JACC Vol. 14, No. 3 September 1989:631-8

631

Echocardiography in Infective Endocarditis: Reassessment of Prognostic Implications of Vegetation Size Determined by the Transthoracic and the Transesophageal Approach

ANDREAS MijGGE, MD, WERNER G. DANIEL, MD, GtJNTER FRANK, MD,

PAUL R. LICHTLEN, MD, FACC

Hannover, West Germany

In 105 patients with active infective endocarditis, disease- associated complications defined as severe heart failure (New York Heart Association class IV), embolic events and in-hospital d&h were correlated to the vegetation size determined by both transthoracic and transesophageal echocardiography. A detailed comparison between ana- tomic and echocardiographlc findings, performed in a subgroup of 80 patients undergoing surgery or necropsy, revealed that true valvular vegetations can be reliably identified by echocardiography in the vast majority of patients; the detection rate was significantly higher for the transesophageal (90%) than for the transthoracic (58%) approach, particularly when infected prosthetic valves were evaluated. However, an accurate echocardiographic differentiation between true vegetations and other en- docarditis-induced valve destruction (ruptured leaflets or chordae) is impossible.

The correlation of vegetation size with endocarditis- associated complications showed that patients with a vege-

Since the initial observations by Dillon et al. (1) and Spangler et al. (2), numerous studies have described the echocardio- graphic findings in patients with infective endocarditis. Echo- cardiography has been generally accepted as the technique of choice for noninvasive diagnosis of infective endocarditis because of its potential for direct visualization of endocardi- tis-induced lesions. Vegetations can be detected when the valve-attached mass reaches a diameter of 22 to 3 mm (1,3).

It has been suggested that the presence of echocardio- graphically identified vegetations may be associated with an

From the Divisions of Cardiology and Thoracic & Cardiovascular Sur- gery. Hannover Medical School, Hannover. Federal Republic of Germany.

Manuscript received December 29, 1988; revised manuscript received March 14. 1989, accepted April 7. 1989.

s for rd: Andreas Miigge. MD, Division of Cardiology. Hannover Medical School, P.O. Box 610180. D-3000 Hannover 61, West Germany.

tation diameter >lO mm had a significantly higher inci- dence of embolic events than did those with a vegetation diameter ~10 mm (22 of 47 versus 11 of 58; p < 0.01). Particularly for patients with mitral valve endocarditis, a vegetation diameter >lO mm was highly sensitive in iden- tifying patients at risk for embolic events. Vegetation size, however, was not significantly different in patients with and without severe heart failure or in patients surviving or dying during acute endocarditis. In addition, no significant correlation was found between vegetation size and location of endocarditis or type of infective organism.

These data suggest that the identification of endocardi- tis-induced vegetations can be improved by transesophageal echocardiography and that patients with a large vegetation at the mitral valve are at increased risk for embolic events. Vegetation size, however, is of minor relevance in relation to the degree of heart failure and patient survival.

(_l Am Co11 Cardiol1989;14:631-8)

impaired prognosis based on a higher incidence of conges- tive heart failure (4-6), death (6,7) or systemic embolism (7-9). However, this issue is still controversial (10). More recently, it has been reported (7,11,12) that prognostic impli- cations are more dependent on vegetation size than on the presence or absence of vegetations alone; thus, Stafford et al. (7) observed embolic events and death only in patients with vegetations >5 mm in diameter. In contrast, in other studies (4,10), a significant correlation between the size of vegetations and the incidence of complications could not be found.

In view of these conflicting reports, we reassessed the prognostic significance of size and location of echocardio- graphically detected vegetations. To optimize the imaging quality of the echocardiograms, all patients were examined by transesophageal echocardiography in addition to the transthoracic M-mode and two-dimensional approach. In a first part of the study, the anatomic abnormalities found at

81989 by the American College of Cardiology 0735.1097/89/$3.50

632 MijGGE ET AL. ECHOCARDIOGRAPHY IN INFECTIVE ENDOCARDITIS

JACC Vol. 14, No. 3 September 1989~631-8

surgery or autopsy were analyzed and correlated with the echocardiographic image of vegetations. Furthermore, the detection rates of vegetations by transthoracic and trans- esophageal echocardiography were compared.

