Background. Endocarditis due to vancomycin-resistant enterococci (VRE) is rare, and the literature consists almost exclusively of reports of single cases.
Methods. We report a case of VRE prosthetic valve endocarditis and review 18 cases of native and prosthetic valve VRE endocarditis reported in the literature.
Results. The majority of cases were due to Enterococcus faecium. Nearly all of these infections were hospital acquired, and the vast majority of patients had significant underlying disease processes, including dialysis and transplantation. More than three-quarters of cases were left-sided, and the aortic valve was most commonly involved. Peripheral stigmata of endocarditis were not reported in any of the cases. Approximately 40% of patients developed cardiac complications. Nearly three-quarters of patients survived, despite the difficulties associated with providing bactericidal antimicrobial therapy, and only 4 patients underwent valve replacement.
Conclusions. VRE endocarditis is an uncommon nosocomial infection that affects patients with significant comorbid conditions. Most cases are due to E. faecium, and the aortic valve is involved in at least one-half of cases. One-third of patients require surgical treatment. Optimal antimicrobial therapy remains undefined, but an attempt to identify bactericidal combination therapy should be sought.
Enterococci have become common nosocomial pathogens and are currently the third-leading cause of nosocomial bloodstream infection, accounting for 9% of such infections [1]. Over the past 3 decades, enterococci have become increasingly resistant to antimicrobials. Vancomycin resistance in enterococci emerged in the late 1980s and is now found in 60% of Enterococcus faecium isolates causing nosocomial bloodstream infections [1]. We report a case of prosthetic valve endocarditis due to vancomycin-resistant enterococci (VRE) and review the cases of native and prosthetic valve VRE endocarditis reported in the literature.
A 25-year-old woman with systemic lupus erythematosus and end-stage renal disease underwent mitral valve replacement for severe regurgitation, resulting in hemolytic anemia. Her postoperative course was complicated, requiring 3 returns to the operating room for hemorrhage and cardiac arrest. On day 12 after operation, 1 of 2 blood cultures yielded E. faecium that was resistant to vancomycin (MIC, ⩾32 mg/L) and ampicillin (MIC, ⩾16 mg/L) and susceptible to quinupristin-dalfopristin (MIC, <0.5 mg/L), linezolid (MIC, <2 mg/L), daptomycin (MIC, 4 mg/L), rifampin (MIC, 0.016 mg/L) and high-level gentamicin (MIC, <500 mg/L; actual MIC, 4 mg/L) and coagulase-negative staphylococci, which were considered to be contaminants. On the day of hospital discharge (day 24 after the operation), another blood sample was obtained for culture. When this culture yielded vancomycin-resistant E. faecium (with same susceptibility profile as the previous isolate), linezolid therapy was prescribed. However, the patient never obtained the medication. Notably, she had a femoral dialysis catheter in place that was not removed because of lack of vascular access.
Six weeks after the operation, the patient developed chills, nausea, and vomiting, and blood cultures again yielded VRE. Physical examination revealed a new systolic murmur, and transesophageal echocardiography revealed a vegetation (area, 1.2 × 0.8 cm) on her prosthetic mitral valve. There were no peripheral stigmata of endocarditis. She was deemed not to be a candidate for surgery. She was initially treated with linezolid (600 mg po q12h for 10 days), which was changed to daptomycin (8 mg/kg iv q.d.), gentamicin (90 mg iv after each dialysis session), and rifampin (300 mg po q8h).
Time-kill curves were performed to assess the bactericidal activity of combinations of daptomycin, gentamicin, and rifampin. Each agent was used at 0.5 times the MIC. Synergistic activity was defined as a ⩾2-log10 decrease in the number of colony-forming units (cfu) per milliliter between the combination and its most active component after 24 h of incubation. Bactericidal activity was defined as a ⩾3-log10 decrease in the number of cfu per milliliter in the inoculum after 24 h of incubation. The combination of daptomycin and rifampin resulted in a 2.35-log10 decrease in the number of cfu per milliliter after 24 h of incubation, compared with rifampin (synergistic), and a 2.99-log10 decrease in the number of cfu per milliliter in the inoculum after 24 h (nearly bactericidal) (figure 1). The combination of daptomycin, gentamicin, and rifampin resulted in a 3.90-log10 decrease in the number of cfu per milliliter, compared with rifampin (synergistic), and a 4.53-log10 decrease in the number of cfu per milliliter in the inoculum (bactericidal).
