Enterococcal Endocarditis on Native and Prosthetic Valves

A Review of Clinical and Prognostic Factors With Emphasis on Hospital-Acquired Infections as a Major Determinant of Outcome

Abstract

Enterococci are the third leading cause of infectious endocarditis, and despite advances in diagnosis and treatment, the mortality of enterococcal endocarditis has not changed in recent decades. Although variables such as advanced age, cardiac failure, and brain emboli have been recognized as risk factors for mortality, cooperative multi-institutional studies have not assessed the role of other variables, such as nosocomial acquisition of infection, the presence of comorbidities, or the changing antimicrobial susceptibility of enterococci, as factors determining prognosis.

We conducted the current study to determine the risk factors for mortality in patients with enterococcal endocarditis in a single institution. We reviewed 47 consecutive episodes of enterococcal endocarditis in 44 patients diagnosed according to the modified Duke criteria from a retrospective cohort study of cases of infectious endocarditis. The main outcome measure was inhospital mortality. We applied stepwise logistic regression analysis to identify risk factors for mortality.

Predisposing heart conditions were observed in 86.3% of patients, and 17 had prosthetic valve endocarditis. A portal of entry was suspected or determined in 38.2%; the genitourinary tract was the most common source of the infection (29.7%). Comorbidities were present in 52.2% of cases. Twelve episodes (25.5%) were acquired during hospitalization. Only 3 isolates of Enterococcus faecalis were highly resistant to gentamicin. Eighteen patients (40.9%) needed valve replacement due to cardiac failure or relapse. Comparing cases of native valve and prosthetic valve endocarditis, we found no differences regarding complications, the need for surgical treatment, or mortality. Eight of 44 (18%) episodes were fatal. Age over 70 years (p = 0.05), heart failure (odds ratio [OR], 1.61; 95% confidence interval [CI], 1.15-2.25; p = 0.001), presence of 1 or more comorbidities (OR, 3.2; 95% CI, 1.11-9.39; p = 0.02), and nosocomial acquisition (OR, 8.05; 95% CI, 1.50-43.2; p = 0.01) were associated with mortality. In the multivariate analysis, only nosocomial acquisition increased the risk of mortality.

In patients with enterococcal endocarditis, nosocomial acquisition of infection is an important factor determining outcome. As the incidence of bacteremia and the population of elderly people at risk continue to grow, the hazard of acquiring nosocomial enterococcal endocarditis may increase; hence, major emphasis must be put on prevention.

Abbreviations: CT = computed tomography, HIV = human immunodeficiency virus, MIC = minimum inhibitory concentration, NVE = native valve endocarditis, PVE = prosthetic valve enterococcal endocarditis, TEE = transesophageal echocardiogram, TTE = transthoracic echocardiogram.

INTRODUCTION

Enterococci are the third leading cause of infectious endocarditis, accounting for 5%-11% of cases in most series^{11,45,83,86,100,111,119}. Because infectious endocarditis has tended to affect the elderly and in view of an increasingly aged population with endocarditis risk factors such as degenerative valve disease, it is plausible that the frequency of enterococcal endocarditis will increase in the years to come^83,122. Two recent reports^10,43, in which enterococcal endocarditis accounted for 17% of all cases of infectious endocarditis seen in different geographic areas in the last decade, suggest that enterococcal endocarditis may be of increasing importance. In these studies, enterococcus was second, after Staphylococcus aureus, in the rank order of major pathogens causing endocarditis^10,43.

Mortality of enterococcal endocarditis has remained unchanged for the last 3 decades, ranging from rates of 9%-47% in the 1970s and 1980s^41,70,77,125 to rates of 11%-35% in recent years^{1,48,73,86,93,117}. The reasons for the lack of improvement in prognosis are not known.

Recently 2 large series of patients with enterococcal endocarditis have shown older age, systemic embolization, and heart failure as independent predictors of mortality^73,86. However, these multi-institutional studies have potential limitations, such as heterogeneity in defining variables, in regional or national practice patterns, and in collection techniques. In addition, susceptibility data for antibiotics used to treat enterococcal endocarditis were not provided, and data on outcomes were available only up to the time of discharge⁷³. This limitation prevented analysis of relapses or of the need for surgery after discontinuation of treatment.

On the other hand, to our knowledge the role of nosocomial acquisition of enterococcal bacteremia on the prognosis of patients with endocarditis has not been assessed in detail. This analysis could be relevant because some recent reports have shown that enterococcal endocarditis is frequently acquired during hospitalization^3,20,22,73.

In the current study, conducted in a single institution, we investigated the epidemiologic, microbiologic, and clinical manifestations of enterococcal endocarditis to identify factors determining prognosis. We also present here a review of the most relevant studies on enterococcal endocarditis and the current problems of therapy for this infection.

PATIENTS AND METHODS

Patients with a diagnosis of enterococcal endocarditis seen from January 1988 through December 2005 in a 600-bed university-affiliated hospital were reviewed. Patients were under the care of members of the Division of Infectious Diseases or were seen by 1 or more of the authors by request of the primary service. Data were obtained from "ad hoc" sheets collected by Infectious Disease doctors and, starting in 1996, from a database.

Diagnosis of endocarditis was based on strict case definitions according to the modified Duke criteria for the diagnosis of infective endocarditis⁶². Transthoracic echocardiogram (TTE) was routinely performed when endocarditis was suspected. Beginning in 1993, transesophageal echocardiogram (TEE) was applied to cases with negative TTE and routinely in patients with prosthetic valve enterococcal endocarditis (PVE).

The portal of entry and the source of bacteremia leading to endocarditis was considered definitive when enterococci were isolated from the site before or simultaneous to diagnosis of endocarditis. Urine, intravenous catheters, intraabdominal pus and surgical wound infections were the sites more frequently cultured.

