Antimicrobial agents active against carbapenem-resistant Escherichia coli and Klebsiella pneumoniae isolates in Lebanon

Introduction: It is not yet clear which antimicrobial agents should be used to treat the ominously increasing infections with carbapenem-resistant (CR) bacteria. We therefore investigated the activity of different antimicrobial agents against CR Escherichia coli and Klebsiella pneumoniae in Lebanon. Methodology: This retrospective study assessed the minimum inhibitory concentrations (MICs) of three carbapenems (by Etest), as well as the in vitro activity of eight other antimicrobials (by disk diffusion) against CR E. coli (n = 300) and K. pneumoniae (n = 232) isolates recovered at a major University Medical Center in Lebanon. Results: Higher percentages of isolates showing carbapenem MICs of ≤ 8 μg/mL were noted among the CR E. coli compared to the CR K. pneumoniae for ertapenem (48% vs 27%), imipenem (74 % vs 58%) and meropenem (82% vs 63%). Among the eight other antimicrobials, activity was generally higher when the MICs for the three carbapenems were ≤ 8 μg/mL. Regardless of the MIC level of the three carbapenems, very low susceptibility rates (≤ 33%) were noted for ciprofloxacin, trimethoprim-sulfamethoxazole and aztreonam against both E. coli and K. pneumoniae isolates. With Amikacin, higher susceptibility rates were seen against E. coli isolates (81%-97%) than against K. pneumoniae isolates (55%-86%), also reflecting higher activity than gentamicin (44%-54%). The best activity (66%-100%) was observed for tigecycline, colistin and fosfomycin against both CR species. Conclusions: Based on the in vitro findings in this study, the combination of a carbapenem showing an MIC of ≤ 8 μg/mL together with an active colistin, tigecycline, or fosfomycin, would offer a promising treatment option for patients infected with CR E. coli or K. pneumoniae.


Introduction
Globally, infection with carbapenem resistant (CR) Escherichia coli (Ec) and Klebsiella pneumoniae (Kp) as well as other CR Enterobacteriaceae (CRE) had been on the rise, often accompanied by high rates of resistance to a wide range of antimicrobials.These pathogens, together with the dwindling antimicrobial armament, constitute a most concerning contemporary threat to health in many countries [1][2][3].The Center for Disease Control and Prevention in the USA noted an increase in the percentages of CRE from 1.2% in 2001 to 4.2% in 2011, where Klebsiella spp.accounted for the highest increase in proportion, a rise from 1.6% to 10.4% [4].In Lebanon, a notable increase was observed at our medical center in the percentages of CR, from 2010 to 2016 for Ec (0.1% to 5%) and Kp (0.7% to 8%), as well as in other hospitals [5,6].
The ominous spread of these pathogens together with the scarcity of treatment options results in a serious health impact, seen in high rates of morbidity and mortality (up to 75% of infected patients) and increasing health costs.This situation is a major challenge for the treating clinicians and infection control professionals [1][2][3]7,8].In Lebanon, similar problems and challenges have been encountered since CRE pathogens were introduced in this country several years ago [9,10].As a result, several studies have been conducted to describe the phenotypic and molecular features of CR E. coli and K. pneumoniae in this country [5,[11][12][13][14][15].
Because of the very limited available options to treat CR isolates, this study was undertaken to identify the types and percentages of antimicrobial agents active against CR E. coli and K. pneumoniae resistant to ertapenem, imipenem, and/or meropenem.

Bacterial Isolates
Consecutive non-duplicate isolates of carbapenem resistant E. coli (300 isolates) and K. pneumoniae (232 isolates) at the Clinical Microbiology Laboratory, Department of Pathology and Laboratory Medicine, American University of Beirut Medical Center (AUBMC) during the period March 2008 to June 2016 were investigated.Identification of the isolates was based on standard biochemical methods [5].
