnosed rate. Embolism in patients with cardiomyopathy is more frequent than in patients with chronic left ventricular aneurysms, probably because in the latter, the wall adjacent to the thrombus is static. There is, therefore, less stimulus to dislodge the thrombus. Left ventricular thrombus can be detected noninvasively in patients with dilated cardiomyopathy as in myocardial infarction and left ventricular aneurysm. In one series (57), 123 patients with diffuse hypokinesia were studied; in about 60%, this was due to cardiomyopathy and in the remainder it was due to coronary disease. Thirty-six percent of the entire group had left ventricular thrombus by echocardiography. Of the 96 patients followed up, 11 % developed systemic embolism over a mean follow-up period of21 years (range 9 months to 4 years; 192 patient-years). The risk of embolism was not increased in patients with left ventricular thrombus (40%) by echocardiography. Neither was the cause of the hypokinesia related to the embolic risk, emboli being equally frequent in those with coronary disease. The echocardiogram seems unlikely to be useful in defining a high risk population by the detection of intracardiac thrombus in patients with cardiomyopathy. Other factors that determine the risk ofembolism in cardiomyopathy have been scantily addressed. There are poorly substantiated suggestions that patients with alcoholic cardiomyopathy have a particularly high incidence of embolism. A 100% incidence of mural thrombus at postmortem study has been reported in a well defined group of patients with alcoholic cardiomyopathy (58). There is a high incidence of embolism in peripartum cardiomyopathy (59) and in the cardiomyopathy of hypereosinophilia (60). Patients with hypertrophic cardiomyopathy and atrial fibrillation have a high incidence of embolism, particularly in association with rhythm changes (61). In contrast, cardiac rhythm does not seem to be a major consideration in the majority of patients with dilated cardiomyopathy, presumably because the thrombi that are the source of the emboli are usually ventricular, while in hypertrophic cardiomyopathy, they are atrial. Fourteen percent of patients with sinus rhythm in one series (54) of patients with dilated cardiomyopathy developed emboli, while only 2 of 10 patients developing emboli in another series (62) had atrial fibrillation. Eight of 15 patients with emboli had sinus rhythm in the series of Hatle et al. (55). Hemodynamic factors have only rarely been correlated with the likelihood of embolism in this group of patients. In one study (57), there was no difference in echocardiographic left ventricular dimensions or ejection fraction between patients with and without left ventricular thrombus. Thrombus morphology was not, in this study, associated with embolic rate. Suggested management. Little information is available regarding the role of antithrombotic treatment in the diffusely dilated left ventricle due to ischemic heart disease. Although there is no prospective study of anticoagulation in dilated cardiomyopathy, the data from the Mayo Clinic (54) support the use of anticoagulants in patients with dilated cardiomyopathy. No patient on anticoagulant therapy developed a thromboembolic event in 101 patient-years of follow-up, whereas emboli occurred in 18% of patients not receiving anticoagulants followed up for 624 patient-years (that is, an event rate of 3.5/100 patient-years). Although

7 828 JACC Vol g. No 6 December I9g6 76B-g7B the patients were not randomized to the two treatment groups, the groups were comparable. Thus, it is recommended that patients with dilated cardiomyopathy receive long-term anticoagulation with coumadin to moderate intensity (32,54). Furthermore, this is particularly indicated in the presence of heart failure, prolonged bed rest or other conditions that predispose to venous thromboembolism, an important additional indication for anticoagulant therapy in these patients. The stage in the natural history of the disorder at which therapy should be started is not known, but the number of emboli in the Mayo Clinic study (54) in the first year after diagnosis suggests that early anticoagulation may be valuable. The role of antiplatelet agents is unknown. We see no role for noninvasive detection of thrombi in decisionmaking in these patients. Infective Endocarditis Thrombosis and the Pathology of Endocarditis Normal valvular endothelium is resistant to infection. Trauma to the valve promotes deposition of fibrin and platelets (63), producing the lesion of nonbacterial thrombotic endocarditis. Experimentally, many forms of stress can produce nonbacterial thrombotic endocarditis (for