Methods Study patients. Patients meeting the following criteria

were selected: acute infective endocarditis documented by clinical signs or histologic findings, or both. Between 1984 and 1987, 105 consecutive patients (75 male, 30 female; mean age 49.7 * 14.4 years) with clinically (n = 25) or histologi- cally (surgery/autopsy, n = 80) proved infective endocarditis were studied prospectively at the Hannover Medical School. In the 25 patients diagnosed clinically, the clinical criteria for endocarditis were fever (lOO%), a new regurgitant cardiac murmur (100%) and anemia (100%) associated with bacter- emia (80%). In the other 80 patients, histologic specimens were obtained during valve replacement (n = 62) or at necropsy (n = 18). Except for two patients with histologi- cally documented endocarditis, all patients undergoing surgery or autopsy also had clinical signs of infective en- docarditis. Patients with vegetations diagnosed by echocar- diography alone, but with no clinical signs and no histologic proof of endocarditis were not included. In the 105 patients, a total of 116 valves were diseased (90 native valves [47 aortic, 39 mitral and 4 tricuspid] and 26 prosthetic valves [ 14 in the aortic and 12 in the mitral position]).

typical vegetations (valve destruction with thromboticl infectious material containing platelets, fibrin, erythrocytes and granulocytes), 2) bacterial colonies (a thin layer of thromboticlinfectious material), 3) valve destruction (without thrombotic material as judged by eye inspection and without valve ruptures), 4) ruptured chordue tendineae (with or without macroscopic visible thrombotic material), 5) ruptured valve leaflets (with or without thrombotic material), or 6) no vegetations (although there was histologically proved en- docarditis). The anatomy could be evaluated in 80 patients with 91 diseased valves. Acute or postinflammatory lesions proved by histologic examination were present on all 91 valves.

Echocardiographic examination. Transthoracic M-mode and two-dimensional echocardiographic studies were per- formed by standard techniques with use of a 2.25 or 3.5 MHz phased array transducer system. Transesophageal two- dimensional echocardiographic studies were performed with a 3.5 or 5.0 MHz phased array transducer attached to the tip of a modified gastroscope (Diasonics Echoscope, Diasonics Cardiollmaging Inc.; Hewlett-Packard model 21362A). Pa- tients fasted for ~4 h and received a local pharyngeal anesthetic agent as the only premeditation. Informed con- sent was obtained from all patients. The investigations were carried out without complications with the patient in a left lateral decubitus position. AU transesophageal examinations were performed primarily for diagnostic reasons and were usually completed within 5 min.

Detition of complications. Patients underwent a detailed Each cardiac valve was evaluated for thickening, calcifi- physical examination including a personal interview, and cation, prolapse of the leaflets and, in particular, for the their charts were carefully reviewed to detect episodes of presence or absence of “typical vegetation,” defined as a pulmonary or systemic arterial embolism. A sudden appear- “sessile” or “pendulating” valve-attached mass. A vegeta- ance of neurologic disorders suggestive of cerebral embolism tion was considered as “definite” when shaggy echoes in the was observed in 28 patients. The diagnosis of cerebral M-mode study and a corresponding mass without restricted embolism was based on computed tomographic findings valve motion in the two-dimensional echocardiogram were (showing a nonhemorrhagic stroke) in 20 patients and could found. A vegetation was considered as “possible” when the be proved by autopsy in 3 more patients; in the remaining 5 findings were suggestive of a vegetation, but no clear differ- patients, the diagnosis of cerebral embolism was based on entiation from a preexisting lesion with thickened or calcified clinical signs alone. In addition, embolism to peripheral leaflets was possible. A “sessile” vegetation had to be organs was documented by autopsy or chest X-ray exami- completely attached to the valve, whereas a “mobile” nation in five patients (lung in one, spleen and kidney in one, vegetation showed a pedunculating part prolapsing into the multiple organs in three). In total, 33 patients had suffered ventricle or atrium. The size of vegetations was measured in from a major embolic event. Cutaneous manifestations of various planes in the precordial and transesophageal endocarditis (Osler nodes, petechiae, Janeway lesions, splin- echocardiogram, and the maximal diameter was used for ter hemorrhages) were not considered to be due to embolic subsequent analysis. The mean size of vegetations was 10.6 events because their pathogenesis is still questionable and mm (range 3 to 28); therefore, vegetations >lO mm in immunologically mediated mechanisms may be involved diameter were defined as “large,” and those 110 mm in (13). diameter were defined as “small.”

Twenty-four patients experienced severe heart failure de- fined as class IV according to the New York Heart Associa- tion classification. Death, defined as in-hospital mortality due to endocarditis complications, occurred in 18 patients.