For screening of carbapenemase-producers, ertapenem disks (10 g) were used.Those isolates showing < 25 mm zone of inhibition were then tested by Etest to confirm the MICs of ertapenem, imipenem and meropenem.

Quality Control
The American Type Culture Collection (ATCC) quality control strains of E. coli (ATCC 25922) and Pseudomonas aeruginosa (ATCC 27853) were used to ensure correct performance of the Etest and disk diffusion methods.When received in February 2016, the quality control reference strains for colistin [CDC E. coli reference strains MCR-1 (AR-Bank # 0349, specified colistin MIC of 2-4 µg/mL) and MCR-1 (AR-Bank # 0346, specified colistin MIC of 4 µg/mL)] were also used to ensure the quality of testing.

Statistical Analysis
Our data was analyzed using Stata v. 13 software Package (Stata Corp LP, College Station, Texas, USA) for determining the p value, considering less than ≤ 0.05 as significant.
The distribution of the MICs (µg/mL) for ertapenem, imipenem and meropenem among these CR E. coli and K. pneumoniae isolates are shown in Figures 1 and 2. The MIC (µg/mL) level for the three carbapenems ranged between ≤ 0.25 and ≥ 32.The  figures reflect the distribution of the MICs up to ≥ 12 µg/mL and not up to ≥ 32 µg/mL due to space limitation.
The distribution of the CR E. coli and K. pneumoniae isolates according to their overall carbapenems MICs, as well as carbapenems MICs of ≤ 8 or ≥ 8 µg/mL in relation to their susceptible rates to other antimicrobial agents, namely: amikacin, gentamicin, ciprofloxacin, trimethoprimsulfamethoxazole, tigecycline, colistin, fosfomycin and aztreonam are presented in Table1.
Three of the antimicrobial agents tested, ciprofloxacin, trimethoprim-sulfamethoxazole and aztreonam, showed very low susceptibility rates (≤ 33%) regardless of the MICs level against both E. coli and K. pneumoniae isolates, and with little difference among the three carbapenems.Among aminoglycosides, the susceptibility rates of gentamicin were low, ranging between 44% and 54%, for both E. coli and K. pneumoniae isolates, and with little difference among the 3 carbapenems.On the other hand, the E. coli isolates showed higher susceptibility rates to amikacin (range: 81% to 97%), compared to the K. pneumoniae isolates (range: 55% and 86%), as noted for the three carbapenems and regardless of the MICs category.Tigecycline, colistin and fosfomycin showed high activity (66%-100%) against both CR E. coli and K. pneumoniae, regardless of the MICs levels of ertapenem, imipenem and meropenem (Table 1).

Discussion
The clinical microbiology laboratory (CML) has a key role in promptly detecting and reporting multi-drug resistant (MDR) pathogens to guide targeted antimicrobial therapy, which will influence outcome.In this context, the CML at AUBMC reported the first alert about CRE in Lebanon, when the first imipenem resistant K. pneumoniae that harbors the bla-OXA-48 gene was recovered from the urine of an 8-year-old girl in 2008 [9].
Two years later (July 2010), three Iraqi patients presented at our medical center seeking treatment.CR E. coli and K. pneumoniae were isolated from different specimens of these patients, and the isolates harbored not only the bla-OXA-48 gene but also the very threatening bla-NDM-1 gene, producing the novel New Delhi Metallo ß-lactamase (MBL) gene [10].Subsequently, the incidence of CR Enterobacteriaceae continued to rise [5].Such critically resistant pathogens warranted publishing an overview to inform the medical and paramedical community on the local and regional epidemiology of carbapenem resistance, the mechanisms involved, screening and detection methods, as well as their treatment and control [11].Generally, physicians rely on the in vitro antimicrobial susceptibility results of pathogens to provide optimum therapy to their patients.Empirical combination therapy for CRE infected patients is usually based on the local resistance epidemiology.Definitive therapy, however, should be guided by determination of the in vitro antimicrobial susceptibility profiles especially when considering combined treatment for CRE pathogens.This is very important since only very few drugs remain as last-resort agents (e.g.colistin, fosfomycin, tigecycline, and aminoglycosides) [18][19][20][21][22].
