Mortality-Related Risk Factors and Novel Antimicrobial Regimens for Carbapenem-Resistant Enterobacteriaceae Infections: A Systematic Review

Introduction

The global emergence of antimicrobial resistance poses a threat to human health.¹ Carbapenem-resistant Enterobacteriaceae (CRE) or carbapenemase-producing Enterobacteriaceae (CPE) are Gram-negative bacteria that are resistant to the carbapenem drug class.² The major resistance mechanisms of CRE are: enzyme production, efflux pumps and porin mutations.³ Of these, the production of carbapenemase including KPC, NDM, OXA-48, IMP, and VIM is the main resistance mechanism among CRE.⁴ The KPC enzyme accounts for a high proportion and has the ability to hydrolyze not just carbapenems but also several other antibiotics, leading to high mortality rate.⁵ CRE has become a major public health problem in the last decade due to the gradual increase in carbapenem resistance and the lack of effective antibiotics.^6,7

The infection types of CRE are mainly bloodstream, pneumonia, respiratory, and urinary tract infections (UTIs). CRE infection is associated with increased mortality.^2,6,8 In particular, CRE-caused bloodstream infections (CRE-BSIs) are associated with extremely high mortality, 30%-80%.^9,10 A recent meta-analysis that included 62 studies showed a mortality rate of 54.3% for BSIs and 13.5% for UTIs associated with carbapenem-resistant K. pneumoniae (CRKP).¹⁰

Several studies have evaluated risk factors for CRE mortality but results were inconsistent. These risk factors included Pitt bacteremia score, immunocompromised status, previous exposure to carbapenems, lack of infection source control, and inappropriate antibiotic treatment, etc.^11–14 Nevertheless, many studies have considered antibiotic use as significant risk factor for CRE infection and death.^15,16 The main treatment options for CRE infections are regimens utilizing carbapenem, tigecycline, colistin, aminoglycoside, or ceftazidime/avibactam (CAZ-AVI). The Infectious Diseases Society of America (IDSA) and the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) have provided recommendations for treating CRE infections.^1,17 However, the optimal antimicrobial regimen for CRE infections is unknown as there are controversies regarding which is the safest and most effective antibiotic regimen among the available choices. More importantly, an increasing body of evidence suggests that therapy should be individualized according to the source and severity of the infection or other related factors.¹⁸ Thus, it is difficult to establish the “gold standard” for treating CRE infections.

Our systematic review aimed to explore mortality risk factors in patients with CRE infections and to focus on current evidence on antimicrobial regimens for CRE infections. The results may provide clinical insight into reducing mortality in CRE patients and develop appropriate antibiotic regimens that offer a better prognosis for patients.

Methods

Mortality-Related Risk Factors

Search Strategy

Two authors independently searched the PubMed, Embase, and Cochrane Library databases to identify relevant studies from January 2012 to January 2022. The search strategy contained five core components, which were linked using the AND operator: (1) carbapenem (eg, carbapenem antibiotics), (2) resistance, (3) Enterobacteriaceae (eg, Klebsiella pneumoniae, Escherichia coli), (4) mortality (eg, death rate, case fatality rate), and (5) risk factors (eg, health correlates, the population at risk). Subject headings and free texts (ie, Medical Subject Headings [MeSH] terms) were identified for the five core components. In addition, relevant articles were selected by searching the references identified by this strategy. The complete search strategies are provided in the Supplementary Material (Part 1).

Selection Criteria

The CDC defines CRE as members of the Enterobacterales order resistant to at least 1 carbapenem antibiotic (meropenem, imipenem or ertapenem) or producing a carbapenemase enzyme.¹ Studies were eligible for inclusion if they 1) were hospitalized patients with CRE infections, 2) reported mortality-related risk factors, and 3) were prospective/retrospective observational cohort, case-control studies or randomized controlled trials (RCTs). Exclusion criteria were 1) studies not published in English, 2) reviews, case reports, or experimental studies, 3) studies conducted in patients ≤14 years, 4) studies that did not differentiate between infection and colonization, 5) studies that did not differentiate CRE and other bacteria, 6) studies that had unclear definition and ineligible analysis, and 7) studies that did not provide adequate information.