Definition of anatomic findings. The macroscopic in situ findings at autopsy or surgery were classified as follows: 1)

Statistical analysis. For statistical analysis, the chi-square test with Yates’ correction for small numbers was used. The association between vegetation size and incidence of embo- lism, death or severe heart failure was analyzed with use of analysis of variance (ANOVA, SPSS, version X). In patients with more than one diseased valve, the largest vegetation

JACC Vol. 14, No. 3 MtiGGE ET AL. 633 September 198963 1-8 ECHOCARDIOGRAPHY IN INFECTIVE ENDOCARDITIS

Table 1. Comparison Between Echocardiographic and Anatomic Findings in 80 Patients with 91 Diseased Valves (histologically proved endocarditis)

Anatomic Findings “Definite” Veg “Possible” Veg No Veg

(valves) (valves) (valves)

Typical vegetations Bacterial colonies Valve destruction Rupture of chordae tendineae

With vegetations Without vegetations

Rupture of’ valve leaflets With vegetations Without vegetations

No vegetations

52 1 1 7 2 2 5 2 -

4 - - 2

6 - 5 - -

1 1

Time interval between the echocardiographic examination and the in situ inspection: 17 T 2 days. Veg = vegetation.

was used for subsequent analysis. Patients with no vegeta- tion evident on the echocardiogram were evaluated in the group of patients with a small vegetation. A p value ~0.05 was considered as significant. Mean values -t SEM are given unless otherwise noted.

The association between embolic episodes (Emb) and echocardiographically detected “large” vegetations (Veg) >I0 mm in diameter was determined by calculating sensi- tivity, specificity, positive and negative predictive values as follows:

Sensitivity = Emb with Veg > 10 mm

Emb with Veg >lO mm t Emb with Veg ~10 mm’

Specificity = No Emb with Veg 510 mm

no Emb with Veg 5 10 mm + no Emb with Veg > IO mm ’

Predictive value (+) = Emb with Vea > 10 mm

Emb with Veg >lO mm t no Emb with Veg >I0 mm

Predictive value (-) = No Emb with Veg ~10 mm

no Emb with Veg ~10 mm + Emb with Veg ~10 mm ’

Results Comparison of echoeardiographic and anatomic findings

(Table 1). In 80 patients with endocarditis of 91 valves proved by surgery or autopsy, the macroscopic appearance of the diseased valves could be compared with the findings obtained by echocardiography. In these 80 patients, the echocardiographic studies were performed 17 k 2 days before surgery or autopsy. In 62 (97%) of 64 diseased valves with anatomically documented vegetations, echocardiog- raphy detected a “definite” vegetation (considered as a “typical” vegetation [n = 541 or a vegetation associated with ruptured chordae tendineae [n = 41 or ruptured leaflets [n = 61). However, echocardiography could not differentiate pre- cisely between vegetations and rupture of leaflets or chordae tendineae found in seven valves without additional vegeta- tions at surgery or autopsy. In 18 other cases where the involved anatomic site revealed “bacterial colonies” (n = 11) or “valve destruction” (n = 7), the echocardiogram showed “definite” (n = 12) or “possible” (n = 4) vegeta- tions or none (n = 2). The one patient with a “definite” vegetation on the echocardiogram, but no corresponding findings during anatomic inspection, had suffered from brain embolism in the interval between the echocardiographic study and surgery.

Comparison of transthoracic and transeaophageal echocar- diography (Table 2). In the same subgroup of 80 patients with histologically proved endocarditis, the transthoracic and the transesophageal echocardiographic findings regard- ing the detection of a vegetation or valve destruction, or both, were compared. Transthoracic echocardiography was able to visualize a “definite” vegetation in 53 (58%) of 91 infected valves; 17 valves (19%) were considered as showing a “possible” vegetation. In contrast, transesophageal echo- cardiography visualized a “definite” vegetation in 82 (90%) of 91 valves; on 6 valves (6%) findings were suggestive of a vegetation (p < 0.05). The relatively low detection rate of the

Table 2. Detection of Vegetations by Transthoracic and Transesophageal Echocardiography in 80 Patients With 91 Diseased Valves

Transthoracic Echo Transesophageal Echo

Valve n “Definite” Veg “Possible” Veg “Definite” Veg “Possible” Veg

Aortic valve 37 25 8 35 2 Mitral valve 30 20 7 28 2 Tricuspid valve 2 2 - 2 - Prosthetic valve 22 6 2 17 2

Total (%) 91 53 (58%) 17 (19%) 82 (90%) 6 16%)

Veg = vegetations: Echo = echocardiography.

634 MiiCiGE ET AL. JACC Vol. 14, No. 3

ECHOCARDIOGRAPHY IN INFECTIVE ENDOCARDITIS September 1989531-8

30 .