In our study, the in vitro activity of these last-resort agents against the CR E. coli and K. pneumoniae isolates detected at our institution are the core of the following discussion.These results will be considered especially in relation to updates reported in the recent scientific literature [18,19].
Colistin has significant activity against various carbapenemase-producing isolates and is often used in combination therapy (e.g., with aminoglycosides, aztreonam, carbapenems, rifampin, tigecycline, or fosfomycin) with a high (82.1%)clinical cure rate [18,23,24].Caution in its use is advised, to avoid development of colistin-resistant Ec & Kp.In our study, the overall high susceptibility rates (82%-84 % ) of both Ec and Kp to colistin makes it an important first option to use in the treatment of such CR infections (Table 1).
Fosfomycin has also been considered as part of a combination regimen (including at least one more active agent, e.g. with tigecycline and colistin) in the treatment of invasive infections with CRE pathogens [25].Its use in combinations, however, resulted in varied synergistic activity against CRE depending on the type of pathogen and the antibiotics used [26].For example, Samonis et al. [27] reported that the synergistic effect of fosfomycin was 55% to 79% with carbapenems, 7.1% to 36% with colistin, 21% to 30% with tigecycline, and 25% to 43% with gentamicin.In this study, fosfomycin showed high activity (89-91%) against the CR Ec, while the susceptibility of CR Kp was lower (66-73%).We conclude that fosfomycin can remain a viable option for the treatment of CR pathogens especially Ec (Table 1).These levels are close to those we reported in 2012 on the fosfomycin susceptibility of ESBL-producing pathogens, where its activity was higher against ESBL-producing E. coli isolates (86%) than against K. pneumoniae isolates (62%) [28].
Aminoglycosides have been used in combination with other classes of antibiotics with different results [18,27].For example, some K. pneumoniae carbapenemase (KPC) and OXA-48 -producers in K. pneumoniae isolates remain susceptible to gentamicin, while this is rare for New Delhi metallo beta-lactamase (NDM) -producers.An in vitro synergistic effect was observed when netilmicin and fosfomycin were tested against a wide range of resistant pathogens.Also, a combination of carbapenems and aminoglycosides showed in vitro and in vivo (in animals) synergistic effects against some KPC-producing K. pneumoniae isolates, but studies with carbapenemases other than KPC are lacking [18,27].A lack of effect was noted when a combination of gentamicin and fosfomycin was tested against KPC-2-producing K. pneumoniae.In our current study, the overall higher activity of amikacin compared to that of gentamicin against both CR Ec and Kp may be attributed to the type of carbapenemase gene involved.For example, in the clinical setting, a study from Greece reported a favorable outcome in five patients infected with KPC-2-producing K. pneumoniae who were treated with aminoglycoside plus colistin (plus tigecycline in 2 patients) [29].
Tigecycline has also been recommended as part of the initial combination regimen in the treatment of patients with CRE infections in different body sites, other than UTI, especially when tigecycline MIC is ≤ 1 mg/L (the typical MICs 90 was 0.5 mg/L for E. coli and 1-2 mg/L for K. pneumoniae).The overall crude mortality when treating with tigecycline was higher with monotherapy than with combined treatment (40-80% vs 0-33%) [18].In our study, tigecycline showed very high in vitro activity against both CR Ec and Kp, making it an excellent agent in combined treatment options for such CR pathogens.Low in vitro resistance rates against tigecycline were also reported in a 2008 study from our institution, in MDR and ESBL Ec (0%) and Kp (3%) [29].The very low rates noted in the previous and current studies indicate a sustainable activity of tigecycline against resistant isolates at our center.
