Introduction
Due to its elaborated and tedious treatment regimens, as well as its associated morbidity and mortality, infective endocarditis (IE) frequently poses a challenge to the attending physician in clinical practice. Endocarditis can be caused by various different microorganisms, with
Enterococci spp. representing the third most common group of pathogens, with an estimated frequency of 10% of all cases [
1]. Among all cases of enterococcal IE,
Enterococcus faecalis further represents the predominant strain of
Enterococci, with a prevalence of almost 90% [
2].
To date, the European Society of Cardiology (ESC) recommends both a combination of two cell wall inhibitors, namely ampicillin plus ceftriaxone for 6 weeks (AC), or one cell wall inhibitor with an aminoglycoside antibiotic, i.e., amoxicillin/ampicillin plus gentamicin for 2–6 weeks (AG), with equivalent class IB recommendations for the treatment of
E. faecalis IE [
3]. However, alarmingly high rates of high-level aminoglycoside-resistant strains of
E. faecalis and
E. faecium (HLAR; aminoglycoside Minimal Inhibitory Concentration (MIC) > 500 mg/L) have been reported in previous studies [
2,
4,
5], which may preclude the proclaimed synergistic bactericidal effects of a combination of cell wall inhibitor plus aminoglycoside [
6].
Furthermore, treatment with aminoglycosides is associated with an increased risk for adverse effects, such as drug-induced nephrotoxicity or ototoxicity [
7‐
9], and the dosing and monitoring of these antibiotics can be quite challenging in clinical practice, especially in patients with already impaired kidney function, elderly or obese patients [
10,
11].
Previous observational studies reported that combination therapy with AC could be non-inferior to treatment with AG in patients with
E. faecalis IE in terms of mortality and relapses, as well as beneficial regarding the risk for drug-induced adverse events [
12‐
14]. However, there is currently no high-level evidence available to support this notion. Therefore, we aimed to provide a comprehensive meta-analysis on all comparative studies regarding combination therapy of AC vs. AG in
E. faecalis IE to further support therapeutic decisions in affected patients.
Methods
The study was conducted according to the current PRISMA guidelines [
15], the principles of Good Clinical Practice and the Declaration of Helsinki. Prior to literature research, the study protocol was registered at the international prospective register of systematic reviews PROSPERO (ID: CRD42021268601) [
16].
Data sources and search strategy
Literature search was performed by two independent reviewers (MM and LS) using predefined search terms (‘endocarditis’, ‘aminoglycoside’, ‘gentamicin’, ‘beta lactam’, ‘cephalosporin’, ‘ceftriaxone’, ‘penicillin’, ‘ampicillin’, ‘amoxicillin’, ‘enterococcus’ and ‘enterococ*’) in the databases PubMed/MEDLINE, Cochrane Database of Systematic Reviews (CDSR), Cochrane Central Register of Controlled Trials (CENTRAL), Cochrane Clinical Answers (CCAs), EBM Reviews—Database of Abstracts of Reviews of Effects, Web of Science and Latin American and Caribbean Health Sciences Literature (LILACS) in June 2021. Additionally, manual search was performed by screening all references of eligible studies.
Study selection
Studies were considered eligible if (P) patients included were ≥ 18 years of age and had infective endocarditis with
Enterococcus faecalis, (I) treatment with ampicillin–ceftriaxone (AC) was compared to (C) ampicillin–gentamicin (AG) and (O) outcomes on in-hospital mortality, nephrotoxicity and adverse events requiring drug withdrawal were reported. Exclusion criteria were: studies including patients < 18 years of age, studies on prophylactic treatment, animal studies or in vitro studies, studies on blood culture negative infective endocarditis (BCIEN), studies where data could not reliably be extracted, overlapping data, abstract-only papers, conference papers, editorials, books, articles without available full text, case reports and case series, as well as systematic reviews. No limits were set regarding the language and publication date of studies. Infective endocarditis was defined by the modified Duke criteria [
17].
Outcomes
The primary outcome was in-hospital mortality. Secondary outcome measures were 3-month mortality, nephrotoxicity, adverse effects requiring drug withdrawal, relapse within 3 months of follow-up (defined as isolation of E. faecalis in blood specimens within the follow-up period) and treatment failure requiring change of antibiotic regimen.
Statistical analyses
Statistical analyses were performed with R (version 4.0.2., R Core Team (2013), R Foundation for Statistical Computing, Vienna, Austria;
http://www.R-project.org/) using the packages ‘Rcmdr’, ‘ggplot2’, ‘meta’, ‘metafor’, ‘dplyr’ and ‘dmetar’. Odds ratios and 95% confidence intervals (CI) were calculated using random-effects models with the Mantel–Haenszel method, the Sidik–Jonkman estimator for
τ2 and the Hartung–Knapp adjustment. Continuity correction was performed with treatment arm continuity correction (TACC). Between-study heterogeneity was assessed by calculating
τ2, Higgin’s &Thompson’s
I2, Cochran’s
Q, prediction intervals and by Baujat plot. Risk of bias assessment was conducted using the rob.summary() function of the ‘dmetar’ package for R, which is based on the Cochrance Risk of Bias Tool. A
p value < 0.05 was considered statistically significant.
Discussion
Infective endocarditis (IE) remains one of the most challenging diseases in modern cardiology. With an ever-increasing number of cardiac and extracardiac devices implanted, as well as an increasingly old and multimorbid population, the incidence of endocarditis has been slightly rising in recent years in parts of the world [
23‐
28]. Furthermore, despite the availability of clinical practice guidelines and improvements in therapy itself, mortality of IE remains high with an estimated 12-month mortality rate of 15–30% reported in most studies [
3,
27,
29]. Besides morbidity and mortality exerted by the infection itself, the intensive and tedious antibiotic treatment regimens of IE are in part associated with an increased risk for adverse outcomes, such as drug-induced nephrotoxicity or ototoxicity in patients treated with aminoglycosides, rendering endocarditis and its necessary therapies a potentially debilitating disease entity for affected patients [
30].