example, severe infections and high cardiac output states). Similar lesions in all stages of development have been seen in humans and appear to lead to all cases of infective endocarditis (64). The same pathologic findings are seen in marantic endocarditis without infection (for example, with some malignancies and connective tissue disorders). In experimental animals, treatment with anticoagulants (65), aspirin (66) and sulfinpyrazone (67) has been used in an attempt to reduce the deposition of platelets and fibrin during the induction of endocarditis. Although vegetation size was reduced by sulfinpyrazone in one study (67) of sterile vegetations, aspirin in large doses had no effect in another study (66). Survival of rabbits with experimental infective endocarditis was reduced by warfarin (65), even in the presence of penicillin. There are no data regarding the effect of antiplatelet agents or anticoagulants on the prevention of endocarditis in human subjects (63). Embolism in Endocarditis Natural history, epidemiology and risk factors. Systemic embolism, including to the brain, is the major complication of endocarditis after congestive failure. Cerebral embolism with stroke occurs in 6 to 31 % of cases treated currently (68), in 17% in one large series from the Massachusetts General Hospital (69). Strokes occur most commonly in the territory of the middle cerebral artery. Microemboli to the cerebral circulation are also common; sometimes silent, they may lead to confused mental states or mood changes. Emboli to other organs may not be clinically apparent; autopsy data revealed splenic and renal emboli in 44 and 56% of patients, respectively (70). Coronary emboli with small areas of myocardial infarction may occur in up to 40 to 60% of autopsy cases (64). Pulmonary emboli are common in right-sided endocarditis. Although these phenomena are common, emboli occluding large vessels like the femoral artery are uncommon, and are more usually due to fungal or marantic endocarditis. Left atrial myxoma may also embolize to large peripheral arteries. Emboli remain an important cause of death in endocarditis, causing 16% of the deaths in one series (71) performed in the early 1970s. Emboli occasionally occur despite successful antibiotic therapy, even long after the illness. Emboli occurring in the early phase of the illness are most likely to be in the context of Staphylococcus aureus infection (69,71). The slower growing but less common fungi (Aspergillus and Candida) and the carbon dioxiderequiring bacteria (Hemophilus spp., Cardiobacterium hominis and Actinobacillus) produce large vegetations with a high risk of embolism (68). In the later phase of the illness, streptococcol infection accounts for almost half of the cases of embolism. Some studies (69) find that mitral valve injections lead to embolism more commonly than do aortic value injections, while others (71) reach the opposite conclusion. Overall, it seems likely that aortic valve involvement leads to emboli more frequently (72). The reported rate of recurrence of emboli is variable; in the Massachusetts General Hospital series (69), only I of 37 patients had a second major cerebral embolus. However, 47% of patients with cerebral embolism had systemic embolism, and almost 30% of patients with stroke (both embolic and hemorrhagic) had a premonitory transient ischemic attack. Also, recent unpublished retrospective data from our institution (72) in a group of patients with echocardiographically detected vegetations show that the rate of recurrent emboli is twice that of first emboli. This opinion is not universal: to quote a recent review (73), "there are no data to suggest that a patient with one or two major emboli is at increased risk of having more..." Despite these reservations, we believe that the occurrence of one embolic event does presage a further event. Finally, there is little doubt that patients with mechanical prosthetic valve endocarditis have a high rate ofembolism, particularly if anticoagulant treatment has been discontinued for some reason (74). Patients in whom vegetations are detected by echocardiography are at higher risk oj embolism than are those in whom they are not detected. A recent review (75) summarized the findings in 442 patients from several studies. Thirty-four percent of patients with vegetations on echocardiography had systemic embolism, whereas this occurred in only 7% of patients without vegetations. Other workers (76,77) have shown that patients with large vegetations are more likely to develop a first embolism than are those with