Aztreonam testing was included in this study because MBLs can hydrolyze carbapenems and all the available beta-lactams with the exception of aztreonam.Whether aztreonam remains an option for treating infections due to MBL producers that test susceptible to this agent remains to be determined; currently there is no clinical experience with aztreonam for the treatment of invasive infections due to CRE [30].In our study, however, a very low aztreonam susceptibility (14-25%) was shown by both CR Ec and CR Kp isolates, indicating that the majority of these isolates were not MBL producers.An earlier study on the CR genes involved in K. pneumoniae and E. coli isolates from our center indicated the prevalence of bla-OXA-1, bla-CTXM-15, bla-TEM-1, bla-CMY-2, bla-OXA-48, and porin channel genes, and few had the NDM-1 gene [12].
The antimicrobial agents discussed above have been considered for possible clinical relevance to treat patients infected with CRE.The question of the superior effectiveness of mono-or combined antimicrobial treatment of patients infected with CRE pathogens was excellently reviewed by Rodríguez-Baño and colleagues in Spain and Karaiskos and Giamarellou in Greece, looking at microbiological diagnosis and treatment of infections with carbapenemase-producing Enterobacteriaceae [18,19].These comprehensive studies suggest a more favorable outcome (e.g.reduced mortality) using a combined antimicrobial treatment for patients infected with CRE pathogens.Besides, combined therapy may maximize bacterial killing (synergistic effect) and minimize bacterial resistance.For example, Tumbarello et al. 2012 [31] reported a higher mortality rate among patients treated with monotherapy compared to combined therapy (54.3% vs 34.1%); the combination included tigecycline, colistin, and meropenem.
In addition, the inclusion of a carbapenem in the combined treatment showed promising results.For example, Qureshi et al., reported lower mortality when carbapenem was used in combination with colistin or tigecycline, compared to a combination that lacked a carbapenem (12.5% vs 66.7%) [32].Moreover, the effect of carbapenems in the combination was noticeable if the carbapenem MIC was ≤ 8 mg/L, and it was added to one or two in vitro fully active drugs (including colistin, tigecycline, an aminoglycoside or fosfomycin) for such infections.Carbapenems, despite being hydrolyzed by carbapenemases, may retain some activity against carbapenemase-producing isolates [18,19,33,34].Such findings were vividly reflected in the study by Daikos et al. [20] who reported a lower mortality rate with treatment by a carbapenemcontaining combination with another active drug, such as an aminoglycoside, colistin or tigecycline, than with a combination without a carbapenem (19.3% vs 30.6%).It was important that the carbapenem had a MIC ≤ 8 mg/L [20].Akova et al. found similar results [35].
Interestingly, the efficacy of using dual carbapenem (i.e., meropenem plus ertapenem) therapy has been shown in animal model of infection (i.e., mouse pneumonia) and in humans infected with KPC producers.The suggested explanation was that ertapenem most likely acts as a "suicide" molecule for carbapenemase activity, whereas the more active drug, meropenem, retains its efficacy [36].
In our study the carbapenem rates of MIC ≤ 8 mg/L for ertapenem, imipenem and meropenem were higher when tested against E. coli: 48%, 74% and 82%, respectively, than against K. pneumoniae: 27%, 58% and 63%, respectively.Such rates, especially for imipenem and meropenem, in combination with the other in vitro active agents, can provide an adequate option in treating patients, particularly those with CR E. coli infection.
Studies at our medical center since 2008 have been addressing molecular characterizations, resistance mechanisms, genes involved, and other aspects of CR Ec and Kp isolates [9,10,[12][13][14].The first CRE detected in Lebanon was a K. pneumoniae recovered from a 7year-old female child, which was characterized by PCR experiments using primers for multiple β-lactamase and carbapenemase genes: TEM, SHV, CTX-M, GES, KPC, IMI, OXA-1 ⁄ 2 ⁄ OXA-10 ⁄ 18, OXA-23 ⁄ 58, IMP, VIM, SPM, GIM and SIM.The only positive result was for a blaOXA-48-like gene [9].Interestingly, this was only the fourth report about this gene worldwide.Subsequently, increasing numbers of CRE have been noticed; some harbored the very threatening bla NDM-1 gene, detected for the first time in Lebanon, in E. coli and K. pneumoniae isolates recovered from Iraqi patients coming for medical care at AUBMC [10].