Duration of the infection was ascertained by the case history of the patient taking into account the duration of fever and other symptoms until diagnosis.

Nosocomial endocarditis was defined as endocarditis developing >72 hours after admission in association with a hospital-based procedure performed during the current hospitalization or during another hospitalization within the preceding 8 weeks²⁰. Endocarditis was considered associated with an intravenous catheter when signs of endocardial infection were documented in a patient with concurrent or recent (<8 wk) catheter-associated bacteremia. Endocarditis was considered associated with a genitourinary procedure when it developed during hospitalization or within 8 weeks of a preceding hospitalization and was temporarily linked to manipulation of the genitourinary tract (Foley catheter, curettage, transurethral resection of the prostate gland)²⁰.

Emergent endocarditis, that is, development of endocarditis after the onset of bacteremia, was defined by the absence of clinical evidence of endocarditis at the time the bacteremia was first documented¹⁹. This category was applied exclusively to patients with nosocomial endocarditis.

Methods for blood cultures changed over the years but at least 3 sets of cultures were taken from each patient with suspected endocarditis. Commercial bottles were always used (Hemoline, BioMérieux), and in the last 12 years the Bactec system was used. Identification of enterococci was performed following conventional methods¹⁸. Species-level identification was made using the API 20 strept identification system (BioMérieux Vitek, Inc).

Susceptibility testing of enterococci to antimicrobials was done following the National Committee for Clinical Laboratory Standards guidelines⁸⁴. Benzylpenicillin, ampicillin, and vancomycin susceptibility testing was performed by broth dilution methods or E-test. High-level resistance to aminoglycosides was detected by agar dilution methods and was defined as resistance to gentamicin or streptomycin ≥2000 μg/mL⁸⁴.

Pathologic studies used in the study consisted of the pathologic findings obtained at autopsy (4 cases) and observations at the time of surgical intervention.

Mortality per episode was calculated. Only the first episode of endocarditis was considered in the analysis of the risk factors for mortality. All deaths that occurred during the hospitalization as a consequence of infection, including operative mortality and any death that occurred within 30 days of discharge, were considered related to endocarditis.

Patients were followed in the outpatient clinic with regular visits at 1, 3, 6, and 12 months after discharge. Traditionally, investigators have presumed that an episode of endocarditis caused by the same species within 6 months after the initial episode represents a relapse. However, the agreement between time-based clinical criteria and molecular analysis of the isolates is imperfect¹⁴. For this reason, in the current study relapse was defined as recurrence of enterococcal bacteremia and symptoms of infection within 8 months after completion of treatment for the initial episode of endocarditis. A period of 8 months was chosen because this is the longer period found in the literature associated with relapse of enterococcal endocarditis after a 6-week course of penicillin and gentamicin⁹³.

Statistics

Categorical variables were reported as percentages and continuous variables as the mean ± SD. Differences between continuous variables were analyzed using the Student t test or nonparametric test in samples without normal distribution. Differences between noncontinuous variables were analyzed using the chi-square test or the Fisher exact test when required. Significance was set at p < 0.05. Stepwise logistic regression analysis was applied to variables that yielded significant results in the univariate analysis to identify risk factors for mortality.

RESULTS

Forty-four patients with 47 episodes of enterococcal endocarditis were reviewed. Three patients had 2 episodes with an interval longer than 1 year (15, 25, and 32 mo, respectively). In order to assess the epidemiologic data, clinical findings, and mortality, only the first episodes in 44 patients were analyzed. We cared for 1-3 cases of enterococcal endocarditis yearly and found no differences in the distribution of cases throughout the study period.

Patients were 25 men and 19 women with ages ranging from 26 to 93 years (median age, 56.8 yr). Diagnosis of endocarditis was definitive in all cases. In 22 we obtained evidence of endocarditis based on the presence of vegetations obtained during surgery (18 patients) or autopsy in the absence of prior surgery (4 cases); in addition, 32 had evidence of endocardial involvement by echocardiography (15 by TTE and 17 by TEE). The detection rate for vegetation of TTE and TEE in patients with PVE was 35.2% and 88.8%, respectively (6 of 17 vs. 8 of 9 cases, respectively; p < 0.05).

Thirty-eight (86.3%) patients had predisposing cardiac diseases: 17 (38.6%) had prosthetic valves, which were 13 metallic valves (Hall-Kaster, Björk, St Jude) and 4 bioprosthesis valves (Hancock); 10 (22.7%) had degenerative valvular disease; 4 (9%) had rheumatic valve disease; 3 (6.8%) had congenital cardiac disease (1 ductus arteriosus and 2 bicuspid aortic valves); 3 (6.8%) had previous endocarditis; and 1 had mitral valve prolapse. In 5 (11.3%) patients an underlying valve condition was not determined.

Twenty (45.4%) patients had extracardiac diseases: 6 (13.6%) had cirrhosis of the liver, 5 (11.3%) had gastrointestinal or genitourinary neoplasia, 4 (9%) had obstructive uropathy, and 3 (6.8%) were human immunodeficiency virus (HIV)-infected drug abusers. In addition, 1 patient was a renal transplant recipient and another 1 patient had an urethrocutaneous fistula.