25.

20-

15-

10-

5-

. . . . I.

. .

. .

. .

.

. .

.

.A.

.T-z z.

_t .z. .-.

.

. . .

‘[ . .

: 1.

0 .- pW3.

k81 n=24

NYHA I-111 NYHA IV No embolism embolism

Figure 1. Comparison of echocardiographically measured vegeta- Figure 3. Comparison of echocardiographically measured vegeta- tion size in 105 patients with infective endocarditis and class I-III or tion size in 105 patients with infective endocarditis with and without class IV heart failure. Vegetation size zero represents patients embolic episodes. Vegetation size zero represents patients without without echocardiographically detectable vegetations. echocardiographically detectable vegetations.

transthoracic technique is mainly due to the high number of prosthetic valves in which the difference between the two echocardiographic approaches was most striking.

Prognostic implications of vegetation size (Fig. 1 to 7). In the total study group of 105 patients, the size of vegetations in the echocardiogram was compared in patients with and without severe heart failure (Fig. l), in patients surviving or dying during the acute phase of endocarditis (Fig. 2) and in patients with or without major embolism (Fig. 3). Only the

Figure 2. Comparison of echocardiographically measured vegeta- tion size in 105 patients with infective endocarditis dying or surviv- ing during the acute phase of the disease. Vegetation size zero reDresents patients without echocardiographically detectable vege-

301 .

25. . .

20.

15.

lo-

5-

. . :‘: . . . . . . . .

SL -I -em -am e . .

.-.

.-.a

Z’ ”

.

. . . . . . .

:: . I

O- -_ pZ”S. “e

rk87 kl8

swvlvon halls

30

25

20

15

10

5

0

.

. . . . .

. . . T .&e&o

z-x. :: . ..U

:ZS . . . . . I . . - . . . . .

n=72

pzo.oz

. . . .

n=33

latter comparison showed a significant difference: 47 patients with a large vegetation (>lO mm in diameter) had a higher incidence of embolic events (46.8%) than that of the 49 patients with a small or the 9 patients with no vegetation (18.9%; p < 0.05). In 16 patients, the echocardiographic examination was performed after one or more embolic events; in the remaining 17 cases, embolism occurred after the initial echocardiographic study. Because an embolic event might reduce the size of vegetations, statistical analy- sis was repeated after exclusion of the 16 patients who had an embolic episode before the echocardiographic study. This second analysis, including the remaining 89 patients, also revealed that the incidence rate of embolism was lower (6%) in the 50 patients with a small vegetation than the rate (36%) in the 39 patients with a large vegetation (p < 0.05). The average size of vegetations in the 17 patients with an embolic event after the echocardiographic examination was 14 + 7 mm compared with 9 + 5 mm in the 72 patients without embolism. Sensitivity, specificity and positive and negative predictive values of a vegetation size > 10 mm for detection of patients with embolism are shown in Table 3.

When vegetation size was analyzed according to the various underlying bacteria isolated in at least two different blood cultures, the data of 97 patients could be evaluated (8 patients had blood cultures with more than one or with different bacteria and were therefore excluded from the analysis). Statistical evaluation showed no significant differ- ences between the bacteriologic groups (Fig. 4).

The mean size of vegetation was compared in all 116 diseased valves. The mean vegetation size in native aortic and mitral valves did not differ significantly (Fig. 5). In contrast, vegetations of the tricuspid valve were significantly

JACC Vol. 14. No. 3 MiKiGEETAL. 635 September lYW631-8 ECHOCARIXOGRAPHY IN INFECTIVE ENDOCARDITIS

Table 3. Association Between Embolic Events and Vegetation Size >I0 mm in 105 Patients With Infective Endocarditis

Patients Without Emb Patients With Total Collective Before Echo Mitral Valve Endo

(tl = 105) (n =z 89) (n = 31)

Emb with Veg >lO mm 22 14 II No Emb with Veg >I0 mm 2s 25 7 No Emb with Veg ~10 mm 47 47 13 Emb with Veg ~10 mm I1 3 0 Sensitivity 66.7% 82.3% 100% Specificity 65.3% 6S.X 65.0% Positive predictive value 46.8% 35.9% 61.1% Negative predictive value 81.0% 94.c% 100%

Echo = echocardiography: Emb = cmbolic episodes: Endo = endocarditis: Veg = vegetations.

larger, and vegetations of a prosthetic device were signifi- cantly smaller on the average than were vegetations of native aortic and mitral valves (variance analysis: p < 0.01) (Fig. 5).

