Author + information
- Received February 26, 2001
- Revision received October 24, 2001
- Accepted January 16, 2002
- Published online April 3, 2002.
- Catherine Graupner, MD*,
- Isidre Vilacosta, MD*,* (, )
- JoséAlberto SanRomán, MD†,
- Ricardo Ronderos, MD‡,
- Cristina Sarriá, MD§,
- Cristina Fernández, MD*,
- Ricardo Mújica, MD‡,
- Olga Sanz, MD†,
- Juan Victor Sanmartín, MD§ and
- Angel González Pinto, MD∥
- ↵*Reprint requests and correspondence:
Dr. Isidre Vilacosta, Instituto de Cardologia, Hospital Universitario de San Carlos, 28040, Madrid, Spain.
Objectives This prospective study was designed to assess the current clinical course, risk factors, microbiologic profile and echocardiographic findings of patients with left-sided endocarditis and perivalvular complications.
Background Periannular complications worsen the prognosis of patients with endocarditis. The relation between these complications and the clinical and microbiologic data has not been clearly defined.
Methods In this clinical cohort study, 211 patients with left-sided endocarditis, according to the Duke criteria, were prospectively recruited. All patients underwent conventional and transesophageal echocardiography. The mean follow-up interval was 151 days.
Results Perivalvular complications were detected in 78 patients (37%). The incidence of periannular extension of infection in native and prosthetic valves was 29% and 55%, respectively. The presence of prosthesis (relative risk [RR] 1.88, 95% confidence interval [CI] 1.35 to 2.64) and previous endocarditis (RR 1.78, 95% CI 1.16 to 2.7) were the only pre-existing heart conditions associated with perivalvular complications. Aortic infection (RR 1.8, 95% CI 1.23 to 2.66) and the development of atrioventricular (AV) block (RR 2.55, 95% CI 1.91 to 3.41) were related with the existence of these complications. Coagulase-negative staphylococci were very common in patients with perivalvular complications (RR 1.77, 95% CI 1.21 to 2.59), and small vegetations were more frequent in these patients (RR l.45, 95% CI 0.95 to 2.22). An operation was more frequently performed in patients with perivalvular complications, but mortality was similar in patients with and without these complications.
Conclusions Aortic infection, prosthetic endocarditis, new AV block and coagulase-negative staphylococci were independent risk factors of periannular complications. The period between symptom onset and diagnosis, the incidence of pericardial effusion and persistent signs of infection were similar between patients with and without perivalvular complications. Patients with perivalvular complications did not demonstrate a difference in the presence or size of vegetations or the frequency of embolism. An operation was more frequently performed in these patients, but mortality was similar in both groups.
Periannular extension of infection is one of the most fearful complications in patients with infective endocarditis. It is well recognized that this particular condition casts a shadow over the prognosis of patients with endocarditis (1–9). Accurate detection and delineation of perivalvular complications are crucial in overall patient management and surgical guidance. Clinical variables for the diagnosis of these complications are inadequate (2). Until relatively recently, transthoracic echocardiography (TTE) was the mainstay of imaging in patients with infective endocarditis. Now, transesophageal echocardiography (TEE) is the imaging method of choice for the assessment of patients at risk of periannular extension of endocarditis (10).
Although several groups have contributed to our understanding of some clinical and echocardiographic findings in patients with endocarditis and periannular complications, there are many limitations—the studies were retrospective (1,2,7,11); they included a small number of patients (2,6,10–14); TEE was not routinely used (2,6,12–14); and some series had a selection bias, as they included patients who had an operation or died (1,2,7,13,14)—and the relationship between these complications and the clinical data has not been clearly defined.
The aim of this prospective study was to assess the current clinical course, predisposing risk factors, microbiologic agents and echocardiographic findings of a large group of patients with left-sided endocarditis and perivalvular complications anatomically proven or demonstrated by TEE.
The clinical cohort study consisted of 211 patients with left-sided endocarditis who were prospectively recruited between April 1996 and June 2000. All patients with the diagnosis of endocarditis in the five centers were included in the study. There were 135 men and 76 women, with a mean age of 57 ± 15 years. According to the Duke criteria (15), 193 patients (91.5%) had definite and 18 patients (8.5%) had possible infective endocarditis. Sixty-six patients (31%) had infections of prosthetic valves, and 34 had early and 31 had late prosthetic valve endocarditis. These patients were clinically followed for 1 to 942 days (mean follow-up interval 151 days). Since the beginning of the study, all historic, clinical, laboratory, echocardiographic and microbiologic data were recorded and entered into a computer data base. At operation or necropsy, the existence of periannular complications was specifically sought.