Of the 3 HIV-infected patients, 1 was a man who was admitted with enterococcal tricuspid valve endocarditis. He was treated with penicillin plus gentamicin during 8 days but escaped from the hospital. He was readmitted 2 months later with relapsing enterococcal endocarditis involving the tricuspid and aortic valves. He developed severe renal and cardiac failure and died. The second patient was a woman who had had a previous episode of staphylococcal endocarditis. She developed mitral valve enterococcal endocarditis and cardiac failure. After a week on antimicrobial therapy with ampicillin plus gentamicin, valve replacement was successfully performed. The third patient developed aortic valve endocarditis that was treated with ampicillin plus gentamicin for 4 weeks. One month later endocarditis relapsed; echocardiographic signs of impending cardiac failure were determined, and valve replacement was promptly performed.

The valves involved were mitral valve, 19 (43.1%); aortic valve, 15 (34%); mitral and aortic valves, 7 (15.9%); tricuspid and aortic valves, 1 (2.2%); and ductus arteriosus, 1 (2.2%).

In 17 (38.6%) patients a portal of entry was suspected or definitively documented, including the genitourinary tract in 14 (31.8%) patients. One patient developed endocarditis after gastrointestinal surgery; 1 patient developed enterococcal bacteremia and endocarditis from a central venous catheter; and 1 developed endocarditis after liver biopsy.

Microbiology

Enterococcus faecalis (40 isolates) and E. durans (1 isolate) were the causative agents isolated from blood cultures and/or valvular tissue in 41 episodes of endocarditis. In 6 episodes, the isolates were identified merely as enterococci. Fourteen (35%) isolates of E. faecalis were highly resistant to streptomycin, but only 3 (6.8%) strains had a minimum inhibitory concentration (MIC) to gentamicin ≥2000 μg/mL. The MIC90 for gentamicin was 64 μg/mL, and MICs90 for benzylpenicillin, ampicillin, and vancomycin were 2 μg/mL, 1 μg/mL, and 2 μg/mL, respectively. Resistance to penicillins or vancomycin was not found.

Clinical Findings

Table 1 shows the most common symptoms and findings in patients with native valve endocarditis (NVE) and PVE. The mean duration of infection was 30 ± 18 days and 9 ± 6 days (p = 0.02) in patients with NVE and PVE, respectively. Only 2 patients had symptoms of enterococcal endocarditis for more than 3 months. These outliers were not taken into account when estimating the duration of infection.

TABLE 1: Clinical Findings of 44 Patients With Enterococcal Endocarditis

Most patients had a subacute course characterized by fever, malaise, generalized aches, and murmurs. Patients with PVE had a significantly shorter duration of symptoms than patients with NVE. Twelve (44.4%) patients with NVE and 9 (52.9%) patients with PVE developed cardiac failure. Brain emboli were observed in 5 (18.5%) and 4 (23.5%) patients, respectively.

Atypical presentations were occasionally seen. Enterococcal endocarditis followed a chronic course in an elderly man who had recurrent cerebral infarcts and developed what was considered multi-infarct dementia. He had a mild systolic murmur but fever was not observed. Finally he was brought to the hospital because of sudden worsening. On admission he was febrile; 3 blood cultures yielded E. faecalis, and TTE revealed mitral vegetations. Antimicrobial therapy was refused, and he died 7 days later.

Three patients, all with NVE, developed metastatic complications. One patient with aortic valve enterococcal endocarditis was brought to the emergency ward because of fever for 10 days and acute pain in the left upper quadrant of the abdomen that radiated to the shoulder and worsened during inspiration. A computed tomography (CT) scan showed a 53 × 60-mm hourglass splenic abscess that was drained percutaneously (Figure 1). Cultures yielded a monomicrobial growth of E. faecalis.

FIGURE 1: Splenic abscess complicating enterococcal endocarditis.

Another patient had a history of fever and pain in the back for 3 weeks. On admission, conjunctival petechiae and a systolic murmur were noted. Three blood cultures yielded E. faecalis. A CT scan revealed spondylodiscitis involving the bodies of L-1 and L-2 (Figure 2). Antimicrobial therapy was given for 5 weeks and he had an uneventful recovery.

FIGURE 2: Enterococcal endocarditis presenting with pyogenic spondylodiscitis.

A 68-year old man with degenerative mitral-aortic valve disease developed fever and generalized aches. A high-pitched systolic mitral murmur was noted and he was admitted for further evaluation. Blood cultures yielded enterococci, and he started treatment with ampicillin and gentamicin. Twenty-four hours later he suddenly noted a severe headache and fainted. A CT scan showed cerebral bleeding. Arteriography revealed a small mycotic aneurysm in a peripheral branch of the median cerebral artery. Fortunately bleeding subsided and he recovered after a 6-week course of antimicrobial therapy.

Hospital-Acquired Enterococcal Endocarditis

Twelve (27.2%) episodes of enterococcal endocarditis were hospital acquired. Table 2 shows the most relevant clinical findings in these patients.

Evidence for endocarditis was present at the time of the first positive blood culture or developed shortly after in most patients. Emergent endocarditis was diagnosed in 3 patients (Table 3). One was a 73-year-old man with colonic carcinoma who developed fever after a diagnostic colonoscopy. One of 2 blood cultures yielded enterococci and Proteus mirabilis. He was empirically treated with cefotaxime plus metronidazole for 5 days, and fever rapidly subsided. Ten days later, fever relapsed and new blood cultures yielded E. faecalis. TTE showed sessile vegetation on a calcified aortic valve. The second patient was a 64-year-old woman with endometrial carcinoma who developed fever 72 hours after radical hysterectomy. Two of 3 blood cultures yielded Escherichia coli, Bacteroides species, and enterococci; and E. coli and Bacteroides species, respectively. She was treated with clindamycin, gentamicin, and cefotaxime. Fever subsided and she was dismissed after 10 days on antimicrobial therapy, but was readmitted a week later because of relapsing fever. Three blood cultures yielded E. faecalis and a diastolic murmur was determined. TTE showed degenerative changes and vegetations in the aortic valve. Signs of severe aortic insufficiency and impending cardiac failure were observed. Ampicillin plus gentamicin was started but permission for valve replacement was denied, and she died soon.