8 JACC Vol. 8, No.6 December B-87B 83B small vegetations (0,1 em). Similar observations have been made in our institution (72); patients with vegetations of an area greater than 1.8 cm 2 as judged by two-dimensional echocardiography were more than 2.5 times more likely to have emboli than were those with smaller vegetations. Furthermore, in one study (78), multiple emboli occurred in patients with large vegetations. Although the echocardiographic detection of vegetations defines a population with a high incidence of embolic events, the predictive value in the individual is not high; in general, 50 to 75% of patients have echocardiographic vegetations, whereas only about 33% will have clinically apparent emboli. The detection of large vegetations may be of greater predictive value. Can anticoagulant therapy prevent embolism in native valve infective endocarditis without excessive risk? The role of anticoagulants in native valve infective endocarditis is controversial with contradictory opinions being expressed in the literature. Portions of established vegetations may embolize despite adequate anticoagulation, An alternative view is that anticoagulants could prevent layering and extension of additional friable thrombus, Clinical studies have not provided an answer. One author states: "Anticoagulants do not prevent emboli in infective endocarditis" (79). For example, in the Massachusetts General Hospital series (69), a subset of seven patients were treated with anticoagulants and yet five experienced embolism. In contrast, other investigators (32) clearly believe that anticoagulants are effective, There are no data regarding the use of antiplatelet agents in this setting, Although the possible benefits of anticoagulation are unclear, the risk of cerebral hemorrhage in patients with infective endocarditis is certainly high, In the early 1940s, heparin was used in the treatment of patients with endocarditis in an attempt to improve penicillin penetration into vegetations (80), However an unacceptably high rate (at least 20%) of cerebral hemorrhage occurred (81), after only short courses of heparin. Warfarin therapy shortens survival in experimental endocarditis in rabbits (65), In the Massachusetts General Hospital series (69), three of seven patients receiving anticoagulants developed hemorrhagic cerebral infarction. Cerebral hemorrhage occurred in only 10 of 211 patients not receiving anticoagulants, Anticoagulation increases the likelihood ofcerebral hemorrhage in endocarditis. which may be due to bleedingfrom mycotic aneurysms hemorrhagic transformation of embolic infarction or subarachnoid or intracerebral hemorrhage with no identifiable cau~e. Mycotic aneurysms are detected in life in a small percent of patients (69,70) and in up to 10% at autopsy, Because these aneurysms may resolve spontaneously, the true incidence is unknown. Such aneurysms commonly arise after septic emboli to the vessel wall. Perhaps 50% of patients with a mycotic aneurysm will have had preceding cerebral signs or symptoms. Certainly, the absence of neurologic manifestations reduces the likelihood of a mycotic aneurysm. In summary, the diagnosis of a mycotic aneurysm is a very strong contraindiction to anticoagulation (32). Thus, in a patient in whom anticoagulation is necessary (for example for cardiopulmonary bypass), full neurologic evaluation including angiography should be considered, particularly in the presence ofprior neurologic manifestations. Anticoagulants in prosthetic valve endocarditis. It would seem appropriate not to institute prophylactic anticoagulation in patients with endocarditis on a bioprosthetic valve, although continuing established antithrombotic therapy is a reasonable policy. To our knowledge, there is little information regarding these issues. Several studies have addressed the matter of continuing anticoagulants in patients with mechanical prosthetic valve endocarditis. A retrospective study from the Mayo Clinic (74) strongly supports continuing anticoagulants in these patients; stroke occurred in 3 (8%) of 38 patients on anticoagulant therapy and in 10 (71 %) of 14 patients in whom anticoagulants were withdrawn or were at doses producing inadequate anticoagulation. Other studies (82,83) report a high incidence of neurologic complications despite anticoagulation. For example, one study (82) reports that 10 (24%) of 42 patients on anticoagulant therapy experienced embolic (four patients) or hemorrhagic (six patients) events. In each study, hemorrhage was associated particularly, but not exclusively, with excessive anticoagulation. Although anticoagulants should be continued in the setting of prosthetic endocarditis, it has to be accepted that the risk of both embolic and hemorrhagic events remains high. Suggested management. In