Time-kill curve for combinations of daptomycin (Dap), gentamicin (Gent), and rifampin (Rif), with each drug at 0.5 times the MIC.
The patient required transfer to the intensive care unit (ICU) twice for management of heart failure. After 2 weeks of treatment, she decided to forgo further therapy. Antimicrobial therapy was discontinued, and the patient was transferred to the palliative care unit. However, 2 weeks later, she significantly improved and decided to resume antimicrobial therapy. She ultimately received 11 weeks of treatment with daptomycin, gentamicin, and rifampin. There was no relapse, as demonstrated by blood culture, but 4 months after the completion of antimicrobial therapy, she presented to an outside hospital with abdominal pain and hypotension, and she died within 72 h. The exact cause of her death and whether it was related to endocarditis are unknown.
A literature review of VRE endocarditis was performed using the Medline/PubMed database. Multiple searches were performed, all of which were limited to English-language articles published since 1988, the year in which vancomycin resistance in an enterococcal isolate was first reported. The first search used the search term "endocarditis" and yielded 1367 articles. A second search, performed with the search term "vancomycin resistant enterococcus endocarditis," yielded 77 articles. A third search was performed using the search terms "glycopeptide resistant enterococcus endocarditis"; this search yielded 49 articles. The abstracts from the articles retrieved from all 3 searches were reviewed, and the references from pertinent articles were reviewed to identify additional reports. Articles that presented aggregate patient data (e.g., clinical trials in which data on individual patients were not reported) were excluded.
A total of 22 relevant publications were identified. Of these studies, specific case information was available for 17 cases of native valve VRE endocarditis and for 7 cases of prosthetic valve VRE endocarditis (including the present case). These cases were then evaluated with use of the modified Duke criteria [2]. Cases that met the criteria for definite and possible endocarditis were included. Additionally, whenever the authors of an article claimed that their case(s) met the Duke criteria without delineation of the specific components, the cases were included in the study. Vancomycin resistance was defined as a vancomycin MIC of ⩾32 mg/L. If the authors claimed that the isolate was resistant to vancomycin without reporting the MIC, the case was included; however, if the reported MIC was <32 mg/L, the case was excluded.
Our analysis yielded 9 definite cases of native valve VRE endocarditis, 1 possible versus definite case, and 2 possible cases. Additionally, 6 cases of definite prosthetic valve VRE endocarditis were identified (including the case that we present), as well as 1 possible case. One case categorized as prosthetic valve endocarditis for purposes of analysis had concomitant prosthetic and native valve involvement [3].
Five cases were excluded from the analysis. In 1 case, the Duke criteria for possible or definite endocarditis were not met [4]; in 2 cases, the infecting organism had a vancomycin MIC of <32 mg/L [5, 6]; and in 2 cases, the patients had polymicrobial endocarditis [7, 8].
Of the 19 cases of VRE endocarditis identified, 14 (74%) were reported from the United States, 3 (16%) were reported from Europe (2 from Italy and 1 from the United Kingdom), and 1 (5%) each was reported from South Africa and Canada. Eighteen cases appear to have been hospital acquired. In 3 cases, patients had VRE colonization but not infection before the development of endocarditis. Six patients had active VRE infection before they developed endocarditis; of these patients, 5 had intravascular infection (2 hemodialysis-access infections, 2 central venous catheter–related infections, and 1 case of septic thrombophlebitis); the other patient had an infected hematoma and VRE peritonitis. Classic clinical findings of vascular and immunologic phenomena associated with endocarditis (e.g., arterial emboli, conjunctival hemorrhages, Janeway lesions, Osler nodes, and Roth spots) were not reported for any of the cases.