TABLE 2: Characteristics and Outcome of 12 Patients With Nosocomial Endocarditis due to Enterococcus faecalis

TABLE 3: Clinical Findings and Outcome of 3 Patients With Hospital-Acquired, Emergent Enterococcal Endocarditis

The third patient was a woman admitted because of subarachnoid bleeding. She developed catheter-associated bacteremia with positive blood cultures for E. faecalis highly resistant to aminoglycosides. TTE was negative, treatment with ampicillin was given for 2 weeks, and bacteremia subsided. Six weeks later enterococcal bacteremia relapsed, and TTE showed mitral valve vegetations. She was treated with ampicillin 18 g/d for 6 weeks. Fever and bacteremia subsided. Because of the severity of the brain damage, she remained in hospital until her death 5 months later. Blood cultures taken periodically were negative.

Two more patients had endocarditis caused by E. faecalis highly resistant to gentamicin. Both developed the infection during admission for other reasons. One of these has been described in detail elsewhere²¹. The second was a man with acute myocardial infarction who developed acute mitral insufficiency. Valve replacement was performed but he followed a stormy course; he remained in the ICU and developed catheter-associated enterococcal bacteremia and endocarditis. The isolate was gentamicin-resistant E. faecalis but was susceptible to streptomycin. Despite antimicrobial therapy, a prosthetic leak was noted and he developed acute cardiac failure. Because permission for valve replacement was not obtained, he died due to acute pulmonary edema.

In 3 patients, enterococcal endocarditis followed endoscopic resection of the prostate gland. Only 1 of these patients had a previously known valvulopathy. He was a 62-year-old man with a prosthetic mitral valve who was given intravenous ciprofloxacin as endocarditis prophylaxis. Two weeks later he developed fever, chills, chest pain, and shortness of breath. Three blood cultures yielded E. faecalis, and a TEE showed prosthetic dehiscence and mitral vegetations. He was put on antimicrobial therapy with ampicillin plus gentamicin but developed cardiac failure and underwent successful valve replacement. The other patients had unrecognized degenerative valvular disease diagnosed by echocardiography during the evaluation for enterococcal bacteremia.

Five (41.6%) patients with hospital-acquired enterococcal endocarditis died. Acute cardiac failure was the main cause of mortality. Due to the severity of the underlying conditions, valve replacement was not undertaken in 2 patients, and both patients died. Valve replacement was performed in 3 patients, 2 of whom survived.

Pathology

Surgical pathology was studied in 6 patients with PVE (5 with metallic valves and 1 with a porcine bioprosthesis) and 12 with NVE. Differences between the groups were noted.

Patients with metallic prostheses had vegetation in the sewing ring (5 cases), dehiscence (3 cases), and abscess (1 case). Tears and vegetation were seen in the cusps of the single patient with a bioprosthesis. Vegetation and valvular regurgitation were observed in all patients with NVE. In addition, tears were observed in 5 cases. Myocardial abscesses were not observed in this group of patients.

Autopsy was performed in 4 cases. Valvular pathology consisted of vegetation in the aortic valve in 2 cases, in the mitral valve in 1 case, and in both valves in another case (Figure 3). Rupture of papillary muscles and perforation of aortic cusps were also observed. Emboli with infarcts in the spleen, kidneys, and myocardium were seen in 2, 2, and 1 cases, respectively. In addition, cirrhosis of the liver and bladder cancer were found in 2 of these patients. Chronic proliferative glomerulonephritis was observed in 1 patient who had a protracted course of at least 6 months. This patient developed multi-infarct dementia that finally determined his death. In the rest of the cases, acute cardiac failure seemed to be the primary cause of death.

FIGURE 3: Enterococcal endocarditis of the aortic valve. See vegetations and partial destruction of the left coronary cusp and small perforations in the noncoronary cusp.

Treatment and Follow-Up

Cell wall-active agents were given from 26 through 46 days (mean, 32 d) and gentamicin from 23 through 36 days (mean, 30 d). Dosages of gentamicin were designed to achieve trough levels ≤1 μg/mL. Most patient with normal renal function were treated with doses of gentamicin of 3 mg/kg per day.

Patients with PVE were treated longer than patients with NVE (38 d vs. 29 d, respectively). Reversible low-grade renal dysfunction was observed in 12 (27.2%) patients. In addition, 2 developed hearing loss. Two patients, a drug-abuser mentioned above and an 84-year-old man with chronic endocarditis and multiple cerebral emboli, did not receive effective antimicrobial therapy and died.

During the acute phase of the infections 18 (40.9%) patients needed valve replacement due to cardiac failure. Six (35.2%) patients with PVE and 12 (44.4%) with NVE needed surgical treatment to achieve cure (p = not significant [NS]).

Thirty-two patients were followed for a median of 10 months. Among patients who were given combined treatment with penicillins plus aminoglycosides, only 3 relapsed (6.8% of relapses after the first episode of endocarditis). Two patients had NVE and 1 had PVE. One of the patients with NVE was a 34-year-old man with a renal transplant who developed endocarditis caused by E. durans on a bicuspic aortic valve. The microorganism was susceptible to ampicillin (MIC 1 μg/mL) and gentamicin (MIC 16 μ/mL) and he was treated with ampicillin plus gentamicin for 4 weeks. Three weeks after discontinuation of treatment, he developed fever and relapsing enterococcal bacteremia. In vitro tests showed the same susceptibility pattern. A new course of ampicillin (12 g/24 h) plus gentamicin (1 mg/kg per 8 h) was extended for 6 weeks. Once again, after discontinuation of therapy, fever and bacteremia relapsed. He did not have emboli or cardiac failure but was considered for valve replacement, which was successfully performed.