native valve endocarditis, there is no evidence for efficacy, whereas there is certainty that anticoagulant treatment increases the risk of hemorrhage. Therefore, anticoagulation is not recommended to prevent emboli, Neither is there good evidence that anticoagulation reduces the risk of recurrent embolism once an initial event has occurred. If anticoagulation becomes necessary (for example, if cardiopulmonary bypass is needed), consideration should be given to the possibility of a mycotic aneurysm. In mechanical prosthetic valve endocarditis, anticoagulation should be discontinued temporarily if cerebral embolism does occur. Care should be taken to minimize the likelihood of excessive anticoagulation. Bioprosthetic valve endocarditis should be treated in the same way as infection on a native valve, with previous antithrombotic therapy being continued. Again, care has to be taken to avoid excessive anticoagulation. Atrial Fibrillation Natural history, epidemiology and risk factors. As discussed elsewhere in this symposium (84) patients with atrial fibrillation and valvular heart disease have a high rate ofembolism. However, nonvalvular atrial fibrillation occurs

9 848 lacc Vol x, No 6 December B-87B much more commonly and is numerically a more important potential cause of embolism. Atrial fibrillation is a relatively common arrhythmia, occurring in 2 to 3 people per 1,000 from ages 25 to 34 years and 30 to 40 per 1,000 from ages 55 to 64 years in the Framingham Study (85). In people aged 62 to 90 years in Edinburgh, Scotland, the rate was 50 to 90 per 1,000 (86). In an earlier report (87) from the Framingham Study, the risk of stroke in nonvalvular atrial fibrillation was 41.8 per 1,000 patient-years, a relative risk of 5.6 (87). An overall annual stroke risk in nonvalvular atrial fibrillation of 3 to 5% has been suggested (32), with a 35% cumulative lifetime incidence. Nonvalvular atrial fibrillation is found in 45% of patients with cardiac embolic strokes, accounting for 5 to 6% of all strokes (32). Of course, many of the patients with nonvalvular atrial fibrillation have other associated cardiac conditions (for example, hypertensive or ischemic heart disease). They, therefore, differfrom the patients with "lone" atrial fibrillation. Thrombus formation in the left atrium in nonvalvular atrial fibrillation is likely related to stasis, but the role of additional endocardial abnormalities has not been addressed. The best predictor of stroke in patients with nonvalvular atrial fibrillation is a first stroke since recurrence occurs in up to 10 to 20% patients per annum (32). The rate of embolism is partly determined by the etiology of the arrhythmia. In patients with coronary disease, the incidence of clinical embolism is relatively low. However, because coronary disease is so common, it is a frequent cause of atrial fibrillation and embolism. Subclinical emboli, often of uncertain clinical significance, are common at autopsy in these patients. In one autopsy series (88), 35% of patients with atrial fibrillation due to ischemic heart disease had systemic emboli, while only 7% of a control group with ischemic heart disease and sinus rhythm had emboli; in most patients, embolization involved the brain (73%). Many patients with coronary disease and atrial fibrillation, who are at an increased risk of embolism, also have other relative indications for anticoagulation (for example, congestive failure with its high risk of pulmonary embolism). Patients with cardiomyopathy have already been dealt with. Patients younger than 65 years of age with lone atrial fibrillation (with no other detectable heart disease) have a low incidence of embolism; 94% of98 patients were free ofembolic events at 10 years in a study from the Mayo Clinic (89). The rate is much higher in older patients with the bradycardia-tachycardia syndrome. Thus, in one study (90) of 100 such patients, 16 had had systemic embolism during a median period of 4 years (range of a to 50). In this study, other risk factors included age greater than 50 years and prolonged atrial asystole, emphasizing the effect of stasis on thrombus formation in the left atrium. Atrial fibrillation occurs in congenital heart disease, but the risk of stroke is not well documented. A high incidence of embolism was reported by Forfang et al. (91) in secundum atrial defect with atrial fibrillation. Thyrotoxic atrial fibrillation is associated with a high risk of embolism (8 to 39%, average 14). In one series (92), 12 of 30 patients with thyrotoxic atrial fibrillation developed emboli, 53% of which were cerebral; as only 4 patients were in chronic atrial fibrillation, the rate of embolism was