Quality Assessment

The quality of cohort or case-control studies was assessed based on the Newcastle-Ottawa Scale (NOS) score. Studies with a NOS score ≥ 5 were further analyzed. The scoring details are shown in the Supplementary Material (Part 1).

Data Extraction

Two authors independently extracted relevant data and information from included studies. The following information was collected: first author, publication year, country, study period, study design, pathogen, infection type, the definition of resistance, mortality day, sample size, the numbers of non-survivors, and characteristics of the study population. Data and information on mortality-related risk factors were also extracted.

Data Synthesis

The risk factors were divided into ten groups according to clinical characteristics: demographics, comorbidities, clinical severity assessment scores, hospital-related factors, invasive procedures, type of infection, antibiotic use, clinical index, CRE strain-related factors, and other factors (such as antibiotic resistance and dialysis).

The significance of the association between risk factors and CRE mortality was investigated by examining the statistical data reported in the study. All risk factors with a significant association in the univariate or multivariate analysis were included in the statistical analysis, and odds ratios (OR) for the associations were recorded. Subsequently, we calculated the proportions of studies that reported significance for each risk factor. We also calculated the sample size of each study.

Antimicrobial Regimens

A literature search was performed using the PubMed database from January 2012 to January 2022 to identify studies investigating the treatments of CRE infections. The search strategy contained four core components, which were linked using the AND operator: (1) carbapenem-resistant Enterobacteriaceae, (2) antibacterial agents, (3) treatments and (4) infections. Subject headings and free texts (MeSH terms) were identified for each core component. The search strategy is provided in the Supplementary Material (Part 2).

Studies were eligible for inclusion if they: 1) were hospitalized patients with CRE infections, 2) studied antimicrobial regimens of patients, 3) had reported clinical outcomes of patients treated for CRE infections, 4) were prospective/retrospective observational cohort, case-control studies or randomized controlled trials (RCTs). Exclusion criteria were 1) studies not published in English, 2) reviews, case reports, or experimental studies, 3) studies conducted in patients ≤ 14 years, 4) studies did not differentiate between infection and colonization, 5) studies did not differentiate CRE and other bacteria, 6) studies had unclear definition and ineligible analysis, 7) studies did not provide adequate information, and 8) studies did not included more than 30 cases. The primary outcome of the systematic review was 30-day mortality. When 30-day mortality was unavailable, 14-day mortality, 28-day mortality and in-hospital mortality were extracted.

Results

Mortality-Related Risk Factors

Results of Included Studies

In total, 448 articles were identified through database searching, and 12 additional articles were identified from reference lists. After removing duplicates and literature published before 2012, 391 articles were screened for eligibility, and 289 were excluded after reading the abstract and title. The remaining 102 full-text articles were assessed for eligibility, and 33 studies were included in the analysis.^{11–14,19–47} The flow of the study selection is shown in Figure 1.

Figure 1 Flow diagram of included studies (Mortality-Related Risk Factors). *1. Children/not in adults; 2. Case report/review/experimental studies; 3. CRE colonization; 4. Risk factors about CRE infection; 5. The studies of Extended-Spectrum β-lactamase Producing Enterobacterales (ESBL-E), Pseudomonas aeruginosa, Acinetobacter baumannii and Gram-positive bacteria.

Study Characteristics

Table 1 shows the characteristics of the 33 studies from nine countries or regions. All were observational studies, 28 of which were retrospective and 5 of which were prospective, including 6 case-control, 27 cohort studies. Among the 33 studies, 9 were multicenter studies, and 24 were single-center studies. The sample size ranged from 39 to 661. The most frequently investigated pathogen was CRKP, followed by CRE (including K. pneumonia, Escherichia coli and other CRE pathogens), and CPE. The primary infections were BSIs (15 studies), followed by any infections (mainly pneumonia and UTIs, 14 studies).

Table 1 Characteristics of the Eligible Studies (Mortality-Related Risk Factors)

CRE Mortality-Related Risk Factors

Binary logistic regression analysis model was used to analysis the mortality-related risk factors in all the included studies. The proportion of studies demonstrating an association between chosen risk factors and the mortality of CRE in univariate analysis is shown in Table 2. In particular, only those factors examined in at least two eligible studies were presented. Table 3 shows the significant risk factors ranked according to the proportion of reports. The most reported significant risk factors were antibiotic use (92.9% of studies; 26/28) followed by comorbidities (88.7%; 23/26), hospital-related factors (82.8%; 24/29), and clinical severity assessment scores (82.1%; 23/28) base on univariate analysis.