Definition of terms
At necropsy or operation (by direct inspection), an abscess was defined as a region of necrosis containing purulent material that had no communication with the cardiovascular lumen. When a perivalvular cavity that had communication with the cardiovascular lumen was found, it was called a “pseudoaneurysm” (Fig. 1) (10,13). A fistula was defined as an anatomic communication between two neighboring heart cavities. Adequate antibiotic therapy was defined as high intravenous doses of antibiotic combinations known to be bactericidal in vitro against the isolated microorganisms; for patients whose cultures were negative, empiric antibiotic regimens were chosen. Early prosthetic valve endocarditis was defined as developing one year or less after surgery, and late prosthetic valve endocarditis as occurring over one year after surgery. Persistent infection was defined as fever lasting for more than seven days after starting adequate antibiotic treatment, septic shock or positive blood cultures after 48 h of the initiation of adequate antibiotic therapy.
All patients underwent both modalities of echocardiographic imaging (TTE and TEE). The echocardiographic studies were done with commercially available instruments (Toshiba [Tokyo, Japan], Advanced Technology Laboratories [Best, The Netherlands] and Hewlett-Packard [Palo Alto, California]). The TEE studies were performed with a 5-MHz transducer. The TEE study was performed after the standard procedure, and the probes used were monoplane in 11 patients, biplane in 87 patients and multiplane in 113 patients. The images were stored on hard copies, as well as on videotape, for analysis. During follow-up (mean 14 ± 10 days), TEE was repeated in 118 patients. A valvular vegetation was defined as a localized mass of shaggy echoes adherent to a valve leaflet or prosthetic material with a distinctive erratic motion. M-mode tracings of the masses were performed to better define the motion of these structures. Nonspecific valvular thickening was not interpreted as a vegetation. The morphologic characteristics of vegetations were analyzed by TEE. The vegetative mass was measured in various planes, and the maximal diameter was used for subsequent analysis; in cases of multiple vegetations, the largest vegetation was measured. Once a periannular lesion was visualized, it was further defined by subtle manipulations of the transducer. These were aimed at optimizing the size of the cavity, defining its location in relation to adjacent structures and determining whether it communicated with other chambers. Blood in the lesion was assessed by color Doppler flow imaging. The largest diameter and maximal area of pseudoaneurysms and abscesses were determined by planimetry. A perivalvular abscess was defined as a circular or more irregularly delineated region (>10 mm) of reduced echo density, without flow in its interior (Fig. 2) (5,16). When flow was detected into these perivalvular cavities, the lesion was defined as a pseudoaneurysm, and then it appeared as a pulsatile echo-free perivalvular pouch (Fig. 3) (5,16). A fistula was defined as a color Doppler tract communicating two adjacent cardiac chambers (5). Valvular regurgitation and its severity were graded by color Doppler imaging, using semiquantitative standard criteria (17,18).
The Student ttest or analysis of variance was used to compare continuous variables (expressed as the mean value ± SD), whereas categorical variables were compared using the chi-square test. Adjusted relative risks (RR) and their 95% confidence intervals (CI) were calculated by using the estimated regression coefficients and their standard errors in the logistic regression analysis. The existence of interactions was evaluated. Variables with a p value < 0.15 on univariate analysis were selected for multivariate analysis. The null hypothesis was rejected in each statistical test at p < 0.05. For data analysis, the statistical program SPSS, version 10.0 was used (SPSS Inc., Chicago, Illinois).
Perivalvular complications were detected in 78 patients (37%, group I). There were 50 abscesses, 34 pseudoaneurysms and 12 fistulae documented by direct macroscopic inspection, histologic examination or echocardiography in the patients who had no operation (n = 17). In the remaining 133 patients (group II), none of these lesions could be demonstrated. The incidence of periannular extension of infection in native and prosthetic valves was 29% and 55%, respectively.