very high. Stability of cardiac rhythm and left atrial size have not been clearly related to the risk ofembolism in nonvalvular atrial fibrillation. Detection of left atrial thrombus by echocardiography is unreliable, and by other modalities (for example, computed tomography) is not established. In patients with atrial fibrillation, recurrent stroke after an initial event is common, occurring in up to 10 to 20% per year (32). Thus, the best predictor of stroke in these patients is an initial event. Role ofanticoagulation. The first stroke in patients with nonvalvular atrial fibrillation is likely to be clinically very significant. Therefore, prevention of stroke in these patients is important. There are no data regarding the use of antiplatelet agents in this context. Anticoagulation has been used frequently, but there are no good controlled studies. However, it is likely that they are efficacious, by analogy with their efficacy in valvular heart disease, and in the prevention of embolism in acute myocardial infarction. In a comparative study with historic controls, Freeman and Wexler (93) showed a reduction in the rate of thromboembolism in patients with" arteriosclerotic" heart disease and atrial fibrillation from 4.9% in untreated patients to 0.6% in patients receiving anticoagulants, with a similar reduction in embolism detected at autopsy. A careful analysis of the risks and benefits of anticoagulation in the bradycardia-tachycardia syndrome (94) suggests that anticoagulation is beneficial in this condition, even when the assumptions made were generally unfavorable to anticoagulation. Suggested management. Prevention of recurrent embolism should be considered in each of the patient groups discussed here by long-term moderate intensity anticoagulation (prothrombin time with rabbit thromboplastin 1.2 to 1.5 times control). This recommendation is based on analogy with venous thromboembolism, in which stasis probably plays a similar pathologic role, and against which moderate intensity oral anticoagulation provides effective prophylaxis (32). However, because many of the patients with nonvalvular atrial fibrillation are elderly, careful consideration of the potential risks and benefits in the individual is crucial. Aspirin might well be considered in some cases, although in the absence of evidence for its efficacy. Anticoagulation for cardioversion for atrial fibrillation. Several studies document the risk of embolism after cardioversion for atrial fibrillation (reviewed by Mancini and Goldberger [95]) as being 1 to 3%, the highest figure of 5.3% being found in a comparative study (96) in which

10 JACC Vol 8. No 6 December B-87B 85B embolism occurred in 0.8% of patients on anticoagulant therapy. Although this risk may seem low, it is a much higher rate per patient-year than for most other considerations. This comparative study suggests that long-term anticoagulation before cardioversion reduces the risk of embolism. However, the issue is one of administering shortterm anticoagulation before cardioversion in patients who would otherwise not be given an anticoagulant. There are no data regarding the rate of development of intraatrial thrombus or the time needed for it to become adherent to the atrial wall once present. A 2 week period is assumed to be needed (97) and, therefore, 2 to 3 weeks of moderate intensity anticoagulation is recommended to allow thrombus to become adherent before cardioversion. This should be continued for about 3 to 4 weeks thereafter because atrial mechanical activity may take 3 to 4 weeks to resume despite electrical sinus rhythm (98). Indeed, the cardiac condition predisposing to atrial fibrillation will, in many patients, warrant long-term anticoagulation. Whether all patients should receive an anticoagulant before cardioversion is unknown. The risk of embolism is higher in patients with cardiac conditions that in themselves produce a significant risk of embolism and in whom longterm anticoagulation would be considered. In lower risk patients (for example, those with ischemic heart disease), some authorities would use aspirin given for 2 to 3 weeks before cardioversion. References I. Dalen JE, Hirsh 1. ACCP-NHLBI National Conference on Antlthrombotic Therapy Che,t (suppl): IS-106S. 2. Cabm HS. Roberts We. Left ventricular aneurysm. mtraaneurysmal thrombus and sy,temlc embolus in coronary heart disease. Chest : Stratton lr. Mural thrombi of the left ventricle. Chest 1983, Nixon IV. Left ventricular mural thrombus Arch Intern Med 1983;143: Arvan S. Mural thrombi m coronary artery disease: recent advances m pathogenesis. diagnosis. and approaches to treatment Arch Intern Med 1984;144: Chesebro