The incidence of embolism was evaluated separately for the various locations of the diseased valves as well as for native and prosthetic valves (Fig. 6). The evaluation in- cluded 45 patients with native aortic endocarditis (11 with embolism), 31 patients with native mitral valve endocarditis (11 with embolism) and 25 patients with prosthetic valve endocarditis (10 with embolism): the 4 patients with native tricuspid valve endocarditis (1 patient with embolism) were excluded because of the small number of cases. There was no significant difference in risk of embolic episodes between patients with a small or large vegetation, as long as en-

Figure 4. Comparison of vegetation size in 97 patients with various bacteria isolated in at least two different blood cultures. Values represent mean diameter *I SD. Variance analysis showed no significant differences between the groups. Str. virid. = Streptococ- cus viridans; Strepto. = Streptococci other than Streptococcus viridans; S. aureus = Staphylococcus aureus; others = other rare bacteria; sterile = sterile blood cultures.

Str.virid. Strepto. Saureus Others Sterile

Bacteriology

docarditis was confined to a native aortic valve or a pros- thetic valve (in either the aortic or the mitral position). However, among patients with a vegetation at the mitral valve, those with a large vegetation had a significantly higher incidence of embolism than did patients with a small vege- tation (p < 0.001). The association between vegetation size and embolic risk in patients with mitral valve endocarditis is included in Table 3; sensitivity and negative predictive value reached a level of 100% in this series of patients.

Patients with a “mobile” vegetation had a significantly higher itwidence ofembolic episodes than did patients with a “sessile” vegetation (Fig. 7). However, the mobility of vegetations was not a further independent variable to dis- criminate the risk of patients for embolism because 40 (85%)

Figure 5. Comparison of vegetation size in aortic (AV), mitral (MV), tricuspid (TV) and prosthetic (PV) valve endocarditis (total group of 105 patients with 116 diseased valves). Values represent mean diameter tl SD. Variance analysis showed a significant difference among the four subgroups (p < 0.01) indicating that tricuspid valve vegetations were larger and prosthetic valve vegeta- tions were smaller than in the other valve subgroups.

w TV

Heart valves

PV

636 MirGGE ET AL. JACC Vol. 14, No. 3 ECHOCARDIOGRAPHY IN INFECTIVE ENDOCARDITIS September 19896316

p<o.o01

AV MV

Heart Valves

PV

Figure 6. Incidence of embolism in 45 patients with native aortic (AV), 31 patients with native mitral (MV) and 25 patients with prosthetic valve (PV) endocarditis and vegetations >lO mm (black bars) or 5 10 mm (shaded bars) in size.

of the 47 patients with a “large” vegetation also had a “mobile” vegetation whereas only 28 (48%) of the 58 patients with a small vegetation had a mobile vegetation (p < 0.001).

Discussion Identification of vegetations by echocardiography. Echo-

cardiography is the only noninvasive method available today allowing direct visualization of endocarditis-induced lesions. The detection rate for valvular vegetations depends on the approach used. Use of M-mode echocardiography alone allows the identification of vegetations in about 57% of cases (evaluation of 14 studies with 332 positive results in 583 patients (3,5,6,12,14-23)). The combined use of M-mode and two-dimensional imaging increases the detection rate to about 73% (550 positive results in 755 patients (4,7- 11,24-34)). In our subgroup of 80 patients with endocarditis proved at surgery or autopsy, a “definite” or “possible” vegetation was found in about 77% (70 of 91 diseased valves) with use of transthoracic M-mode and two-dimensional echocardiography. Recently, the diagnostic value of echo- cardiography has been further improved by the introduction of the transesophageal approach. This technique is of par- ticular importance in patients in whom transthoracic echo- cardiography fails to provide an adequate imaging quality, as in patients who have advanced obesity, chest deformity or emphysema, or who are seen early after thoracic surgery or under artificial ventilation. In addition, the transesophageal technique allows an improved evaluation of prosthetic valves and provides an increased detection rate for perival- vularlesions such as abscesses (35-38). In the present series,

p-O.042

sessile

Mobility of vegetations

Figure 7. Incidence of embolic episodes in 68 patients with a “mobile” and 37 with a “sessile” vegetation. black area = embo- lism.

the transesophageal approach identified a vegetation or endocarditis-induced valve destruction, or both, in about %% (88 of 91 diseased valves). A similar high detection rate using transesophageal echocardiography was recently re- ported by Erbel et al. (39).