Currently, guidelines by the European Society of Cardiology (ESC) recommend either a combination of ampicillin plus ceftriaxone (AC) or amoxicillin/ampicillin plus gentamicin (AG) in the treatment of
Enterococcus faecalis infective IE [
3]. However, among the aforementioned antibiotic drugs, gentamicin is associated with the highest risk of drug-induced adverse events, including potentially irreversible complications such as vestibulotoxicity or ototoxicity [
7,
8]. As guidelines currently state an equivalent class IB recommendation for both antibiotic regimens [
3], the aim of this meta-analysis was to investigate the therapeutic non-inferiority of AC versus AG, and to elucidate whether AC would be associated with a lower risk of drug-induced adverse events.
After including four retrospective, nonrandomized, observational cohort studies [
12,
18‐
20] in this meta-analysis, we found that the AC regime was indeed non-inferior to AG in
E. faecalis endocarditis, as depicted by no statistically significant differences in hospital mortality (
p = 0.740), 3-month mortality (
p = 0.072), relapses (
p = 0.8066) or treatment failure (
p = 0.1847) between the two treatment arms investigated. Furthermore, the presumed lower toxicity of the AC regime was confirmed by a significantly lower prevalence of nephrotoxicity (OR 0.45 [0.26–0.77],
p = 0.0182) and drug withdrawal due to adverse events (OR 0.11 [0.03–0.46],
p = 0.0160) when compared to patients receiving AG.
Our findings, showing non-inferiority of the AC regime, as well as a significantly reduced risk of drug-induced adverse events, give rise to speculation that AC might be favourable over AG for affected patients with
E. faecalis endocarditis. Especially in highlight of an increasingly old and multimorbid population, with a steadily incrementing prevalence of chronic kidney disease [
31], the aforementioned findings should be taken into consideration when clinical decisions on antibiotic treatment in enterococcal IE are to be made. Moreover, the fact that high-level aminoglycoside-resistant (HLAR) strains of
Enterococci increasingly limit the use of AG in IE [
6,
32] further strengthens the notion to abandon AG in favour of AC in this disease entity.
However, while exhibiting a good tolerability profile [
33] and broad antimicrobial spectrum, treatment with ceftriaxone was previously identified as a major risk factor for
Clostridium difficile infection, as well as for colonization with vancomycin-resistant
E. faecium (VRE) or other multi-resistant bacteria [
34‐
37]. The link to colonization with these pathogens
, and its potential implications for the individual patient, as well as the patient’s environment, should be considered when therapeutic decisions on antimicrobial treatment in IE are to be made.
Furthermore, both AC and AG are currently implemented as therapeutic regimens administered via repeated intravenous infusions. However, the Partial Oral Treatment of Endocarditis (POET) trial [
38] recently found that a change to oral antibiotics after an initial intravenous phase was non-inferior to continued intravenous treatment in patients with IE. Since a reduction of hospital stay diminishes the risk of complications, associated costs, as well as psychological distress for our patients [
38], outpatient treatment appears to be a promising opportunity in patients with
E. faecalis IE as well. However, treatment with cephalosporins was not investigated in the POET trial, and bioavailability of these antibiotic drugs is unfortunately comparatively low [
39]. Nonetheless, novel formulations are being investigated at the moment, which could help to put outpatient treatment with AC in patients with
E. faecalis IE into practice in the near future [
40,
41].
Finally, other antibiotic regimens are also depicted in the ESC guidelines on enterococcal IE, and can thus also be considered in affected patients [
3]. However, vancomycin with gentamicin has a weaker class IC recommendation than AC or AG, and treatment with daptomycin is currently recommended in patients with strains resistant to aminoglycosides, beta-lactams and vancomycin only. Furthermore, a recent study by Cerón et al. showed worse microbiological and clinical responses in patients treated with daptomycin when compared to standard treatment [
22].
In conclusion, treatment with AC was non-inferior to treatment with AG, and it was associated with a reduced prevalence of nephrotoxicity and drug withdrawal due to adverse events. Thus, combination therapy with AC appears favourable over AG in patients with E. faecalis IE.
Limitations
Our study has several limitations. First, all four included studies were relatively small, resulting in a total
N of only 599 patients included in the current meta-analysis. The low number of included patients, as well as the potential statistical implications thereof, has to be considered when interpreting the results of our study. Moreover, the definition of acute renal failure was slightly different between included studies (also see Table
1), which is why a certain bias on the secondary outcome measure ‘nephrotoxicity’ cannot be excluded. Of note, we only included data of patients with creatinine increase ≥ 50% from the study by El Rafei et al. [
20,
42] to meet the definitions of the other included studies. In one study, the definition of acute renal failure was initially missing [
19]; however, the corresponding author could provide the definition used when contacted.
Also, between-study heterogeneity was assessed by calculating
τ2,
I2,
Q and prediction intervals. Here, we found no substantial heterogeneity; however, a low number of included studies (
k = 4) can result in a bias of heterogeneity measures [
43,
44], which is a known problem of small meta-analyses. Hence, a certain degree of between-study heterogeneity cannot certainly be excluded (see also Supplementary Fig. 2). Also, a low
k is a problem when assessing publication bias, especially when small study effects should be calculated. Hence, a certain degree of publication bias can also not certainly be excluded. However, for visual estimation, contour-enhanced funnel plots are additionally depicted in Supplementary Fig. 3. The possible biases arising from a low number of included studies should be considered when interpreting the findings of our meta-analysis.