IH. Ezekowitz MD. Badimon L. Fuster V. Intracardiac thrombi and systemic thromboembolism: detection. mcidence and treatment. Ann Rev Med : Meltzer RS. VIsser CE, Fuster V. Intracardmc thrombi and systemic embolization. Ann Intern Med 1986;104: hnson RC, Crissman RS. DiDio LJA. Endocardial alteration, m myocardial mfarction Lab Invest 1979;40: Mallory GK, White PD. Salcedo-Salgar 1. The speed of healmg of myocardial mfarctlon: a study of the pathologic anatomy m seventy two cases Am Heart ;18' Hochman IS. Platia EB. Bulkley BH. Endocardial abnormalities m left ventncular aneurysms. a climcopathologlc study. Ann Intern Med 1984;100: II. Roberts WC, Ferran, Vl PathologICal aspects of certain cardiomyopathie, Circ Res 1974,34/35(suppl 11): Asmger RW. Mikell FL. EI,perger 1. Hodges M. Incidence of left ventricular thrombosis after acute transmural myocardial mfarction Serial evaluation by two-dimensional echocardiography N Engl 1 Med 1981 ;305: Haugland 1M, Asmger RW, Mikell FL, Elsperger 1, Hodges M. Embolic potential of left ventncular thrombi detected by two dimensional echocardiography. Circulation 1984;70: Mikell FL, Asmger RW. Elsperger Kl, Anderson Wl, Hodges M. Regional stasis of blood m the dysfunctional left ventricle: echocardiographlc detection and differentiation from early thrombosis. Circulation 1982; Fulton RM, Duckett K. Plasma fibrinogen and thromboemboli after myocardial infarction. Lancet li: Simpson MT. Oberman A. Kouchoukos NT. Rogers Wl. Prevalence of mural thrombi and systemic embolization with left ventncular aneurysm: effect of anticoagulant therapy. Chest 1980;77: Reeder GS. Lengyel M. TaJlk Al, Seward JB. Smith He. Danielson GK. Mural thrombus m left ventricular aneurysm: incidence, role of angiography. and relation between anticoagulation and embolization. Mayo ClIn Proc 1981;56: Ezekowltz MD. WIlson DA. Smith EO. et al. Companson of indium III platelet,cmtlgraphy and two-dimensional echocardiography in the diagnosis of left ventricular thrombi. N Engl 1 Med 1982; 306' hannes,en KA. Nordrehaug le, von der Lippe G. Left ventricular thrombus and cerebrovascular accident in acute myocardial mfarction. Br Heart ;51 : Stratton lr. Lighty GW lr, Pearlman AS. Ritchie ll. Detection of left ventricular thrombus by two-dimensional echocardiography: sen SItivity. specificity. and causes of uncertamty. Circulation 1982; 66: A,mger RW. Mikell FL, Sharma B, Hodges M. Observations on detecting left ventricular thrombus with two dimensional echocardiography. emphasis on avoidance of false positive diagnosis. Am 1 Cardiol 1981;47: Visser CE, Kan G, David GK. Durrer D. Two dimensional echocardiography m the diagnosis of left ventncular thrombus: a prospective study of67 patients with anatomic validation. Chest 1983;83: Ezekowltz MD, Burrow RD. Heath PW. Streitz T. Smith EO. Parker DE. Diagnostic accuracy of mdium-iii platelet scmtigraphy m Identifying left ventncular thrombi Am J Cardiol 1983;51: Hellerstem HK. Martin JW. Incidence of thrombo-embolic lesions accompanying myocardial infarction. Am Heart ;33: DaVies Ml, Woolf N. Robertson WB. Pathology of acute myocardial mfarction with particular reference to occlusive coronary thrombi. Br Heart ;38: Keatmg EC, Gross SA. Schlamowltz RA. et al. Mural thrombi in myocardial infarctions: pro,pectlve evaluation by two-dimensional echocardiography. Am 1 Med 1983;74: Visser CA, Kan G. Lie KI. Durrer D. Left ventricular thrombus followmg acute myocardial infarction: a prospective serial echocardlographic study of 96 patients. Eur Heart ;4: Kothan AJ. PaczkowskI K. Baker KM, et al. Ventncular thrombi in acute myocardial infarction. mcidence. complications. and effects of anticoagulation (abstr). 1 Am Coli Cardiol 1984;3: Ezekowitz MD. Kellerman OJ, Smith EO. Streizt TM. The detection of active left ventncular thrombosis during acute mycardial infarction usmg mdium-iii platelet scintigraphy Chest 1984; Solandt DY, Nassim R. Best CH. Production and prevention ofcardiac mural thrombosis in dogs. Lancet 1939; 11: Working Party on Anticoagulant Therapy in Coronary ThrombOSIS to the Medical Research Council Assessment of short-term anticoagulant admmistratlon after cardiac mfarctlon. Br Med ;1: Sherman DG. Dyken ML. Fisher M, Harrison MJG. Hart RG. Cerebral embolism. Chest 1986;89(suppl):82S-98S