Advantages of transesophageal echocardiography. Be- cause analysis of the impact of vegetation size on prognosis in patients with infective endocarditis requires optimal im- aging quality of valve structures, this study involved transe- sophageal echocardiography in addition to the precordial approach. Imaging of the heart from the esophagus provides a clear visualization of virtually all cardiac structures that is not disturbed by bones, lungs and the precordial chest wall (40,41). In addition, because of the close relation between the esophagus and the heart, higher transducer frequencies can be applied, thereafter leading to improved resolution and more accurate measurement of vegetation size. As shown by earlier studies from our laboratory (35,42), transesophageal echocardiography, although accompanied by minor discom- fort, can also be performed easily in conscious patients at almost no risk.

Comparison with anatomic findings. The comparison be- tween the anatomic findings documented at surgery or autopsy and the echocardiographic results showed that vegetations attached to diseased valves can be identified on the echocardiogram in the vast majority of patients. A reliable differentiation between vegetations and other en- docarditis-induced lesions, such as rupture of leaflets or the valve apparatus, is generally not possible. This point is only of minor clinical relevance regarding the echocardiographic assessment of the diagnosis; it is, however, of major impor- tance concerning the associated embolic risk because rup- tured leaflets, in contrast to vegetations, usually do not

JACC Vol. 14, No. 3 MirGGE ET AL. 637 September 1989:631-8 ECHOCARDIOGRAPHY IN INFECTlVE ENDOCARDITIS

embolize. In this regard, previous studies did not provide sufficient information about the true anatomic findings. This anatomic inhomogeneity of echocardiographically detected “vegetations” in earlier studies may contribute to the con- troversies concerning the prognostic implications.

Prugnos.tic implications. The present study clearly shows that vegetations >lO mm in diameter are associated with a higher incidence of embolic episodes. Although this associ- ation was significant in our total study group, further detailed analysis revealed that it was based on the subgroup of patients with a vegetation on a native mitral valve. Earlier studies that did (7,8,9,12,26) or did not (4,lO) find an in- creased embolic risk in patients with vegetations focused only on the presence and size of vegetations but not on their location. Our findings may therefore partially explain previ- ous discrepancies concerning the prognostic implications of echocardiographically detected vegetations.

In all studies published to date, a certain percentage of patients underwent echocardiographic examination after an embolic event had already taken place: this increases the number of patients with embolic events in the subarour, with a small (or-even no) vegetation and may obscure; possible significant difference in the embolic risk. In the present study, we classified our patients with embolic events into those with embolism before and those with embolism after the echocardiographic examination. This classification per- mitted improved sensitivity and negative predictive value of echocardiographically detected vegetations > 10 mm in rela- tion to identification of patients with increased embolic risk.

of the present study to address these questions. Second, it is well known (43) that some embolic events during the course of infective endocarditis remain clinically silent. Therefore, the true number of embolic events remains unknown in this as well as in comparable studies. Furthermore, the natural course of patients with a large vegetation can only be partially assessed in studies of this type. A certain percent- age of patients undergo surgery before a potential embolic event because of refractory septicemia or advanced heart failure. Thus, some patients with a possible late embolic episode may be missed.

Clinical implications. Although the present study as well as previous reports cannot avoid certain limitations, our results indicate that the echocardiographic identification of endocarditis-induced vegetations can be improved by the transesophageal approach and that patients with a vegetation > 10 mm at the mitral valve are exposed to an increased risk for embolic episodes. In these patients, we recommend early surgery independent of the degree of heart failure and responsiveness of septicemia to medical therapy.

We thank R. Schtitzenmeister for excellent technical assistance and A. Schafft for careful preparation of the figures.

References I. Dillon JC, Feigenbaum H, Konecke LL, Davis RH, Chang S. Echocar

diographic manifestations of vaHular vegetations. Am Heart J 1973; 86:69&704.

Other endocarditis-associated complications, such as severe heart failure and death, showed no significant depen- dence on vegetation size. These results are in agreement with two previous reports (4,10), but Wong et al. (28) found an increased need for surgery in patients with a large vegetation and recommended early surgery in patients with vegetations > 10 mm. However, their study included only 31 patients studied by two-dimensional echocardiography, and most patients had mitral and tricuspid valve disease (n = 22). In onr study group, there was -an approximately equal distribution of infected aortic and mitral valves in the pa- tients with large or small vegetations (Fig. 5).

2. Spangler RD. Johnson MC, Holmes J, Blotmt G. Bchocardiographic demonstration of bacterial vegetations in active infective endocarditis. 3 Clin Ultrasound 1973;1:1268.