Table 2 Proportion of Studies Demonstrating an Association Between Chosen Risk Factors and the Mortality of CRE in Univariate Analysis

Table 3 The Proportion of Risk Factors Groups in Univariate Analysis and Multivariate

Furthermore, in univariate analysis, the individual significant risk factors associated with CRE mortality were the Sequential Organ Failure Assessment (SOFA) Score (100% of studies; 5/5), inappropriate definitive therapy (100%; 2/2), the Pitt bacteremia score (85.7%; 6/7), hematologic malignancy (72.7%; 8/11), septic shock (76.2%; 16/21), and bloodstream infections (75%; 6/8). Additionally, in Table 2, no risk factors with an OR value < 1 are shown. The most reported risk factors were microbiological eradication (100% of studies; 4/4), colistin-based combination (80%; 4/5), appropriate antibiotic therapy (55%; 11/20), and combination therapy (50%; 8/16).

The summary of studies that reported a significant association with CRE mortality in multivariate analysis is shown in Table 3. Antibiotic use accounted for the highest proportion of studies (81.0%; 17/21), followed by type of infections (77.8%; 14/18), and clinical severity assessment scores (73.7%; 14/19).

Antimicrobial Regimens

Characteristics of Included Studies

In total, 28 eligible studies were included.^{11,13,21,26,31,32,36,38–41,43,48–63} The characteristics of the studies are presented in Table 4. Only studies that reported mortality as a treatment outcome were presented. Of the 28 studies, 15 were multicenter, and 13 were single-center studies. All were observational studies (24 retrospective; 4 prospective, 5 case-control, and 23 cohort studies). The sample size of the included studies ranged from 36 to 595.

Table 4 Mortality of Infections Caused by CRE Among Different Antibiotic Treatment Regimens

Ten studies focused on CPE, 12 CRKP, and 5 CRE. The primary infections were BSIs (17 studies), followed by pneumonia, respiratory, and urinary tract infections. In 22 studies, mortality at 28 or 30 days was provided. Three studies reported 14-day mortality, and 3 reported in-hospital mortality.

Antimicrobial Therapy and Outcome

All studies reported antibiotic treatment regimens and their associated mortalities. Mortalities ranged from 10% to 59% in 12 studies in patients who received combination therapies. However, mortalities ranged from 8.3% to 64.7% in 11 studies in patients who received monotherapies. Except for the 9 studies focused on CAZ-AVI therapy, 8 studies demonstrated significantly lower mortalities using combination therapies, 4 studies reported lower mortalities using monotherapies, and the remaining 7 studies reported no difference. However, in CAZ-AVI studies, 5 studies identified no substantial differences between CAZ-AVI monotherapy and CAZ-AVI combination therapies. Only one study reported significantly lower mortality in CAZ-AVI combination therapies.

Colistin, tigecycline, aminoglycosides, carbapenems, and CAZ-AVI were the most commonly used monotherapy antibiotics (Table 4). The mortality rates of monotherapies were attributed to use of: polymyxins, 40% to 66.7% in seven studies; tigecycline, 18.2% to 68.4% in six studies; aminoglycosides, 5.8–41% in four studies; carbapenems, 20% to 56.4% in eight studies; and CAZ-AVI, 22% to 47.6% in four studies (excluded 0% and 100%). In contrast, the corresponding mortalities of combination therapies were 22.6%-68.3%, 35%-69%, 32%-50%, 19.9%-53.8%, and 27%-41.5%, respectively. Carbapenem-containing therapies were associated with lower mortality than carbapenem-sparing therapies in three studies.^48,55,57 Two studies reported no significant differences between these two types of therapies.^38,39 In addition, CAZ-AVI- based therapies had substantially lower mortality than other regimens in six studies.