11 868 lace Vol S, No 6 Dccemher l'is6 76B- 87B 33. Komrad MS, Coffey CEW, Coffey KS, et al. Myocardial infarcllon and stroke. Neurology 1984;34: Lapeyre AC Ill, Steele PM, Kazmler Fl, Chesebro le. Vlietstra RE, Fm,ter V Systemic embolism in chromc left ventricular aneurysm: incidence and the role of anllcoagulallon. 1 Am Coil Cardiol 1985, Arvan S. Left ventricular mural thrombi secondary to acute myocardial infarction: predisposing factors and embolic phenomenon. 1 Clin Ultrasound 1983; II : Kinney EL. The SIgnificance of left ventricular thrombi in patients with coronary heart disease: a retrospective analysis of pooled data. Am Heart ;109: Visser CA, Kan G, Meltzer RS. Embolic potential of left ventricular thrombus after myocardial infarction: a two-dimensional echocardiographic study of 119 patients 1 Am Coil CardlOl 1985;5: Thompson PL, Robinson IS. Stroke after acute myocardial Infarction: relallon to infarct size. Br Med ;2: Steffensen KA Coronary occlusion treated WIth small doses of heparin. Acta Med Scand 1969;186: Hilden T, Iversen K, Raaschou F. Anticoagulants in acute myocardial Infarction. Lancet 1961; n: Nordrehaug le, lohannessen K-A, von der Lippe G. Usefulness of high-dose anticoagulants In preventing left ventricular thrombus In acute myocardial Infarction Am 1 Cardlol 1985; Tramarin R, Pozzoli M, Febo 0, et al. Two-dimenSional echocardiographic assessment of anllcoagulant therapy In left ventricular thrombosis early afteracute myocardial Infarclton Eur Heart ;7: WeInreich 01, Burke IF, Pauletto Fl. Left ventricular mural thrombi complicating acute myocardial Infarcllon: long-term follow-up with serial echocardlography. Ann Intern Med 1984:100: Veterans' AdmInistration Co-operative Study Group. Anllcoagulants In acute myocardial Infarcllon. results of a co-operative climcal trial. lama 1973;225' Drapkin A, Merskey C. Anllcoagulant therapy after acute myocardial infarcllon: relation of therapeutic benefit to patients' age, sex and severity of infarction. lama 1973;225: Levine MN, Raskob G, Hirsh 1. HemorrhagIC complications of longterm anticoagulant therapy. Chest I986;89(suppl): 16S-25S 47. Ezekowitz MD. Acute Infarcllon, left ventricular thrombus and systemic embolization: an approach to management. 1 Am Coil Cardiol 1985;5: Hirsh 1, Hull RD, Raskob GE. Epidemiology and pathogenesis of venous thrombosis 1 Am Coil CardlOl 1986;8.I04B-113B. 49. Kakkar VV, Adams Pc. PreventIve and therapeutic approach to venous thromboembolic disease and pulmonary embolism-can death from pulmonary embolism be prevented') 1 Am Coil Cardiol B-158B. 50 Cooley DA, Hallman GL. Surgical treatment of left ventricular aneurysm: experience with excision of post Infarction lesions in 80 patients. Prog Cardiovasc Dis 1968;11: Kremer P, Fiebig R. Tilsner V, Bleifeld W, Mathey DG. LySIS of left ventricular thrombi with urokinase. Circulation 1985;72: Ledain LD, Ohayon IP. Colle IP, Lorient-Roudaut FM, Roudaut RP, Besse PM. Acute thrombotic obstruction WIth disc valve prostheses: diagnostic considerations and fibrinolytic treatment. 1 Am Coil Cardiol 1986;7: Eigle N. Maurer G. Shah PK. Effect of early systemic thrombolytic therapy on left ventricular mural thrombus formation in acute anterior myocardial infarction. Am 1 Cardiol 1984;54: Fuster V. Gersh Bl, Giuliani ER, Tajlk Al, Brandenburg RO. Frye RL. The natural history of idiopathic dilated cardiomyopathy. Am 1 CardloI1981; Hatle L. Orjavik O. StorsteIn O. Chromc myocardial disease. I ClInical picture related to long term prognosis. Acta Med Scand 1976; 199: , 56. Hamby Rl. PrImary myocardial disease MedICIne 1970;49: Gottdlener IS, Gay la. VanVoorhees L, DlbIanco R, Fletcher RD. Frequency and embolic potential of left ventricular thrombus in dilated cardiomyopathy: assessment by two-dimensional echocardiography. Am 1 Cardiol 1983;52: Demakls IG, Proskey A, Rahimtoola SH, et al. The natural course of alcoholic cardiomyopathy. Ann Intern Med 1974;80: lulian DG, Szekely P. PerIpartum cardiomyopathy Prog Cardiovasc DIS 1985;27: Gottdlener IS. Maron Bl, Schooley RT, Harley lb, Roberts WC, FaUCI AS. Two-dimensional echocardlographic assessment of the IdiOpathic hypereosinophilia syndrome: anatomic basis of mitral regurgltallon and peripheral embolizallon. CirculatIOn 1983;67: Henry WL, Morganroth 1. Pearlman AS, et al. Relation between echocardiographlcally determined left atrial size and atrial fibrillation. Circulation 1976;53: Johnson RA, Palacios I. DIlated cardiomyopathy of the adult (part I) N Engl 1 Med 1982;307: WeInsteIn L, Schlesinger 11. Pathoanatomic, pathophysiologic and climcal correlations In endocarditis. N Engl 1 Med 1974,291: ScheId WM. Pathogeneis and pathophysiology of Infective endocardills. In: Sande MA, Root RK. eds. Contemporary Issues In Infectious Diseases, Volume 2 Endocardills. Edinburgh: ChurchIll Livingstone, Hook EW III, Sande MA. Role of the vegetation In experimental Streptococcus vlridans endocarditis Infect Immun 1974:1 0: LeVison ME. CaITIsoza 1, Tanphaichitra 0, SchICk PK, Rubin W. Effect of aspmn on thrombogenesls and on production