We found no association between relapse and the duration of symptoms, involvement of a particular valve, an underlying disease, or PVE.

Six patients died within 1 year of diagnosis of enterococcal endocarditis. Cancer and cirrhosis of the liver were the primary causes of death in 5 cases. In only 1 patient was death associated with the underlying cardiovascular disease. Two patients, both with NVE, needed valve replacement 8 and 10 months, respectively, after the end of treatment because of progressive cardiac failure.

Mortality

Eight of the 44 patients with enterococcal endocarditis died as a consequence of infection (overall mortality, 18.1%). The mortality of patients with NVE and PVE was 22.2% and 11.7%, respectively (p = NS). The main cause of mortality was cardiac failure (6 cases) followed by multiple cerebral emboli and cerebral bleeding due to ruptured cerebral aneurysm.

In the univariate analysis, advanced age (p = 0.05), presence of 1 or more comorbidities (odds ratio [OR], 3.2; 95% confidence interval [CI], 1.11-9.39; p = 0.02), hospital-acquired endocarditis (OR, 8.05; 95% CI, 1.50-43.2; p = 0.01), and cardiac failure (OR, 1.61; 95% CI, 1.15-2.25; p = 0.001) were associated with mortality. Not significant were the site of involvement (aortic vs. mitral valve, or double vs. single valve), PVE, or valve replacement. Only nosocomial acquisition remained a risk factor for mortality in the multivariate analysis.

DISCUSSION

Enterococci account for a large proportion of cases of infective endocarditis seen in a university-affiliated hospital^81-83,86,93. In our institution, enterococcus was the third etiologic agent of endocarditis and accounted for 11% of cases. A 2007 report⁴³ suggests that the importance of enterococci may be increasing, and that these microorganisms have overcome viridans streptococci as a major microbial etiology of infectious endocarditis. However, it must be taken into account that there is a significant disparity between the microbial causes of endocarditis from different centers, and, in a single institution such as the Mayo Clinic in Rochester, Minnesota, among patients referred to the hospital and patients admitted from the local community of Olmsted County, Minnesota^101,111. In community-based practice, enterococcal endocarditis is still a relatively uncommon infection, accounting for only 5%-6% of cases^111,119.

Microbiology, Predisposing Factors, and Portal of Entry

E. faecalis is responsible for the vast majority of cases of enterococcal endocarditis^75,86,93,125. Only a minority of cases are caused by other species, such as E. durans, E. hirae, and E. avium^76,90. Despite the common occurrence of E. faecium bacteremia, endocarditis caused by this organism is rare^{36,66,80,88,106}. In 2 prospective series investigating the risk of developing endocarditis in patients with hospital-acquired enterococcal bacteremia, the isolation of E. faecalis but not E. faecium was significantly associated with endocarditis^3,22. The reasons for the different behavior of these organisms are unknown. Most non-E. faecalis enterococci lack for gelatinase, an aggregation substance that seemed to be of some importance in the pathogenesis of endocarditis^7,14. More recently, ACE-a microbial surface component recognizing adhesive matrix molecules structurally and functionally similar to Can on S. aureus-has been discovered to be present in E. faecalis⁹⁴. This collagen-binding surface component, which may be an important factor in allowing E. faecalis to adhere to heart valves, has not been detected on E. faecium strains, and thus may be part of the explanation of why the risks of endocarditis are different for the 2 organisms.

Enterococcal endocarditis is a disease that occurs in the setting of prior valvular damage or a prosthetic valve^{41,70,75,86,93,123}. The mitral and aortic valves are most frequently involved, with the tricuspid and pulmonic valves much less commonly affected^34,75. Right-sided endocarditis, typically seen in drug abusers, is rarely caused by enterococci. Enterococcal endocarditis in this population involved the mitral and aortic valves mainly⁹². Anecdotally, a case of quadruple-valve endocarditis due to E. faecalis has been reported⁵⁸.

Although in the past enterococcal endocarditis was frequently observed in women during childbearing years^{57,70,75,77,123}, currently most patients are elderly men with a variety of underlying diseases^1,2,73,86,93. The source of infection is most often the genitourinary tract^{1,70,73,86,125}. Urinary tract infection and diagnostic or therapeutic instrumentation of the genitourinary tract including cystoscopy, prostatectomy, cesarean section, and curettage are procedures associated with enterococcal endocarditis^{3,9,20,22,32,41,68,73,75,77,123,125}. Transrectal prostatic biopsy, transjugular intrahepatic portosystemic shunts (TIPS), and extracorporeal shock wave lithotripsy are new procedures that must be added to the list of causes of enterococcal bacteremia and endocarditis^{12,97,124,129}. In at least 1 study⁹³, the gastrointestinal tract was the most common portal of entry. Inflammatory lesions of the gut and biliary tree and malignant lesions of the gut are recognized sources of enterococcal bacteremia and endocarditis^93,125. In addition, colonoscopy and fiberoptic sigmoidoscopy are invasive procedures that have been associated with enterococcal endocarditis^33,85. Similarly, liver biopsy has been found to determine transient bacteremia due to enterococci and other potential endocardial pathogens in 15% of individuals with chronic liver diseases, which may cause endocarditis in patients with predisposing cardiac conditions^20,61.