The underlying mechanisms and genes responsible for the carbapenem-resistance in K. pneumoniae and E. coli revealed the presence of different Beta-lactamase gene profiles including: bla-OXA-1, bla-CTXM-15, bla-TEM-1, bla-CMY-2, bla-OXA-48 and NDM-1 genes in both genera; in addition, the K. pneumoniae isolates were found to lack outer membrane porin (OmpC and OmpF) encoding genes, while E. coli harbored these porin genes [12].In addition, the prevalence of carbapenem resistance encoding genes and their correlation with corresponding MIC 90 against ertapenem, imipenem and meropenem were studied among CR E. coli (n=76) and CR K. pneumoniae (n=54) isolates.The prevalence of blaOXA-48, blaNDM-1, blaTEM-1 and blaCTX-M-15 among the E. coli isolates were 36%, 12%, 20% and 80%, respectively, while among K. pneumoniae isolates they were 37%, 28%, 28% and 72%, respectively.The presence of more than one carbapenem resistance encoding gene and/ or ESBL encoding gene did not have an effect on the MIC 90 value in K. pneumoniae isolates, while in E. coli they resulted in higher MIC 90 values [14].
We also assessed the effects of antimicrobial combination therapy against bacteria with different CR genes using BALB/c mice [13].The mice were injected with carbapenem-resistant Enterobacteriaceae strains, harboring either bla CTXM-15 , bla CTXM-15 and bla OXA-48 , bla NDM-1 , or bla KPC-2 genes.The qRT-PCR revealed a significant decrease of transcript levels in all isolates upon using rifampicin or tigecycline, singly or in combination with colistin.However, variable levels were obtained using colistin singly or in combination with meropenem or fosfomycin.In vivo assessment showed that all combinations used were effective against isolates harboring bla CTXM-15 , bla OXA-48 , and bla NDM-1 .Conversely, the most significant combination against the isolate harboring the bla KPC-2 gene was colistin with either carbapenem, fosfomycin, or kanamycin.Essentially, the findings indicated that combination therapy selected based on the type of carbapenemase produced, appeared to be non-toxic and effective in BALB/c mice [13].Therefore, an approach based on gene detection may be useful to extrapolate for human use, as a step towards optimizing combination therapy and antimicrobial stewardship when in treating patients with CRE infections.The testing and logistics involved would, however, have to be feasible for the prompt and real time therapeutic practice.
The search for new and effective antimicrobial agents for optimal treatment of these highly threatening pathogens is continuing, especially for one that can cope with covering all genes involved in CR status.For example, one of the most recently introduced agents is avibactam with ceftazidime.Although this new agent showed promise for treating infections due to K. pneumoniae KPC producers, it lacks effectiveness against NDM producers [23].

Conclusions
The in vitro susceptibility testing of CRE and other pathogens is of utmost importance to provide local antimicrobial resistance epidemiologic data for selection of possible treatments.Our in vitro study highlighted the available options of antimicrobial agents to treat infections with CR K. pneumoniae and E. coli isolates in our institution.A larger nationwide study would be needed to reflect the overall situation in Lebanon.However, the control of such pathogens necessitates not only this technical information but also active and adequate antimicrobial stewardship program together with appropriate infection control measures following expert guidelines [30].

Table 1 .
Distribution of CR E. coli (n = 300) and K. pneumoniae (232) isolates according to carbapenem MICs in relation to susceptibility rates (%) of eight other antimicrobial agents.