ofexperimental aortic valvular Streptococcus vmdans endocarditis In rabbits. Blood 1977, Johnson CE, Dewar HA. Effect of sulfinpyrazone on the development of experimental endocardial vegetations. CardlOvasc Res 1982; 16: Lerner PI. Neurologic complications of Infective endocarditis Med CIIn North Am 1985;69: Pruitt AA. Rubin RH, Karchmer AW, Duncan GW. Neurologic complications of bacterial endocarditis. MediCIne 1978,57: Lerner PI, Weinstein L. Infective endocarditis in the antibiotic era. N Engl 1 Med 1966;274: Garvey Gl, Neu HC Infective endocarditis-an evolving disease: a review of endocarditis at the Columbia-PresbyterIan Medical Center MediCIne 1978;57: Goldman ME. WInters S, RelChstein R. Staville K, Gorlin R. Fuster V Early Identificallon of pallents With native valve endocarditis at high risk for major complications: the value of echocardiography (unpublished data), 73. Alsip SG, Blackston EH, KIrklin lw, Cobbs CG. Indications for cardiac surgery in patients with acllve endocarditis. Am 1 Med 1985;78(suppI6B): WIlson WR, Geraci JE, Danielson GK, et al. Anticoagulant therapy and central nervous system complications In patients with prosthetic valve endocarditis. CirculatIOn 1978;57: O'Brien IT, Geiser EA. Infective endocarditis and echocardiography Am Heart J 1984;108: Stewart la, SIlimperi 0, Harris P, Wise NK, Fraker TO, Kisslo la. Echocardiographic documentation of vegetative lesions in infective endocarditis: climcal implications. Circulation 1980;61: Wong 0, Chandraratna PAN, Wlshnow RM, Dusitnanond V. NImalasurIya A. Climcal Implicallons of large vegetations In InfectIOUS endocarditis. Arch Intern Med 1983;143: Nasser FN. Gura GM, Seward JB. TaJlk Al. Embolism In infective

12 lacc Vol 8, No 6 December B-87B 878 endocardills: IdentificatIon of a hlgh-nsk group by two-dimem,ional echocardlography (abstr). Clfculation 1980;62(suppl 111):11/ Cobb, CG. Living,ton WK. SpecIal problems In the management of Infective endocarditis In: Bls,no AL. ed, Treatment of InfectIve Endocarditis New York. Grune and Stratton Loewe L, Rosenblatt P. Greene HJ. Russel M. CombIned pemclllin and heparin therapy of subacute bactenal endocarditis. JAMA 1944,124: McLean J. Meyer BBM, Griffith JM. Hepann In subacute bactenal endocarditis. JAMA 1941;117: Karchmer AW. DIsmukes WE. Buckley MJ. Austen WG. Late prosthetic valve endocarditis: climcal features InfluencIng therapy. Am J Med : Carpenter JL, McAllIster CK and US Anny Collaborative Group. AntICoagulation In prosthetic valve endocarditb. South Med J 1983; Chesebro JH. Adams pc, Fuster V. Antlthrombotic therapy In patient, with valvular heart di,ease and prosthetic heart valves J Am Coil Cardlol 1986;8:4IB-56B. 85. Kannel WB. Abbott RD. Savage DO. McNamara PM. EpIdemIOlogic features of chromc atnal fibrillation' the Framingham Study. N Engl J Med 1983, KItchIn AH, MIlne JS LongItudinal,tudy of bchemlc heart disease In a randomly selected sample of older population. Br Heart J 1977:39: Wolf PA, Dawber TR, Thomas HE Jr. Kannel WB EpIdemIOlogic a"essment of chromc atnal fibnllatlon and n,k of stroke: the Framingham Study. Neurology 1978;28: HInton RC. Kistler JP. Fallon JT. FriedlIch AL. Fbher CM Influence of etiology of atnal fibnllation on Incidence of systemic embolism. Am J CardlOl 1977:40: Kopecky SL. Ger,h B1, McGoon MD. Holmes DR. WhIsnant JP. Frye RL. The natural history of Idopathlc "lone" atrial fibrillation: a 3 decade population-ba,ed study (abstr) Circulallon (suppl I1I).I1I-1 90 Fairfax AJ. Lambert CD. Leatham A. Systemic embolism in chronic sinoatrial dl'order. N Engl J Med 1976; Forfang K, SImonsen S. Andersen A. Efskind L. Atnal septal defect of secundum type In the mldle aged: clinical result, of surgery and correlation, between symptom, and hemodynamics. Am Heart J 1977;94: Bar-Sela S. Ehrenfeld M. Eliaklm M. Artenal embolism In thyrotoxlcosi, with atnal fibnllation. Arch Intern Med 1981;141: Freeman I, Wexler J AntIcoagulants for treatment ofatnal fibnllation JAMA 1963;184: Bed. JR. Pauker SG AnticoagulatIOn and atnal tibrillation In the bradycardia-tachycardia syndrome Med DeCISIon MakIng 1981; I: Manclm GBJ. Goldberger AL. CardlOversion of atnal fibnllatlon: con'lderation 01 embolization, anticoagulation. prophylactic pacemaker. and long term,ucce" Am Heart J 1982;104: BJerkelund CJ, Ornmg OM. The efficacy of anticoagulant therapy In preventing embolism related to DC electrical conver'lon of atnal fibnllation Am J Cardiol 1969; Dunn M, Alexander J. de SIlva R, HIldner F Antlthrombotic therapy In atnal tibnllatlon. Che,t I986;89(,uppl) 68S-74S. 98 Ikram H, NIxon PG, Arcan T Left atnal function after electrical cardlovefslon to sinw. rhythm. Br Heart J 1968;3U'80-3