Clinical Findings

Most patients with enterococcal endocarditis had a subacute course with fever and murmurs^75,123. Occasionally patients had atypical manifestations such as polyarthritis, spondylodiscitis, dementia, and metastatic abscesses^91,116. Splenic abscess with abdominal distention, hiccups, pain in the left flank, and pleural effusion has been reported⁵². Empyema secondary to splenic abscess has also been found in patients with enterococcal endocarditis¹¹². Table 4 shows the most common manifestations of enterococcal endocarditis in series reported from 1954 to 2005, including the current study.

TABLE 4: Clinical Manifestations of Enterococcal Endocarditis Comparing Pooled Series From 2 Periods, "Old Series" (1954-1974) and "Modern Series" (1984-2005), With Present Report*

It is noteworthy that the median age of patients in this series was 58 years, which is lower than that observed in other recent reports^2,73,86,93 but similar to that reported in older studies⁷⁷. The reasons for this disparity are not completely understood, but it is possible that the inclusion of drug addicts, some young patients with cirrhosis of the liver, and 1 woman aged in her 30s who developed enterococcal endocarditis after miscarriage have contributed to decreasing the age of patients in the series. On the other hand, the number of patients with PVE or degenerative valvulopathy was not significantly different from other recent studies^86,93. PVE was present in only 4% cases in the old series, but was present in 28% in modern series and in 38% in our experience^2,77,93,125.

Cardiac failure is the most common complication in patients with enterococcal endocarditis, and seemed to be more frequent now than in the past. Almost half of the patients in the current series developed cardiac failure, and 41% of patients required valve replacement. Valve replacement has been increasingly used in patients with enterococcal endocarditis (Table 5)^2,73,86,93.

TABLE 5: Treatment, Mortality, and Relapse Rate in Patients With Enterococcal Endocarditis, Present and Previous Reports

Hospital-Acquired Enterococcal Bacteremia and Endocarditis

It is believed that infectious endocarditis is a consequence of spontaneous or induced bacteremia from a colonized ecologic niche^75,81. Enterococcal bacteremia is commonly found in the setting of modern medical practice, particularly as a health care-associated infection occurring after invasive procedures of the genitourinary and gastrointestinal tracts^{3,22,35,36,68,69}. Enterococcus is the third most common cause of nosocomial infections in the United States, and the third leading Gram-positive microorganism causing bacteremia^71,126.

Enterococcal bacteremia is much more common than endocarditis. In a survey of studies published since 1980 that included 1036 patients with enterococcal bacteremia, we found that 60 (5.7%) were thought to have endocarditis (Table 6)^{22,29,35,36,68,69,88,101,120,129}. The risk of endocarditis is believed to be much higher in patients with community-acquired bacteremia than in those with nosocomial bacteremia, and the risk of developing endocarditis during an episode of hospital-acquired enterococcemia is estimated at 1%^68,82. However, other authors have found that endocarditis is a serious hazard in patients who develop enterococcal bacteremia during hospitalization^3,22.

TABLE 6: Frequency of Endocarditis in Different Series of Bacteremia Caused by Enterococcus faecalis

A previous report from our institution found that among 75 patients with hospital-acquired enterococcal bacteremia, 8% developed endocarditis²². The presence of valve disease, prosthetic valves, 3 or more positive blood cultures, and bacteremia caused by E. faecalis but not E. faecium were associated with endocarditis²². More recently, Anderson et al³ observed that 39% of their cases of enterococcal endocarditis diagnosed from 1992 to 1999 were hospital-acquired. By univariate and multivariate analysis, the presence of a prosthetic valve and infection with E. faecalis were significantly associated with endocarditis in patients with enterococcal bacteremia. The estimated risk of developing endocarditis in patients with E. faecalis bacteremia was 5.8%, a risk that increased to 64% in patients with prosthetic valves³. According to these data, the widely accepted concept that enterococcal endocarditis is rarely found in patients with health care-associated enterococcal bacteremia must be revised.

Although most patients in the current series had evidence of endocarditis at the time of the first positive blood culture or it developed shortly after, new or emergent endocarditis was observed in a subset of patients with nosocomial bacteremia. Some of these patients had polymicrobial bacteremia from an urogenital source, enterococci were isolated in association with other microorganisms considered true pathogens, or enterococci were isolated in a single bottle of blood culture, and so were not considered clinically significant. As a consequence, patients did not receive specific anti-enterococcal therapy until bacteremia relapsed and the diagnosis of endocarditis was entertained. Remarkably, both polymicrobial bacteremia and the positivity of only a small number of the blood cultures taken for diagnosis have been previously found in some patients with hospital-acquired enterococcal endocarditis³. Whether early treatment of enterococcal bacteremia could abort or eradicate endocarditis in these patients is unknown.

Although several early reports emphasized the role of some invasive procedures in the acquisition of infectious endocarditis during admission in the hospital^{38,98,102,118}, the interest in nosocomial infective endocarditis is relatively new. In most recent studies on hospital-acquired infective endocarditis, enterococci are the microorganisms most frequently isolated, besides staphylococci^{9,20,28,59,72,95,107}. To our knowledge, the impact these hospital-acquired infections may have on the prognosis of enterococcal endocarditis has not been analyzed before, and for this reason, 1 objective of the current study was to assess the role of nosocomial infections in determining the outcome of patients.

Risk factors for mortality that have been identified in previous studies on enterococcal endocarditis included age over 60 years, symptoms longer than 3 months, involvement of the aortic valve, congestive heart failure, and major embolic events^{41,73,75,86,93,125}. In agreement with some of these previous reports, we found that advanced age, the onset of cardiac failure, and the presence of comorbidities were associated with mortality in our patients. However, nosocomial acquisition of enterococcal endocarditis was the most important factor associated with mortality. Patients with hospital-acquired enterococcal endocarditis had a risk of mortality 8 times higher than patients who spontaneously developed the infection in the community. However, how the nosocomial acquisition of the infection could determine an increased risk of death is not entirely clear. We believe that other factors were serious obstacles to success, such as the advanced age of some of these patients, or the coexistence of severe comorbidities that may have limited the patient's ability to survive a severe infection such as endocarditis or that may have precluded valve replacement in patients with cardiac failure. Our findings may be due in part to some of these confounding interactions.