The “Old” Antibiotics

The older antibiotics for treating CRE infections are polymyxins, tigecycline, aminoglycosides, fosfomycin, and aztreonam. Polymyxins and tigecycline have been used as first-line agents to treat CRE infections. However, these monotherapies were often unsatisfactory, and the efficacy was uncertain even when combined with other antibiotics.⁶⁴ Aminoglycosides are limited by nephrotoxicity and are second-line agents due to the availability of newer β-lactams and β-lactamase inhibitor combinations.⁶⁵ The combination of polymyxin and tigecycline showed a good synergistic effect in vitro evaluation.⁶⁶ In vitro synergy was also observed when polymyxins were combined with aminoglycosides or carbapenems.^67,68 However, the clinical effect of synergy has not been identified.

In a single-center retrospective study, the outcome of 89 CRKP-caused BSI cases showed polymyxin-based therapy improved the survival rate compared to tigecycline-based treatment.¹³ Conversely, another nationwide multicenter study (64 patients) analyzed BSIs caused by CRKP (n = 50) and E. coli (n = 14), showing that tigecycline monotherapy was a choice if the strains exhibited the minimum inhibitory concentration (MIC) ≤ 0.5 mg/L, and colistin monotherapy was not suitable.⁴¹ Additionally, another study evaluated the treatment outcomes of a cohort of 36 patients with BSI due to OXA-48-like CPE, found that colistin-based dual combinations and preferably triple combinations were associated with significantly better outcomes when compared to non-colistin-based regimens (P < 0.001).⁴³ Similarly, combination therapy, mostly polymyxin B plus amikacin, showed a survival benefit compared with other regimens in patients with Klebsiella pneumoniae carbapenemase-producing K. pneumoniae (KPC-KP) BSIs.⁵⁴

Despite renal toxicity and second-line status, aminoglycosides still have potential roles in treating CRE infections, especially when combined with newer agents. A small sample size study demonstrated that aminoglycosides had effectively treated CRKP-BSIs if the pathogen was susceptible to aminoglycoside, showing a 75% clinical cure rate.⁶⁵ Two observational studies revealed that aminoglycosides had better clinical outcomes compared to polymyxins or tigecycline in patients with CRE bacteriuria.^69,70 A prospective cohort study in patients with CRKP-UTIs observed that patients treated with aminoglycosides (adjusted hazard ratio HR 0.34, 95% confidence interval CI 0.15–0.73, P=0.0049) were less likely to fail compared to patients treated with tigecycline (adjusted HR 2.92, 95% CI 1.03–5.13, P=0.0425).⁷⁰ Similarly, the clinical success of aminoglycosides was 78.9% compared to other antibiotics (37.0%, P=0.007) in kidney transplant recipients with polymyxin-resistant CRE infections.⁶⁹ Data are limited regarding tigecycline, fosfomycin, and aztreonam treatments for CRE infections.

Carbapenems

Due to the increasing resistance to CRE, carbapenems are no longer reserved as a last-resort therapy for high-risk CRE infections. However, carbapenems in treating CRE infections are still widely debated.⁶ Our analysis showed that carbapenem-containing treatment had lower mortality than other regimens. Dosing strategies of carbapenems for CRE infections include using high doses with prolonged infusion, double carbapenems, or combination with other antibiotics.

Two comparative studies on the efficacy of double carbapenems versus other antibiotics for CRE treatment showed similar results.^55,71 A case-control (1:2) observational two-center study that involved critically ill adults demonstrated significantly lower mortality in patients treated with double carbapenems than standard treatment (ie, colistin, tigecycline, or gentamicin) (47.9% vs 29.2%, P = 0.04).⁵⁵ Likewise, a single-center retrospective study observed that the double carbapenem regimen was also effective compared with the best available regimens in patients infected with CRE, including those with severe clinical conditions, and even in extremely high meropenem MICs.⁷¹

A large sample study suggested that high-dose carbapenem-based combination therapy was a protective factor (HR 0.69, 95% CI 0.47–1.00, P=0.05) for CRKP-caused BSIs, even in high-level carbapenem resistance.⁵⁷ In a small sample study, 19 critically-ill patients with BSIs caused by KPC-KP (MICs ≥16 mg/L) were given combination therapy including meropenem, tigecycline, plus colistin or gentamicin. Meropenem was administered as an extended 3-hour infusion (2 g every 8 hours). High-dose meropenem failed to reach pharmacokinetics/pharmacodynamics targets.⁷² However, another cohort study revealed that high-dose continuous-infusion meropenem optimized using real-time TDM (Therapeutic Drug Monitoring) improved clinical outcomes in the patients infected with KPC-KP (meropenem MIC ≤ 64 mg/L).⁷³ Real-time TDM-guided meropenem may represent a valuable adjunct for optimized care.⁷⁴ Tigecycline and colistin were the two antimicrobials most commonly combined with meropenem,⁷³ but their clinical effects of synergy are not entirely clear.