The clinical characteristics of the study group are summarized in Table 1. The valve distribution frequency and microbiologic etiologies can be calculated from the incidence of several variables (Table 2). Age, gender and underlying systemic disorders (e.g., chronic renal failure, intravenous drug use, immunodeficiency, diabetes, neoplasia, chronic obstructive pulmonary disease, chronic anemia) were not significantly different between the two study groups. The presence of prosthesis (RR 1.88, 95% CI 1.35 to 2.64) and previous endocarditis (RR 1.78, 95 CI 1.16 to 2.7) were the only two preexisting heart conditions associated with perivalvular complications. The latency period between the onset of symptoms and the diagnosis of endocarditis was similar in both groups (Table 1). Anticoagulation and embolization during the total course of endocarditis or after initiation of an adequate antibiotic treatment were not different between the two groups (Table 1). The development of atrioventricular (AV) block during the course of endocarditis occurred in 15 patients from group I and in 3 patients from group II (13 patients had first-degree AV block and 5 patients had second- and third-degree AV block) (RR 2.55, 95% CI 1.91 to 3.41). The sensitivity, specificity and positive and negative predictive values of a new AV block for the detection of periannular complications were 18%, 97%, 83% and 67%, respectively. Twelve patients developed bundle branch block, but it appeared in both groups with a similar frequency (Table 1).
Aortic infection (RR 1.8, 95% CI 1.23 to 2.66) was associated with the existence of perivalvular complications (Table 2). The type of prosthesis (mechanical or biologic) and the type of prosthetic endocarditis (early or late) were not particularly associated with periannular complications (Table 2).
At follow-up, persistent signs of infection were not different between the two groups (Table 1). Severe valvular dysfunction was echocardiographically documented in 112 patients. A higher frequency of valvular insufficiency was observed in group I versus group II (60.3% vs. 48.9%) (Table 1). Nevertheless, this result did not reach a statistically significant difference.
Although it was not statistically significant, heart failure was more frequent in patients with periannular complications (Table 1), and most of them (76%) had heart failure at hospital admission. Fifteen patients had pericardial effusion; in all patients, the effusions were small, and their frequency was similar in both groups.
When the microorganisms of both groups were compared, no statistical differences were found (p = 0.29). Nevertheless, when coagulase-negative staphylococci were compared with the rest of the microorganisms, the former were more common in patients with perivalvular complications (RR 1.77, 95% CI 1.21 to 2.59) (Table 2). The proportion of patients with unknown microorganisms was similar in both groups.
When the incidence of periannular complications due to coagulase-negative staphylococci was stratified by the type of valve (native or prosthetic), no statistically significant differences were found.
On multivariate analysis, the variables that were independently related to the development of periannular complications were prosthesis, aortic position, staphylococcal coagulase-negative infection and new AV block (Fig. 4).
The sensitivity, specificity and positive and negative predictive power of TEE to detect periannular complications were 80%, 92%, 89% and 85%, respectively. Eight patients with aortic abscesses were overlooked by the TEE approach. There was no discernible relationship between the type of TEE probe used and the incidence of missed perivalvular infections. No pseudoaneurysms were missed, and there were few missed aortic abscesses (recognizable, in retrospect, as periannular thickening in 3 patients). The frequency and size of vegetations, as studied by TEE, were not significantly different between the two study groups (Table 2). The size of the pseudoaneurysm by TEE at diagnosis in patients who underwent surgery and in those who did not (17.9 ± 12 mm vs. 18.5 ± 14 mm; p = 0.91) and that of patients who died and in those who survived at follow-up (20.17 ± 13 mm vs. 16.9 ± 12 mm; p = 0.5 1) were similar. The abscess size by TEE at diagnosis in patients who underwent surgery and in those who did not (19.95 ± 29 vs. 7.4 ± 3.1 mm; p = 0.35) and that of patients who died and those who survived at follow-up (20 ± 30 vs. 16 ± 25 mm; p = 0.73) were not statistically different.
There were 39 patients with abscesses, 23 with pseudoaneurysms and 11 with both. Patients with abscesses required surgery more frequently than did those with pseudoaneurysms (36 [92.3%] vs. 13 [56.3%], p = 0.002; RR 1.63, 95% CI 1.13 to 2.64). The incidence of heart failure was higher in patients with abscesses, but it did not reach statistical significance (28 [71.8%] vs. 13 [56.5%], p = 0.27; RR 1.27, 95% CI 0.84 to 1.91). Persistent signs of infection (12 [30.8%] vs. 6 [26.1%], p = 0.78), AV block (6 [15.4%] vs. 3 [13%], p = 1) and death (15 [38.5%] vs. 10 [43.5%], p = 0.79) were similar in both groups.