Susceptibility of Enterococci to Antimicrobials and Treatment

Treatment of enterococcal endocarditis has presented a dilemma since the early days of the antibiotic era, when, in contrast to the dramatic success of penicillin therapy for streptococcal endocarditis, unacceptable rates of failure were observed in patients with enterococcal endocarditis⁶⁵. The relative resistance of enterococci to the bactericidal activity of penicillin and other β-lactam antibiotics was later associated with a diminished affinity of the lower molecular weight penicillin-binding proteins (PBPs 4, 5, 6) of E. faecalis for these agents^25,79. This phenomenon called "penicillin tolerance," which can be identified by several in vitro characteristics, is a significant determinant of the in vivo response of enterococci to penicillin therapy⁵⁶. Researchers' demonstration that the combination of penicillin with streptomycin exhibited a synergistic bactericidal activity in vitro led to the empirical use of these antimicrobials, whose clinical benefits were verified by Geraci and Martin in 1954^32,49,50,96.

Initially, almost all strains of enterococci were killed by the synergistic action of penicillin and streptomycin⁴², but it was soon recognized that not all strains of enterococci were killed by this combination^39,51. The first observation of enterococcal endocarditis caused by a strain resistant to streptomycin and to the bactericidal activity of the combination was reported by Harvard et al in 1959³⁹. Guided by the simple philosophy "better deaf than dead," the patient was treated with penicillin plus neomycin, and although the use of the latter drug caused deafness in the patient, he survived the infection³⁹. However, the therapeutic dilemma represented by enterococcal endocarditis caused by strains with high-level aminoglycoside resistance was not widely recognized until the early 1970s, when a significant number of enterococci with high-level resistance to streptomycin and kanamycin were isolated in different parts of the United States^78,104. Fortunately, researchers found that penicillin-gentamicin exhibited synergistic activity against these isolates and was successful therapy for enterococcal endocarditis¹²⁵.

To our knowledge, the first isolation of enterococci with high-level resistance to gentamicin was in France in 1979⁴⁷. In the past 2 decades, organisms with high-level resistance to gentamicin have rapidly disseminated throughout the world, and in some countries over 50% of enterococci are resistant to gentamicin and to penicillin-gentamicin synergy^{40,74,82,127,128}. Most of these isolates are also resistant to the synergy of all penicillin-aminoglycoside combinations, although, paradoxically, 30% are still susceptible to penicillin-streptomycin synergy¹⁶.

High-level gentamicin resistance is due to the presence of a bifunctional enzyme with both 6′-acetylating and 2″-phosphorylating activities²⁴. The latter activity is responsible for resistance to gentamicin specifically, but the bifunctional enzyme also confers resistance to all available aminoglycosides. However, streptomycin is not affected by this enzyme. Streptomycin resistance, which is mediated by a separate genetic element that codifies an adenylyltransferase, is very common among enterococci, and so a large proportion of highly gentamicin-resistant strains will also prove highly resistant to streptomycin¹⁶.

Despite the worldwide dissemination of aminoglycoside-resistant enterococci and the common occurrence of enterococcal bacteremia in modern medical practice^71,126, the number of cases of endocarditis caused by these strains is small^{1,4,46,53,55,60,63,93,103,113}, and the most recent series of enterococcal endocarditis did not show an increasing number of cases^2,73,86. Recently, Gavaldá et al³⁰ reported a series of 21 cases of enterococcal endocarditis caused by strains highly resistant to gentamicin. This series was collected over 9 years in 13 teaching hospitals in Spain. Most patients were elderly men with aortic endocarditis, and only 8 (38%) patients had health care- associated enterococcal endocarditis²⁹. The scarcity of cases of enterococcal endocarditis caused by aminoglycoside-resistant enterococci is difficult to explain in view of the large dissemination of these strains both in the hospital environment and the community^{40,82,89,120,127,128}.

Table 7 provides a summary of cases of endocarditis caused by E. faecalis highly resistant to gentamicin reported to date in the literature, including the current study. Only cases with a definitive diagnosis of endocarditis and complete clinical information are included. In the current study, the first case was seen in 1988, the second in 1996, and the third and last in 2002. Most of the patients with endocarditis caused by E. faecalis highly resistant to gentamicin were men who acquired the infection during hospitalization for a variety of different conditions. It is noteworthy that many patients remained febrile while being treated with the combination of ampicillin with gentamicin, and at least 2 of them showed "breakthrough" bacteremia. Cardiac failure was common, and valve replacement was performed in 60% of cases. Only 2 patients died, 1 after cardiac surgery and another, for whom valve replacement was refused, due to cardiac failure. From this review we can conclude that infection caused by E. faecalis highly resistant to aminoglycosides, although frequently not responsive to treatment, does not increase the mortality of patients with enterococcal endocarditis.

TABLE 7: Cases of Endocarditis Caused by Enterococcus faecalis With High Resistance to Gentamicin, Present and Previous Reports

Optimal treatment of endocarditis caused by gentamicin-resistant enterococci is an unresolved question. For the strains susceptible to streptomycin, the combination of penicillin, ampicillin, or vancomycin with streptomycin is the best option⁶. For the more common strains resistant to all the available aminoglycosides, a double combination of β-lactam antibiotics has been recommended based on in vitro studies and the results of the experimental model of infectious endocarditis^4,30,67. A combination of ampicillin with ciprofloxacin has been successfully used in a few patients^46,60,113.