Ceftazidime-Avibactam

The CAZ-AVI was approved in 2015 to treat complicated intra-abdominal, urinary tract infections and hospital-acquired pneumonia.⁷⁵ Before introducing CAZ-AVI, combination therapy was associated with lower mortality than monotherapy for CRE infections. However, it seems inconsistent when CAZ-AVI was administered to CRE patients. A relatively large multicenter cohort of 138 patients with KPC-KP bacteremia infections revealed significantly lower mortality when treated with CAZ-AVI-containing regimens as salvage therapy after first-line treatment (36.5% vs 55.8%, P =0.005). The results indicated no significant difference in mortality between CAZ-AVI monotherapy and combination therapy.⁵⁹ Subsequently, the largest study published to date confirmed that combination therapies, including CAZ-AVI, were not associated with any significant change compared to CAZ-AVI monotherapy in mortality (26.1% vs 25.0%, P = 0.79),⁵¹ which was also supported by three other observational studies.^11,58,61 On the contrary, lower mortality was observed in 41 critically ill patients treated with CAZ-AVI combined with antibiotics against CRKP infections (24.4% vs 47.6%, P = 0.028) suggesting that tigecycline, carbapenems, and fosfomycin could be optional concomitant antimicrobials.⁴⁹

Several studies analyzed the efficacy of CAZ-AVI regimens compared to other antibacterial regimens on mortality in patients with CRE infections. Two multicenter observational studies compared the effectiveness of CAZ-AVI versus polymyxins for CRE, demonstrating the superiority of CAV-AVI over polymyxins in treating infections caused by KPC-KP or CRKP.^21,53 Interestingly, a potential survival benefit was found in a large cohort study comprising 577 adults with KPC-KP infections treated with prolonged CAZ-AVI infusions (over three hours).⁵¹ Together, preliminary evidence suggests that CAZ-AVI appears to be a promising antibiotic for treating CRE infections. However, this option requires further evaluation.

Discussion

Most of the research on CRE infections was observational studies with a moderate to high risk of bias. It is challenging to perform RCTs on CRE infections due to the different susceptibility of CRE strains and many confounding factors.⁵⁶ The lack of RCTs has hindered the development of guidelines for managing CRE infections.⁷⁶

Several systematic reviews and meta-analyses have been conducted on specific pathogens, such as CRKP and carbapenem-resistant Acinetobacter baumannii (CRAB), focusing on mortality and predictors.^10,16,77 Additionally, two meta-analyses analyzed the association between CRE and mortality.^8,10 Another systematic review analyzed mortality risk factors with carbapenem-resistant Gram-negative bacterial (CR-GNB) infections.¹⁵ No systematic studies or meta-analyses have evaluated mortality-related risk factors for all CRE pathogens.

The duration of antibiotic treatment is controversial. Some studies reported that patients who received a short course of antimicrobial therapy had a poorer prognosis.^43,44 Other studies revealed that the short duration of antibiotic treatment was a protective factor.⁷⁸ The difference may be due to frequent changes in clinical conditions in critically ill patients with CRE infections, and antibiotic regimens are often modified during treatment. Therefore, it is difficult to evaluate the effect of the duration of antibiotic treatment on clinical outcomes.⁴³ The IDSA does not provide recommendations on the duration of therapy. Instead, IDSA advises clinicians that prolonged treatment is unnecessary against infections by resistant pathogens compared to infections caused by the same bacterial species with more susceptible phenotypes.¹

The protective factors with an OR value less than 1 are mainly regarding antibiotic therapy, such as appropriate antibiotic treatment and combination therapies with a carbapenem, suggesting that the proper use of antibiotics may reduce the risk of CRE mortality. Proper antibiotic use has become an essential measure to prevent and treat CRE infections.⁷⁹