Native valve VRE endocarditis. In the native valve VRE endocarditis group (12 patients) (table 1), there were 8 male patients, 2 female patients, and 2 patients for whom the sex and age were not reported. All cases occurred in adults (age, 34–76 years), with the exception of 1 case that occurred in an infant. The mean age for development of VRE endocarditis was 54 years (median, 59.5 years).
Characteristics of patients with native valve endocarditis due to vancomycin-resistant enterococci (VRE).
All patients had significant comorbidities. Three (25%) had diabetes mellitus. Five (42%) had posttransplantation status (4 solid organ transplant recipients and 1 stem cell transplant recipient). Three patients (25%) were hemodialysis dependent. Five (42%) had undergone recent surgery, and 3 (25%) had various malignancies.
Endocarditis involved a single valve in 9 cases, and in 1 case, there was definite involvement of 1 valve and possible involvement of a second valve. Infection was left-sided in 70% of cases, right-sided in 20%, and bilateral in 10%. The aortic valve was the most commonly infected valve (in 5 cases, plus an additional case that involved the tricuspid valve in which there was questionable involvement of the aortic valve). The mitral valve was infected in 2 cases, the tricuspid valve was involved in 3 cases, and the involved valve was not reported for 2 cases. No patients had pulmonic valve involvement.
In 8 cases (67%), the enterococcal isolate was E. faecium. The isolate was Enterococcus faecalis in 2 cases, and in 2 cases, the species was not reported. Antimicrobial susceptibilities are shown in table 2.
Only 3 major complications were noted among the 12 cases. One patient had aortic valve perforations, a second patient had a valve ring/aortic root abscess, and 1 patient had severe aortic regurgitation requiring valve replacement.
Of the 12 patients, only 3 underwent surgery, all of whom survived. Three (25%) of the 12 patients died, 2 of whom had deaths that were attributed to endocarditis. Of the 4 patients treated with linezolid (all of whose isolates were susceptible to linezolid), 1 died (however, his death was not attributed to endocarditis). Of the 5 patients who were treated with quinupristin-dalfopristin, 1 died (drug susceptibility data were not provided for this case).
Prosthetic valve VRE infective endocarditis. In the prosthetic valve VRE endocarditis group (7 patients) (table 3), there were 4 men and 3 women. The mean age for development of VRE endocarditis in this group was 56 years (median, 64 years).
Characteristics of patients with prosthetic valve endocarditis due to vancomycin-resistant enterococci (VRE).
Three of these patients had prosthetic aortic valves; 2 of these valves were metallic, and the other valve was not described. Three additional patients had prosthetic mitral valves, 2 of which were bioprosthetic, and the valve type was not provided for 1 of the valves. Neither the type nor location of the prosthetic valve was specified for 1 case. The time from valve replacement to the development of endocarditis ranged from 3 weeks to 2 years, with 4 of the 6 cases for which the temporal relationship was reported occurring within the first 2 months. Additionally, the majority of these patients also had comorbidities (2 had diabetes, and 3 were dialysis dependent [2 were undergoing hemodialysis, and 1 was undergoing peritoneal dialysis]). Endocarditis involved the aortic valve in 4 patients (prosthetic valves for 3 and native valve for 1) and the mitral valve in 3 patients. The involved valve was not indicated for 1 case.
The infecting species was E. faecium in 3 cases, E. faecalis in 3 cases, and in 1 case, the species was not reported. Antimicrobial susceptibilities are shown in table 2.
Complications were noted in 4 of 7 patients. Two patients had mechanical valve failure, 1 had a perforated valve leaflet, and an additional patient had severe functional mitral stenosis and a periaortic abscess. Of the 4 patients for whom complications were noted, 2 underwent surgery (both survived), and 2 were treated medically (1 survived).
Of the 7 patients, 5 (71%) survived. Only 2 patients underwent surgery, both of whom survived. Five patients were treated with medical therapy alone, 3 of whom survived.
The enterococci are relatively nonvirulent gram-positive bacteria that are part of the normal gastrointestinal flora. They have become increasingly common nosocomial pathogens because they are resistant to many antimicrobials, can survive in the environment for prolonged periods of time, and can easily contaminate the hands of health care workers who, when noncompliant with hand hygiene practices, carry the organism directly or indirectly from patient to patient.