Surgery was more frequently performed in patients with periannular complications (RR 3.2, 95% CI 2.03 to 5.14) (Table 1). As expected, the most common reasons for valve replacement in our patients were development of heart failure and persistent signs of infection; these were the primary indications for surgery in 51.3% of patients with periannular infection and in 27.1% of patients without these complications (p = 0.001). Twenty-seven patients (15 in group I and 12 in group II; p = 0.053) had surgery because of severe valvular dysfunction without heart failure. Only three patients had surgery because of the existence of perivalvular complications alone. A similar percentage of patients were alive at hospital discharge (64.1% and 72.9%, respectively) (Table 1). Twenty-two patients (36.1%) with periannular complications died during the operation or in the postoperative period, whereas 6 (35.3%) of 17 patients with these complications who did not have surgery died during follow-up. There were 24 deaths in total (6 due to septic shock, 3 due to disseminated intravascular coagulation, 3 due to hemodynamic shock and multi-organ failure, 3 due to neurologic complications, 1 due to aspiration and 6 due to arrhythmia and sudden death) in the patients from group II who had no operation.
Perivalvular extension of infection is a serious complication of bacterial endocarditis. The recognition of this complication has been greatly improved since the introduction of TEE. A number of studies have attempted to determine the risk factors and clinical course of patients with periannular extension of endocarditis; however, because of the heterogeneity of the patient populations reported, some of these aspects are unsettled.
In this prospective, large-scale study, we sought to circumvent many of the problems of previous studies. We focused on the risk factors, clinical course, microbiologic profile and echocardiographic findings in patients with these complications.
Our study supports the work of other investigators, clearly demonstrating the association between endocarditis-associated perivalvular complications and prosthetic valve endocarditis (4,5,19,20). Because prosthetic valve endocarditis usually begins as periannulitis, it is not surprising that infected prosthetic valves had these complications with a higher frequency than did infected native valves. Also, using multivariate analysis, involvement of the aortic valve was an independent variable that significantly predicted the presence of these complications (Fig. 4). Many studies support the association between aortic infection and periannular complications (1–3,5,10,12,13).
In the published data, one of the most convincing and consistently reliable variables predictive of periannular complications has been the appearance of an AV block (1,6,11,21). Our study definitely confirms this association (Fig. 4), which probably reflects the tendency of infection to extend into the membranous septal area, which contains the conducting tissue (22,23). Other variables classically associated with periannular complications, including pericarditis, a history of intravenous drug use and previous aortic stenosis, were not predictive of the presence of perivalvular complications. Historically, several authors have associated pericarditis (both purulent and nonpurulent) with periannular extension of infection (1,3,24); the latter may reach the epicardium through the cardiac wall. We and other investigators (11)could not corroborate this association, probably because nowadays the earlier diagnosis and appropriate treatment of these complications prevent the eventual development of pericarditis. In most patients, the presence of pericardial effusion will probably be reactive to infection or secondary to heart failure, uremia or pulmonary hypertension. A history of intravenous drug use was rare in our patient group, and we could not demonstrate an association of this condition with the development of perivalvular complications. Omari et al. (2)revealed the presence of active intravenous drug use as a significant risk factor for periannular extension of infection. Their patient group was markedly different from ours, as their study contained a large number (40 of 73) of intravenous drug users.
Periannular extension of infection has been associated with a delayed diagnosis and treatment (25). We could not corroborate this classic finding. In this study, the latency period between the onset of symptoms and the diagnosis and treatment of infection was similar in both groups. Most complications occur in the very active period of the disease.
The microbiologic profile of patients with periannular extension of infection has varied from study to study. In some studies, staphylococci were the predominant microorganisms (1,6,10), but not in others (2,5,11). In our study, coagulase-negative staphylococci were more frequently found in patients with periannular complications than in patients without extension of the infection around the annulus (Table 2). It would seem reasonable that virulent pathogens, such as staphylococci, can cause periannular destruction. Staphylococcal strains that cause endocarditis are resistant to platelet microbicidal proteins and elaborate proteolytic enzymes that facilitate spread to adjacent tissues (26).
Transesophageal echocardiography reliably determines the presence of periannular complications (10,27). This study was not designed to compare both echocardiographic modalities. The better performance of TEE over TTE in this setting was established by Daniel et al. (10). Since that study, many other authors have confirmed the outstanding accuracy of TEE in the diagnosis of periannular complications (5,7,16,27–31). We obtained similar results. The diagnostic accuracy of TEE for pseudoaneurysms was better than that for abscesses. In our study, the size of abscesses and pseudoaneurysms did not predict an increased need for surgical intervention and increased mortality overall. In addition, the presence and size of vegetations were not helpful in predicting perivalvular extension of infection (Table 2). This was also observed by Omari et al. (2)in a small group of patients. In fact, on univariate analysis, small vegetations (≤9 mm) were related to the existence of these complications; this could not be demonstrated on multivariate analysis, probably because of a small sample size.