Recently, the results of treatment of 21 patients with enterococcal endocarditis caused by strains highly-resistant to gentamicin with a combination of ampicilin and ceftriaxone have been published³⁰ Patients were collected over a period of 9 years in 13 university-affiliated hospitals in Spain. Of 21 patients included in that prospective study, 7 died during hospitalization and 3 were treated with the combination plus valve replacement. Ten patients treated with antimicrobial alone for 6 weeks were considered cured without relapses after 3 months of follow-up³⁰.

These encouraging observations must be taken cautiously. First, the mortality observed in this group of patients with enterococcal endocarditis was unusually high, which reveals the severity of this infection, and failures have been found by others ¹¹⁵. Second, the follow-up was probably not enough to affirm the cure. Although most relapses of enterococcal endocarditis occur within 8 weeks after discontinuation of therapy¹²⁵, delayed relapses are also found^13,93. In addition, it must be taken into account that ampicillin alone given at high doses during prolonged periods could achieve cure for a significant number of patients. While it is clear that treatment with ampicillin alone is inadequate in many patients, this treatment may cure some patients with enterococcal endocarditis.

The model of infectious endocarditis in animals has provided a valuable means for further understanding the treatment of enterococcal endocarditis^17,37. Under appropriate experimental conditions, enterococcal endocarditis may be successfully treated with ampicillin alone. Thadepalli et al¹⁰⁸ cured 17 of 17 rabbits with 5 weeks of high-dose procaine-penicillin. Remarkably, infection was invariably fatal within 2 weeks in untreated controls¹⁰⁸. Tight¹¹⁰ administered ampicillin for 21 days and sterilized vegetation in all of 14 rabbits with enterococcal endocarditis. Others have found that continuous infusion of ampicillin for 5 days was more effective than intramuscular injection in reducing bacterial counts in cardiac vegetation and eradicated the organisms in 38% of the animals treated¹⁰⁹, which is almost never observed after treatment with combinations of antimicrobials. Although the use of intolerant strains in some of these experiments may have influenced the results, the results suggest that high-dose, prolonged therapy with ampicillin may cure a significant proportion of cases of enterococcal endocarditis.

On the other hand, the clinical experience provides some interesting observations. In the early days of antimicrobial therapy, Geraci and Martin³² reported cure of 7 of 18 (39%) patients with penicillin alone. Other authors have reported on the cure of patients with endocarditis caused by aminoglycoside-resistant enterococci with ampicillin alone^{8,53,63,64,87}. The same criticism can be formulated when assessing other combinations that include ampicillin with drugs that inhibit cell-wall synthesis⁴ or ciprofloxacin^46,113. Although ciprofloxacin and some other fluoroquinolones exhibit significant activity in the experimental model of enterococcal endocarditis^21,93, the combination with ampicillin used to be "indifferent" by in vitro tests of drug interactions and in the experimental model of endocarditis^23,46, so it is doubtful that combining these drugs with ampicillin has increased the antimicrobial activity of the β-lactam.

Valve replacement may be of paramount importance to cure this infection. The review of cases of endocarditis caused by E. faecalis highly resistant to gentamicin shows that half of the patients needed valve replacement to achieve cure (see Table 6). It is worth noting that treatment failure, despite the high serum bactericidal activity on ampicillin therapy, and subsequent cure with valve replacement has been observed^21,93. However, surgery is not indicated in the absence of cardiac failure or major emboli, and the decision to undertake valve replacement merely because of the isolation of aminoglycoside-resistant enterococci is not supported by the available information. Until new drugs with improved activity against enterococci are developed, treatment of endocarditis caused by highly aminoglycoside-resistant strains will continue to be a largely empirical endeavor. In these patients, ampicillin plus ceftriaxone or high-dose ampicillin monotherapy for 6-8 weeks seems to be reasonable treatments based on in vitro and in vivo observations and a limited clinical experience. Patients who remain febrile, develop complications, or relapse after prolonged therapy should be considered candidates for surgery.

CONCLUSION

The ubiquity of enterococci in the human gut, their ability to colonize other ecologic niches and medical paraphernalia, their resistance to the bactericidal activity of β-lactam antibiotics, and their capacity to develop resistance to any class of antimicrobial agent introduced into practice^{5,27,44,54,106,130}, mean that these organisms will continue to be phenomenal pathogens and a major challenge to our ability to develop effective antimicrobial agents. In this scenario major emphasis must be put on prophylaxis.

The current experience and previous reports suggest that the increasing rate of hospital-acquired enterococcal bacteremia on a substrate of people with predisposing heart diseases would certainly account for changes in the spectrum of enterococcal endocarditis compared to studies performed 30-40 years ago. As in the case of endocarditis caused by S. aureus²⁶, changes in health care delivery and the aging of the population may alter the epidemiology and prognosis of enterococcal endocarditis in the current era.

As the incidence of bacteremia and the population of elderly people at risk continue to grow⁷¹, the hazard of acquiring nosocomial enterococcal endocarditis may increase; hence, a great deal of emphasis must be put on prevention. From our review of cases in the current and previous series, we find that many cases may have been prevented^3,19,20,59. Prevention should largely be aimed at first identifying patients at risk, particularly the elderly with undetected degenerative valvular disease. Individuals with conditions whose diagnosis or treatment requires instrumentation or surgery of the genitourinary or gastrointestinal tracts should be identified at the time of hospital admission. Until new recommendations are issued for antimicrobial prophylaxis specifically designed to prevent nosocomial endocarditis, antibiotics should be used according to the most recent recommendations¹⁵.