Over the past 35 years, enterococci have increasingly acquired resistance to antimicrobials, and the first case of vancomycin resistance was reported in 1988 [25]. By the end of 2003, a total of 28.5% of enterococcal isolates that caused infections in ICU patients were resistant to vancomycin [26]. The first case of VRE endocarditis that met the Duke criteria was reported in 1996 [8].
In patients aged <60 years, 5%–8% of cases of native valve endocarditis are due to enterococci [27]. For patients aged >60 years, enterococci account for 14%–17% of cases. For both age groups, enterococci are surpassed in frequency only by Streptococcus species and Staphylococcus aureus. In patients with prosthetic valve endocarditis, enterococci account for 5%–10% of cases that occur in the first 60 days after valve replacement and in 8%–15% of cases >60 days after this procedure [27]. Two-thirds of the prosthetic valve cases we reviewed were early cases of endocarditis.
Two patient populations (patients undergoing dialysis and transplant recipients) appear to be at higher risk for VRE endocarditis (table 4). In our analysis, 26% of patients were transplant recipients (primarily renal transplant recipients), and 32% were patients who were undergoing dialysis. Rates of intestinal VRE colonization have been shown to be higher in these groups [28–34].
The rates of VRE colonization among 5 cohorts of patients undergoing dialysis ranged from 6% to 14% [28–32], with 4 of these studies revealing a colonization rate of >10% [29–32]. When patients undergoing dialysis develop nosocomial enterococcal bloodstream infection, 75% of the cases are due to vancomycin-resistant isolates [1]. Of the 6 patients undergoing dialysis who had VRE endocarditis identified in this study, all but 1 survived.
In 3 cohorts of transplant recipients, 39 (16%) of 240 liver transplant recipients [33], 13 (6%) of 203 kidney and kidney/pancreas transplant recipients at one health care center [33], and 42 (18%) of 234 liver transplant recipients at a different medical center were infected or colonized with VRE [34]. In a 1998 review of the literature on endocarditis in solid organ transplant recipients, Patterson et al. [10] demonstrated that enterococci accounted for 5 (11%) of 46 cases, surpassed in frequency only by S. aureus (30% of cases) and Aspergillus fumigatus (17%). Presentation of enterococcal endocarditis in these patients occurred a median of 3 months after transplantation, and the survival rate was 80%. In the present study of cases of VRE endocarditis, 4 transplant recipients survived, and only 1 underwent valve replacement. One patient who had undergone autologous peripheral blood stem cell transplantation died.
In a comparison of the patients in this study with those in 3 contemporaneous studies of endocarditis (as defined by the Duke criteria) due to vancomycin-susceptible enterococci (VSE) [35–37], the sex distribution is similar (47%–72% of subjects with VSE endocarditis were male, compared with 71% for VRE endocarditis). Age distributions were also similar: in the 3 VSE endocarditis cohorts, the mean or median ages ranged from the late 50s to early 70s, whereas the median age for patients with VRE endocarditis was 60 years. One notable difference is that E. faecalis predominated in the VSE endocarditis groups (accounting for 84%–94% of cases), whereas 69% of the speciated cases of VRE endocarditis were due to E. faecium. This is not surprising, because the prevalence of vancomycin resistance is much greater for E. faecium isolates. In a large study of nosocomial bloodstream infections that included nearly 2000 cases due to enterococci, vancomycin resistance was 30-fold more common for cases due to E. faecium than for those due to E. faecalis (60% vs. 2%) [1].
Similar to other series of VSE endocarditis [37, 38], in our study of VRE endocarditis, the aortic valve was the most commonly infected valve. The lack of endocarditis peripheral stigmata in the VRE cases is consistent with findings reported previously for VSE endocarditis [39].