3. Roy P, Tajik AJ, Guiiiani ER, Schattenberg ‘IT, Gau GT, Frye RL. Spectrum of echocardiographic findings in bacterial endocarditis. Circu- lation 1976:53:474-82.

4. Stewart JA, Silimperi D, Harris P, Wise NK, Fraker TD, Kisslo IA. Echocardioeranhic documentation of venetative lesions in infective en- docarditis: &ical implications. Circulatkn 1980;61:37&80.

5. Hickey AJ, Wolfers I, Wilcken DEL. Reliability and clinical relevance of detection of vegetations by echocardiography in bacterial endocarditis. Br Heart J 1981;46:624-8.

6. Come PC, Isaacs RE, Riley MF. Diiostic accuracy of M-mode echo- cardiography in active infective endocarditis and prognostic implications of ultrasounddetectable vegetations. Am Heart J 1982;103:839-47.

7. Stafford WJ, Petch 1, Radford DJ. Vegetations in infective endocarditis: clinical relevance and diagnosis by cross sectional echocardiography. Br Heart J 1985:53:318-3.

Our data indicate that there is no clear association be- tween vegetation size and the common underlying infective organisms (Fig. 4). However, because specific organisms like fungi were rare in the present series, this lack of association may not be true for all bacteria.

Lbnitations of the study. This study like earlier reported studies, has certain limitations that seem to be unavoidable. First, there is a remaining uncertainty regarding the true vegetation size when measurements are based on echocar- diographic images. The additional use of transesophageal echocardiography improves the detection rate of vegetations and, because of better visualization, probably also improves the accuracy of measurements. However, it was not the aim

8. Wann LS, Haham CC, Dillon JC, Weyman AE, Feigenbaum H. Compar- ison of M-mode and cross-sectional echocardiography in infective en- docarditis. Circulation 1979;60372&33.

9. Mintz GS, Kotler MN, Segal BL, Parry WR. Survival of patients with aortic valve endocarditis: the prognostic implications of the echocardio gram. Arch Intern Med 1979;139:862-6.

10. Lutas EM, Roberts RB, Devereux RB, Prieto LM. Relation between the presence of echocardiogtaphic vegetations and the complication rate in infective endocarditis. Am Heart J 1986;112:107-13.

11. Bardy GH, Talano IV, Reisberg B, Lesch M. Sensitivity and specificity of echocardiography in a high-risk population of patients for infective

638 MirGGE ET AL. ECHOCARDIOGRAPHY IN INFECTIVE ENDOCARDITIS

JACC Vol. 14. No. 3 September 1989:631-S

12.

13.

14.

15.

16.

17.

18.

19.

endocarditis: significance of vegetation size. J Cardiovasc Ultrasonogr 1983:2:23-7.

Sheikh MU, Covamtbias EA. Ah N, Lee WR, Sheikh NM, Roberts WC. M-Mode echocardiographic observations during and after healing of active bacterial endocarditis limited to the mitral valve. Am Heart J 1981;101:37-45.

Gahl K, Mtlgge A, Nonnast-Daniel B. Das klinische Bild der infektiiisen Endokarditis. In: K Gahl, ed. Infektiose Endokarditis. Steinkopff Verlag. Darmstadt, 1984:40-68.

Warm LS, Dillon JC, Weyman AE, Feigenbaum H. Echocardiography in bacterial endocarditis. N Engl J Med 1976;295:135-9.

Jenzer HR, Follath F, Grade1 S, Amann FW. Die Echokardiographie bei infektiiiser Endokarditis. Schweiz Med Wochenschr 1977;107:1572-3,

Pratt C, Whitcomb C, Neumann A, Mason DT, Amsterdam EA, DeMaria AN. Relationship of vegetations on echogram to the clinical course and systemic emboli in bacterial endccarditis. Am J Cardiol 1978;41:384.

Young JB, Welton D, Quinones MA, lshimori T, Alexander JK, Miller RR. Prognostic significance of valvular vegetations identified by M-mode echocardiography in infective endocarditis. Circulation 1978;58(suppl II):II-41.

Sheikh MU, Ali N. Covarrubias EA, Fox LM, Motjaria M, Deja 1. Right-sided infective endocarditis: an echocardiographic study. Am J Med 1979;66:283-7.

Davis RS, Strom IA, Frishman W, et al. The demonstration of vegeta- tions by echocardiography in bacterial endocardttis. Am J Med 1980; bY:57-63.

20. Sheikh MU, Covamtbias EA, Ah N, Sheikh NM, Lee WR, Roberts WC. M-Mode echocardiographic observations in active bacterial endocarditis limited to the aortic valve. Am Heart J 1981;102:6&75.