Few monotherapy studies, except for CAZ-AVI, reported lower mortality outcomes, partly because patients who received monotherapies had less severe symptoms or a quickly controllable source of infection.⁴⁸ The ESCMID guidelines recommended that “old” antibiotics, including polymyxin, tigecycline, and aminoglycosides, be considered in patients with non-severe CRE infections. Newer antibiotics (meropenem-vaborbactam or ceftazidime-avibactam) are used in critically ill patients.¹⁷ In clinical practice, combination therapies are commonly administered to patients with severe infections. However, studies have shown that the efficacy of combination therapies is uncertain. The in vitro synergy of specific antibiotics may not always translate into clinical effects. Dosages and duration of antibiotics and the susceptibility profiles of CRE pathogens may affect the treatment effectiveness.^17,80

The effectiveness of colistin monotherapy was not satisfactory principally because the suboptimal dosing could not reach appropriate plasma concentrations. Still, it would increase the risk of death, particularly in severely ill patients with renal dysfunction.⁸¹ In our analysis, colistin monotherapy was a mortality-related risk factor for CRE infections. Furthermore, a study supported a survival benefit in colistin-based dual combinations, preferably in triple combinations.⁴³ Therefore, combination with other in vitro active antibiotics might be the optimal option when treating CRE patients with colistin. In addition, aminoglycosides were more effective than polymyxins for treating CRE bacteriuria based on the ESCMID guidelines.

Carbapenem-containing regimens for CRE infections have been a long-standing topic of debate. The ESCMID guidelines suggested that clinicians should avoid carbapenem-containing combination therapies for CRE infections unless the MIC of meropenem is 8 mg/L.¹⁷ However, IDSA guidelines recommend that meropenem should be avoided if isolates are carbapenemase producers, despite susceptibility to meropenem.¹ Our review obtained favorable outcomes when carbapenem-containing regimens were administered to CRE patients, and three studies supported a better outcome in patients treated with carbapenem-containing therapies than other treatments.^48,55,57 In contrast, two studies reported no differences.^38,39 Furthermore, high-dose continuous-infusion meropenem optimized by real-time TDM improved clinical outcomes even when there were extremely high meropenem MICs,^73,74 which can be an option for clinicians.

In addition to CAZ-AVI, other novel antibiotics against CRE infections have been approved or in advanced clinical development, including ceftolozane-tazobactam, meropenem-vaborbactam, and imipenem-relebactam.^82,83 We cannot accurately evaluate their effectiveness and safety due to the lack of data available for these new antibiotics against CRE. Preliminary evidence revealed a potential role of CAZ-AVI in patients with CRKP infections. Our review and two other meta-analyses demonstrate no substantial difference between CAZ-AVI monotherapy and CAZ-AVI combination therapy.^84,85 More post-marketing data from real-world studies and RCTs are needed to evaluate the effectiveness and safety of CAZ-AVI in treating CRE infections. However, the drug resistance of CAZ-AVI has gradually increased in recent years, and their effectiveness has been decreased due to β-lactamase production, efflux pumps and target modifications.⁸⁶

Our systematic review has the following limitations: 1) No RCTs were included. 2) The size of included studies was small. 3) Owing to the high heterogeneity of the included studies in terms of study design, patient populations and CRE pathogens, and so comparative statistical analysis or meta-analysis of the results was not possible. and 4) In addition to antibiotic use, risk factors such as complications and septic shock also accounted for a high proportion of the dead patients. We could not control for these variables when analyzing the efficacy of antimicrobial regimens due to the limited data. It needs further investigation whether the patient’s antibiotic regimen was different in the sepsis/non-sepsis group, organ dysfunctions/non- organ dysfunctions, or mild/critical patients’ group.

Conclusions

Our systematic review has explored mortality-related risk factors and antimicrobial regimens of CRE infection. According to our review, antibiotics use, patients’ comorbidities, and hospital-related factors are the most important mortality risk factors in patients with CRE infections. Combination therapies may offer a comparative advantage over monotherapy except for CAZ-AVI. When treating CRE infections, colistin monotherapy should be avoided. Aminoglycosides can be used for CRE bacteriuria. High-dose continuous-infusion meropenem and double carbapenems regimens could be considered. CAZ-AVI appears to be a promising drug for treating CRE infections, especially those involving bacteremia. Clinicians must consider mortality-related risk factors, and treatment should be individualized based on the source and severity of the disease.