Previous studies have suggested an association between perivalvular complications and an increased incidence of embolism (32). In this series, embolization was not more frequent in patients with periannular extension of infection, neither during the whole course of the disease nor after starting adequate antibiotic therapy (Table 1).
Classically, persistent signs of infection have been related to the existence of periannular complications (32,33). In this study, the frequency of these signs was not different between the two groups. Fever may persist for various reasons, and many patients with perivalvular complications will not have persistent bacteremia. In fact, in a recent retrospective, multicenter study of 233 French patients with endocarditis and abscesses, the infection appeared to be under control in 49% of those who had an operation (34). The incidence of persistent signs of infection in patients with abscesses was not higher than that of patients with pseudoaneurysms. Severe valvular dysfunction and heart failure were more frequent in patients from group I, but it was not statistically significant. The patients reported in this series were quite ill; in fact, 65 patients from group II had severe valvular insufficiency.
Our study confirms the results of other investigators (35,36), who clearly demonstrated an association between perivalvular extension of infection and the need for surgery. Most of these patients will undergo surgery because of classic surgical indications (i.e., heart failure and persistent signs of infection), independent of the echocardiographic detection of periannular complications. In our opinion, it is clear that most patients with periannular extension of infection will require surgery to eradicate the infection and to correct hemodynamic disorders. Nevertheless, it is also clear that a small number of these patients (8% in our series) may be treated successfully without surgical intervention. Such patients should probably be monitored closely by serial TEE. In this study, patients with abscesses needed surgery more frequently than did those with pseudoaneurysms, probably because the former had a higher incidence of heart failure.
Some studies suggest an association between periannular extension of infection and increased mortality (37,38). In our work, overall mortality was not statistically different between the two groups. A large proportion of patients (28%) from group II who did not undergo surgery died. We can speculate that had the diagnosis and treatment been earlier, we might have had a lower death rate in this group. Also, our surgical mortality rate in patients from group I was quite high (36%). However, it must be considered that surgery in patients with periannular extension of infection is presumed to be technically difficult. In any case, the aim of this study was not to compare surgical and medical treatments, but to describe the clinical course of patients with perivalvular infection.
Potential limitations of this investigation are the possible dissimilar patient characteristics and treatment modalities of different hospitals. In addition, comprehensive data on long-term outcomes are not provided.
Several interesting conclusions were derived from this large prospective study: 1) aortic infection, prosthetic endocarditis, new AV block and coagulase-negative staphylococci were independent risk factors of periannular complications (Fig. 4). The risk of developing one of these complications will depend on the presence of one or more of these risk factors in patients with endocarditis, and it varies between 19.1% (native mitral valve, non-staphylococcal infection and no AV block) and 93.8% (prosthetic aortic valve, coagulase-negative staphylococcal infection and new AV block). 2) The period from symptoms to diagnosis, incidence of pericardial effusion and persistent signs of infection were similar in both groups. 3) The presence and size of vegetations and the frequency of embolism were not higher in patients with periannular complications. Finally, 4) surgery was more frequent in these patients, but mortality was similar in both groups.
- confidence interval
- relative risk
- transesophageal echocardiography
- transthoracic echocardiography
- Received February 26, 2001.
- Revision received October 24, 2001.
- Accepted January 16, 2002.
- American College of Cardiology Foundation
- Arnett E.N,
- Roberts W.C
- Middlemost S,
- Wisenbaugh T,
- Meyerowitz C,
- et al.
- Ellis S.G,
- Goldstein J,
- Popp R.L
- Saner H.E,
- Asinger R.W,
- Homans D.C,
- Helseth H.K,
- Elsperger K.J
- Helmcke F,
- Nanda N.C,
- Hsiung M.C,
- et al.
- Perry G.J,
- Helmcke F,
- Nanda N.C,
- Byard C,
- Soto B
- John R.M,
- Pugsley W,
- Treasure T,
- Sturridge M.F,
- Swanton R.H
- Shively B.K,
- Gurule F.T,
- Roldan C.A,
- et al.
- Karalis D.G,
- Bansal R.C,
- Hauck A.J,
- et al.
- Afridi I,
- Apostolidou M.A,
- Saad R.M,
- Zoghbi W.A
- Bayer A.S,
- Bolger A.F,
- Taubert K.A,
- et al.
- Blumberg E.A,
- Robbins N,
- Adimora A,
- et al.
- Choussat R,
- Thomas D,
- Isnard R,
- et al.
- Rohmann S,
- Erbel R,
- Gorge G,
- et al.
- Sheldon W.H,
- Golden A