21. Assey MF, Usher BW. Echocardiography in diagnosing and managing aortic valve endocarditis. South Med J 1981:74:55865.

22. Markiewiez W, Moscovitz M. Edoute Y, Finkelstein R, Raz R. Prognos- tic implications of detecting vegetations by M-mode echocardiography. Cardiology 1983;70:194-8.

23. Daniel WG, Migge A, Gahl K. Lichtlen PR. Echokardiographische Diagnostik. In Ref 13, 108-32.

24. Martin RP, Meltzer RS, Chia BL, Stinson EB, Rakowski H, Popp RL. Clinical utility of two dimensional echocardiography in infective en- docarditis. Am J Cardiol 1980;46:379-85.

25. T&da P, Skalsky L, Bommer WJ, et al. Two-dimensional echocatdiog- raphy in infective endocarditis: early and late diagnosis, therapeutic and prognostic implications. Circulation 1980;62(suppl III):III-99.

26. Nasser FN, Gura GM, Seward JB, Tajik AJ. Embolism in infective endocarditis: identification of high-risk group by two-dimensional echo- cardiography. Circulation 198ft:62(suppl III):IH-100.

27. Melvin ET, Berger M. Lutzker LG, Goldberg E, Mildvan D. Noninvasive methods for detection of valve vegetations in infective endocarditis. Am J Cardiol 1981;47:271-8.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

Wong D, Chandraratna PAN, Wishnow RM, Dusitnanound V, Nimala- suriya. A. Clinical implications of large vegetations in infective endocardi- tis. Arch lntem Med 1983:143:18767.

O’Brien IT, Geiser EA. Infective endocarditis and echocardiography. Am Heart J 1984;108:386-94.

Horstkotte D, Bircks W, Loogen F. Infective endocarditis of native and prosthetic valves: the case for prompt surgical intervention? A retrospec- tive analysis of factors atfecting survival. Z Kardiol 1986;75(suppl 2): 16882.

Sareli P, Klein HO, Schamroth CL, et al. Contribution of echocardiog- raphy and immediate surgery to the management of severe aortic regur- gitation from active infective endocarditis. Am J Cardiol 3986;57:413-8.

G&ton LE, Sieger RJ, CriIey JM. Natural history of tricuspid valve endocarditis: a two dimensional echocardiographic study. Am J Cardiol 1982;49:1853-9.

Berger N, Dehin LA, Jelveh M. Two-dimensional echocardiographic findings in tight-sided infective endocarditis. Circulation 1980;61:855-61.

Nakamura K, Koyanagi H, Hirosawa K. Spectrum of the infective endocarditis in the past five years. Jpn Circ J 1982;46:352-61.

Daniel WG, Schriider E, Milgge A, Lichtlen PR. Transesophageal echo- cardiography in infective endocarditis. Am J Cardiac Imaging 1988; 2:78-85.

Daniel WG, Nellessen U, Schroder E, Nikutta P, Nonnast-Daniel B, Milgge A. Transesophageal echocardiography as the method of choice for the detection of endocarditis-associated abscesses. Circulation 1986; 74(suppl II):II-55.

Erbel R, Rohmann S, Drexler M, Mohr-Kahaly S, Meyer J. Diagnostic value of transesophageal echocardiography in infective endocarditis. Ctrculatton 1986;74(suppl 11):11-55.

38. Daniel WG, Miigge A, Frank G. Improved diagnosis of prosthetic valve malfunction by transesophageal echocardiography. Circulation 1988; 78(suppl II):II-606.

39. Erbel R. Rohmann S, Drexler M, et al. Improved diagnostic value of echocardiography in patients with infective endocarditis by transesoph- ageal approach: a prospective study. Em Heart J 1988;1:43-53.

40. Seward JB, Khandheria BK. Oh JK, et al. Transesophageal echocardiog- raphy: technique, anatomic correlations, implementation, and clinical applications. Mayo Clin Proc 1988;63:6#-80.

41. Mitchell MM, Sutherland GR, Gussenhoven El, Taams MA, Roelandt JRTC. Transesophageal echocardiography. J Am Sot Echo 1988; 1~362-77.

42. Daniel WG, Milgge A, Schroder E, WenzlatTP, Grate 1. Transesophageal echocardiography in clinical cardiology-indications, practicability and risk. Circulation 1988;78(suppl 11):11-Z%.

43. Weinstein L, Rubin RH. Infective endocarditis 1973. Prog Cardiovasc Dts 1973;16:23%74.