Interestingly, despite fewer therapeutic bactericidal regimens, the survival rate among patients with VRE endocarditis (74%) is similar to that reported in the contemporaneous VSE endocarditis studies (80%–84%) [35, 37]. When ampicillin-susceptible cases of VRE endocarditis are excluded, the survival rate is 64%. Although some have argued that VRE endocarditis is an indication for valve replacement on the basis of lack of effective antimicrobial therapy [40], it is important to note that 5 of the 7 patients with VRE prosthetic valve endocarditis described in the literature did not undergo valve replacement; of these patients, 3 were cured with medical management only. Of the 5 patients who underwent valve replacement, all survived.
Treatment of enterococcal endocarditis with a bactericidal synergistic combination of antimicrobials should be used when possible. All 3 cases in our review that were caused by strains that were ampicillin susceptible and had high-level gentamicin susceptibility were cured with bactericidal synergistic therapy. It should be noted that high-level resistance to streptomycin is mediated by a different gene than that which determines high-level gentamicin resistance. Therefore, susceptibility testing for both aminoglycosides should be performed.
For those cases caused by ampicillin-resistant strains, identification of a bactericidal regimen using currently approved drugs is essentially impossible, and optimal therapy has not been established. On the basis of our review of published cases, it is not possible to make recommendations regarding therapy, because the number of cases is small, data are missing in some case reports, multiple different regimens were used, and many patients were treated with >1 regimen. Our observations may also be affected by publication bias, because it is possible that successfully treated cases were more likely to be reported and published than were cases for which therapy was unsuccessful. Most of the isolates were susceptible to quinupristin-dalfopristin and linezolid, and several of the patients were treated with these drugs. Even though both drugs are bacteriostatic, some of the patients were cured.
Unfortunately, at the present time, there are no data available from randomized, controlled trials to determine optimal therapy. However, in addition to the case reports analyzed in this review, there are 2 studies that summarized the compassionate use programs for quinupristin-dalfopristin and linezolid that contain aggregate data on the use of these drugs for VRE endocarditis [41, 42]. In an intention-to-treat analysis of 10 patients who received quinupristin-dalfopristin for VRE endocarditis, only 2 survived [41]. In an intention-to-treat analysis of 22 patients with VRE endocarditis treated with linezolid, cure was achieved in 10 (45%) of the patients [42]. In addition, of the 3 patients in our review who received at least 6 weeks of linezolid, all were cured. We are unable to determine to what degree the cases reported individually in the literature overlap with the cases included in the compassionate use studies.
Because our patient was not a candidate for an operation, our goal was to identify a regimen that we believed would offer maximal bactericidal activity. We chose to treat her with daptomycin, gentamicin, and rifampin. It is important to note that daptomycin, a cyclic lipopeptide, has not been approved by the US Food and Drug Administration for the treatment of bloodstream infection or endocarditis. The drug is active against enterococci, including multidrug-resistant strains [43]. Two studies evaluating daptomycin in vitro susceptibility for 219 vancomycin-resistant E. faecium isolates from across the United States [44] and 114 isolates from across Europe [45] both showed an MIC range of 0.25–4 mg/L and an MIC90 of 4 mg/L. An MIC of <4 mg/L is classified as susceptible [46]. The combination of daptomycin and gentamicin was shown to be bactericidal and synergistic against 11 of 11 strains of vancomycin-resistant E. faecium [47], and the combination of daptomycin and rifampin was shown to be synergistic in 13 of 19 strains tested. In our case, daptomycin combined with rifampin and gentamicin showed the greatest killing in vitro.
In conclusion, VRE endocarditis is a very uncommon hospital-acquired infection, with only 19 monomicrobial cases individually reported in the literature. Clinical presentation and outcomes are similar to VSE endocarditis. Optimal management remains unknown, but it appears that valve replacement is not absolutely indicated solely on the basis of infection with this organism, and clinicians should strive to provide combination bactericidal antimicrobial therapy.
We thank Dr. Frank Tally, Dr. Judith Steenbergen, and Diane Anastasiou of Cubist Pharmaceuticals, for performance of the time-kill studies.
Potential conflicts of interest. M.P.S. and M.B.E.: no conflicts.
IDSA Members: For your free access to this journal, log in via the IDSA members area.
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy