The Synergistic Activity and Optimizing Doses of Tigecycline in Combination with Aminoglycosides against Clinical Carbapenem-Resistant Klebsiella pneumoniae Isolates

Carbapenem-resistant Enterobacteriaceae (CRE), especially carbapenem-resistant Klebsiella pneumoniae (CRKP), are among the largest pathogenic threats to humans. The available antibiotic treatment options for combating CRKP are limited. Colistin-resistant Enterobacteriaceae (CoRE) have also been reported worldwide, including in Thailand. Therefore, this study aimed (1) to determine minimum inhibitory concentrations (MICs) and synergistic activities of antibiotics of CRKP, and (2) to determine the probability target of attainment (PTA) and cumulative fraction of response (CFR) using pharmacokinetic/pharmacodynamic (PK/PD) data. Clinical CRKP isolates were obtained from Phramongkutklao Hospital (June to November 2020). Broth microdilution and checkerboard techniques were used to determine the mono- and synergistic activities of antibiotics. Carbapenemase and mcr-1 genes were also identified by polymerase chain reaction (PCR). The optimal antibiotic regimens were evaluated using Monte Carlo simulations. Forty-nine CRKP isolates were collected, 40 of which were CoRKP strains. The MIC50 and MIC90 of tigecycline, amikacin, and gentamicin were 1 and 2 µg/mL, 4 and 16 µg/mL, and 0.25 and 4 µg/mL, respectively. None of any isolates expressed the mcr-1 gene, whereas blaOXA-48 (53.1%) and blaOXA-48 plus blaNDM (42.9%) were detected. Synergistic activity was observed in 8.2% of isolates for tigecycline combined with amikacin or gentamicin. Additive activity was observed in 75.5% of isolates for tigecycline-amikacin and 69.4% for tigecycline-gentamicin, and no antagonism was observed. High-dose antibiotic regimens achieved the PTA target. The general recommended dose of combination regimens began with 200 mg tigecycline and 25 mg/kg amikacin, or 7 mg/kg gentamicin, followed by 100 mg tigecycline every 12 h and 15 mg/kg amikacin or 5 mg/kg gentamicin every 24 h. In conclusion, tigecycline plus aminoglycosides might be a potential regimen against CRKP and CoRKP. The appropriate combination regimen based on MIC-based dose adjustment can improve optimal antibiotic dosing. Further research via clinical studies will be necessary to confirm these results.


Introduction
Carbapenem-resistant Enterobacteriaceae (CRE), especially carbapenem-resistant Klebsiella pneumoniae (CRKP), is among the most present and dangerous pathogens prioritized by the World Health Organization (WHO) [1]. CRE mainly causes a variety of nosocomial infections, such as bloodstream infections, hospital acquired pneumonia (HAP), ventilatorassociated pneumonia (VAP), urinary tract infections, intra-abdominal infections, and skin and soft tissue infections [1,2]. Additionally, the emergence and spread of CRE leads to significant global health issues that cause high mortality and medical expenditure [1,3]. Based on data from the Centers for Disease Control and Prevention (CDC), the death rate among patients infected with CRE bloodstream infections can be up to 50% [4]. Moreover, patients infected with CRE had higher mortality rates than those infected with carbapenemsusceptible Enterobacteriaceae (CSE) [5]. The prevalence of CRE infection has continually trended upwards over the last decade throughout the USA, Europe, and Asia [6][7][8]. In Thailand, the National Antimicrobial Resistance Surveillance Thailand (NARST) reported that the prevalence of CRKP has risen from 1.1% to 13.2% from 2010 to 2020 [9].
Currently, the available antibiotic options are limited for combating CRE. Even the efficacy of the newer beta-lactam/beta-lactamase inhibitors such as ceftazidime/avibactam, meropenem/vaborbactam, and imipenem/relebactam relies on the types of carbapenemases produced by CRE isolates, which is variable in each region. Additionally, treatment of CRE infection with combination therapy is the preferred option, as opposed to monotherapy [10,11]. Combination therapy also improves outcomes by maximizing bacterial killing and preventing bacterial regrowth [12,13].
The current combination regimen consists of mainstay antibiotics, which may be ineffective against CRE isolates, whereas other antibiotics might be more efficacious [12]. The antibiotic regimens may be divided into carbapenem-based regimens (when meropenem MICs ≤ 8 µg/mL), colistin-based regimens, and tigecycline-based regimens, whereas aminoglycosides may be often used as adjunctive antibiotics [14]. However, if the CRKP are susceptible to aminoglycosides, they are considered as part of the antibiotic regimens combined with other active antibiotics.
Tigecycline-based regimens are one option for CRE treatment, especially for intraabdominal and skin/soft tissue infections with low risk of nephrotoxicity [2,11,15]. Currently, colistin-resistant Enterobacteriaceae (CoRE) have been reported in various regions worldwide, including Thailand [16][17][18]. Tigecycline presents an interesting choice of antibiotic against either CRE or CoRE infections. Due to its pharmacokinetic properties, the quite high volume of distribution results in tigecycline penetrates and distributed into tissues rather than the plasma. Therefore, high doses of tigecycline may increase the plasma concentrations, leading to adequate bactericidal activity against CRE isolates [2].
Amikacin and gentamicin are most frequently selected as the first adjunctive agents in combination regimens. Many previous studies revealed the excellent activity against CRE isolates [19][20][21]. In Thailand, K. pneumoniae isolates have been susceptible to amikacin and gentamicin at rates of 94.7% and 82.8%, respectively [22]. The combination of tigecycline with amikacin or gentamicin is one of the most efficacious combinations in the antibioticresistant era of CRKP and colistin-resistant K. pneumoniae (CoRKP). Although the previous studies described the synergistic activities of these combinations, they have not drawn clear conclusions because the variety of study methodologies, antimicrobial susceptibility profiles, and different carbapenemase types have led to inconsistent findings [14,15]. The Monte Carlo simulation is an advanced statistical technique that randomly chooses pharmacokinetic parameters and distributions, integrated with pharmacokinetic equations and pharmacokinetic/pharmacodynamic (PK/PD) indices, to estimate and predict the optimal dosage regimens [23].
In this study, we conducted in vitro experiments with combinations of tigecycline and amikacin or gentamicin to determine antimicrobial susceptibility and the synergistic activity of these antibiotics against CRKP and CoRKP clinical isolates. We also generated an appropriate dosage regimen using PK/PD targets for the treatment of critically ill patients infected with CRKP or CoRKP.

Bacterial Strains
A total of 49 non-duplicate CRKP strains were collected from patients admitted to the Pharmongkutklao Hospital, Bangkok, Thailand, during the period from June to November 2020. Various clinical specimens were obtained from different individual patients infected with CRKP. All studied strains were identified as K. pneumoniae by using Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS). After that, antimicrobial susceptibility testing was performed using the broth microdilution method (Sensititre, TREK Diagnostic Systems, Inc., Cleveland, OH, USA) and interpreted following Clinical and Laboratory Standards Institute (CLSI) 2020 for identification as CRKP [24].
After getting samples, all strains were isolated again on sheep blood agar (Clinical Diagnostics Ltd., Part, Bangkok, Thailand) and purified on MacConkey agar (Clinical Diagnostics Ltd., Part, Bangkok, Thailand). To identify K. pneumoniae, the morphology of colonies on MacConkey agar appeared large and mucoid-pink. Furthermore, meropenem minimum inhibitory concentrations (MICs) were determined by broth microdilution to confirm CRKP strains.
CRKP isolates were defined as non-susceptible to either ertapenem, imipenem, or meropenem according to the procedures described by the Clinical and Laboratory Standards Institute (CLSI) 2020 [24]. All CRKP strains were stored at −80 • C until analysis. Escherichia coli (E. coli) ATCC 25922 was used as a reference strain for quality control.
MIC50 is defined as the MIC value at which ≥50% of the isolates in a test population are inhibited. MIC90 is defined as the MIC value at which ≥90% of the isolates in a test population are inhibited.

Performing FICI
The synergistic effect of antibiotic combinations including tigecycline-amikacin and tigecycline-gentamicin was determined using the checkerboard technique. The studied antibiotic concentrations of a series of two-fold dilutions were 0.125-8 µg/mL for tigecycline, 0.0625-64 µg/mL for amikacin, and 0.0078-8 µg/mL for gentamicin.
Of the checkerboard, the MIC of either antibiotic combination or antibiotic alone is the lowest drug concentration that contains no growth (a clear on the well) [26,27]. The FICI was defined as the aggregate MIC ratio of each combination. It was calculated as follows: where FIC A = MIC of drug A in combination/MIC A alone; FIC B = MIC of drug B in combination/MIC B alone.
A FICI index specified the effects of a given combination, which were classified into 4 categories. Synergism, additive effects, indifference, and antagonism were indicated as an ∑FIC index of ≤0.5, an ∑FIC index of >0.5-1, an ∑FIC index of >1-4, and an ∑FIC index of >4, respectively [28].

Molecular Study of Antibiotic Resistance Genes
The carbapenemase types in all studied K. pneumoniae strains were evaluated using the polymerase chain reaction (PCR) method. Multiplex PCRs consisted of multiple primer sets of carbapenemase genes (bla IMP , bla VIM , bla OXA-48 , bla NDM, and bla KPC ); moreover, the mcr-1 primer also detected mcr-1 gene isoforms. The primer pairs of each gene are presented in Table 1. Table 1. Primer sets for detection genes of antibiotic resistance [29,30].

Type of Resistance Genes Primers
The multiplex PCR conditions for amplification were as follows: preincubation for 3 min at 94 • C, 35 cycles of 30 s at 94 • C, 35 s at 57 • C, and 45 s at 72 • C; and a final extension step for 5 min at 72 • C [31]. The mcr-1 PCR conditions were as follows: preincubation for 5 min at 94 • C, 35 cycles of 30 s at 94 • C, 35 s at 53 • C, and 45 s at 72 • C; and a final extension step for 5 min at 72 • C. The amplified products were analyzed by agarose (1%) gel electrophoresis in 0.5 × Tris/Borate/EDTA (TBE), after staining with ethidium bromide. The criterion for identifying the resistant genes was identical positions of bands between the PCR products and the PCR reference.

Antibiotic Dose Optimization by Monte Carlo Simulation
All pharmacokinetic parameters of critically ill patients were obtained from previous studies. A set of population pharmacokinetic parameters is shown in Supplementary Materials S1 (Tables S1-S3). The relationship between drug concentration levels and time was generated using a two-compartment model for tigecycline and amikacin and a onecompartment model for gentamicin with the linear pharmacokinetic behavior [32][33][34]. For amikacin and gentamicin, creatinine clearance (CrCL) was included to simulate the plasma aminoglycoside concentration-time using the equations containing CrCL as covariate factor (Supplementary Materials S1: Tables S1-S3).
The PD indies for tigecycline are characterized by the area under the curve (AUC) of the unbound tigecycline plasma concentration-time at 24 h to the MIC, whereas amikacin and gentamicin used the C max /MIC ratio. The effective PK/PD targets and indices of tigecycline and aminoglycosides were f AUC 0-24 /MIC ≥ 0.9 and C max /MIC > 8, indicative of good clinical outcomes [35,36]. The safe PK/PD targets of aminoglycosides were AUC 0-24 (mg × h/L) > 700 which occurred nephrotoxicity rate more than 10% [37]. The trapezoidal rule was used to calculate AUC. f AUC was calculated and divided by the given MIC to estimate the desired PK/PD value; the tigecycline unbound fraction of 80% was used.
The regimens of tigecycline ranged from a loading dose of 200-400 mg, followed by 100-200 mg intravenous injection every 12-24 h in general patients, whereas the regimens of amikacin and gentamicin ranged from a loading dose of 15-30 mg/kg, followed by 7.5-20 mg/kg and a loading dose of 3-8 mg/kg, followed by 2.5-7 mg/kg every 24-48 h, respectively. Different PK profiles were simulated only for aminoglycosides with patients having creatinine clearance of <10 mL/min, 10-25 mL/min, 26-50 mL/min, and 51-90 mL/min, respectively. The dosing regimens were simulated over a 24 h period. Details of the antibiotic dosing regimen are shown in Supplementary Materials S1 (Table S4).
The optimal antibiotic regimens were investigated using a 10,000-subject Monte Carlo simulation (Oracle Crystal Ball Classroom Faculty Edition-Oracle 1-Click Crystal Ball 201, Thailand).
MCS is a mathematical technique to simulate 10,000 virtual patients based on mean pharmacokinetic parameters (drug clearance or volume of distribution) and their distribution (standard derivation) in the compartment models (one or two compartments) for computing the plasma concentration and time among individual virtual patients. The Cmax and calculated AUC values from the trapezoidal rule among simulated patients and the probability of target attainment (PTA) of the target PK/PD index were determined.
The PTA was defined as the predicted probability that each antibiotic regimen would meet a PK-PD index and a PK-PD target at each MIC. Additionally, the cumulative fraction of response (CFR) was defined as the sum of the cumulative fraction of each regimen, calculated as the fraction of bacteria multiplied by the PTA of each antibiotic MIC. The optimal dosing regimens defined as the dosing regimens of each antibiotic MIC or the MIC values from the synergistic study reached greater than 90% of PTA for documented therapy and greater than 90% of CFR for empirical therapy. Each dosing regimen for tigecycline, amikacin, and gentamicin, we estimated the PTA to achieve MIC ranging from 0.0625 to 16 µg/mL in the target. CFR calculation was achieved using the MIC distributions.

Ethics Approval
Ethical approval for this study was obtained from the ethics review committee of the Royal Thai Army Medical Department and Phramongkutklao Hospital, Bangkok, Thailand (Ethics number: Q011h/63 issued on 13 July 2020).

PTA and CFR
For tigecycline dosage regimens to meet the PTA at f AUC 0-24 /MIC ≥ 0.9, once daily high-dose tigecycline (400 mg loading dose) was followed by 200 mg supplemental doses every 12 h, which achieved ≥90% PTA with a MIC of 2 µg/mL. Twice daily tigecycline doses achieved the PTA target with a MIC of 1 µg/mL, whereas all tigecycline dosage regimens reached the PTA target of a MIC of 0.5 µg/mL. The assessments of PTA and CFR for different tigecycline dosages are shown in Table 4. The aminoglycoside dosage regimens also reached the PTA target at C max/MIC > 8. At a MIC of 1 µg/mL, all amikacin dosage regimens met the PTA target in critically ill patients. At a MIC of 2 µg/mL, a loading dose of 25-30 mg/kg followed by a maintenance dose of 15-20 mg/kg of amikacin met the PTA target in patients with creatinine clearance levels of more than 50 mL/min. No regimens achieved the target at MIC50 and MIC90. For gentamicin, all dosage regimens met the PTA target in critically ill patients with an MIC of 0.25 (MIC50) and 0.5 µg/mL. At a MIC of 1 µg/mL, 4 mg/kg and 5 mg/kg once daily regimens (extended infusion) reached the PTA target in patients with creatinine clearance at 10-50 and 51-90 mL/min, respectively. No regimens achieved the target with a gentamicin MIC90 of 4 µg/mL, nor did any regimen of aminoglycosides achieve the CFR target when used as single agents, whereas all regimens of aminoglycosides met the target when combined with other antibiotic agents. The PTA and CFR values for different aminoglycoside regimens are evaluated in Tables 5 and 6.

Discussion
In these studies, we determined the microbial susceptibilities of CRKP isolates to tigecycline and aminoglycosides. In our findings (Table 2), one-fifth (20.4%) of CRKP isolates were only susceptible to tigecycline. Nonetheless, previous studies reported tigecycline susceptibility rates ranged from 12 to 100% [18,21,38,39] in Thailand, as well as 58.0-86.1% in other countries [19,40]. It is possible that there are differences in the cutoff tigecycline breakpoint (in a range of between 0.5 and 2 µg/mL). Furthermore, growing rates of CRKP and CoRKP in our setting may lead to increased use of tigecycline [9,18].
From Table 4, all high-dose tigecycline regimens reached the PTA target for isolates with an MIC of 0.5 µg/mL. At MIC = 1 µg/mL, a loading dose of 200 mg of tigecycline followed by a maintenance dose of 100 mg every 12 h (400 mg/day) was able to reach the PTA target; these results were consistent with those of previous studies [41,42]. At MIC = 2 µg/mL, our results showed that a loading dose of 400 mg tigecycline followed by a maintenance dose of 200 mg every 12 h (400 mg/day) also achieved the PTA targets. Based on tigecycline PK/PD, once daily high doses of tigecycline (loading dose of 400 mg followed by maintenance doses of 200 mg every 24 h) also met the PTA target with an MIC of 1 µg/mL. Thus, high-dose tigecycline regimens with once daily maintenance doses should be considered for the treatment of infections caused by CRKP or CoRKP, as these antibiotics have a concentration-dependent killing property and long half-life (~27-42 h) [43].
Most CRKP isolates were found to be susceptible to aminoglycosides (approximately 90%). These findings from Table 2 were consistent with previous findings from NARST. The susceptibility rates to amikacin and gentamicin ranged between 80% and 100% [18,21,22]. Furthermore, we observed that all isolates were only resistant to one of the two aminoglycosides tested. The occurrence of aminoglycoside-modifying enzymes (AMEs) may be one mechanism of resistance, being associated with no cross-resistance to all aminoglycosides [44]. High-dose regimens of aminoglycosides were simulated for critically ill patients in order to increase the probability of achieving target peak concentrations. Our findings (Tables 5 and 6) showed the regimens of amikacin and gentamicin met the PTA targets (at C max/ MIC = 8) with MICs of no more than 2 and 1 µg/mL, respectively. An earlier study showed aminoglycoside dosage regimens (ranged from 5 to 30 mg/kg) achieved the PTA target at MICs of 0.5 µg/mL with the corresponding target of C max /MIC ≥ 10 [33]. Although our study used the minimum PK/PD target of aminoglycosides when compared with the previous study, the dosage regimens did not achieve the targets within the susceptibility breakpoints. Thus, aminoglycosides should be used as the adjuncts for the treatment of infections caused by CRKP or CoRKP.
In prior studies, colistin resistance rates of CRKP isolates were only 25% [39], whereas colistin resistance rates of K. pneumoniae isolates ranged from 17.3% to 76.1% [18,[45][46][47]. Not surprisingly, high rates of colistin resistance were reported because our study only collected K. pneumoniae isolates, which had the highest rates of colistin resistance among Gram-negative bacteria [18].
Generally, colistin-based regimens were one of the active regimens for treatment if the infections were found to be susceptible to colistin [14,48]. Our results found that 80% of CRKP isolates had high MICs of colistin (MICs ≥ 16 µg/mL). A higher colistin MIC could lead to treatment failure and development of antibiotic resistance [49]. Furthermore, a PK/PD study recommended colistin MICs to achieve probability of target attainments (PTAs) in all patients with various creatinine clearances ranging from 0.5 to 16 µg/mL [46]. Hence, colistin combination regimens may not be recommended to treat patients in our setting.
Tigecycline-based regimens are alternative regimens for the treatment of multi-drug resistant bacteria with high MIC values. The results of checkerboard assays from Table 2 showed the sum of in vitro interaction of the combination of tigecycline with amikacin and gentamicin. These assays generated a potent effect greater than or equal to that of the individual antibiotics. Overall, four isolates showed synergism with both combinations, whereas the additive effects of tigecycline plus gentamicin and tigecycline plus amikacin were seen in 34 and 37 isolates, respectively (Table 3). Related studies also reported tigecycline plus amikacin or gentamicin was effective against K. pneumoniae isolates. For tigecycline plus amikacin, one study showed the synergistic effect was 36.4%, whereas another study found the synergistic effect was 70% [19,50]. For tigecycline plus gentamicin, one study showed a synergistic increase of 28.6%, whereas another study showed only 13.3% [19,21]. Notably, our results were not consistent with previous studies, likely because of the differences in tigecycline susceptibility and the type of multi-drug resistance genes present.
The synergistic activities of these antibiotics may be associated with co-resistance genes. Our study showed 8.2% synergy rates, whereas a previous study (Prawang et al.) collected the isolates in the same hospital during 2016 to 2018 and showed 13.3% synergy rates [21]. Although we conducted a study in the same setting, a difference in the percentage of synergy was found. The different results may be caused by the genetic background. Our isolates contained co-resistance of carbapenemase gene (bla OXA-48-like (53.1%) and bla OXA-48-like plus bla NDM (42.9%), whereas the isolates of a previous study contained only bla OXA-48-like (~90%). Notably, the CRKP isolates producing OXA-48-like carbapenemase showed more antibiotic synergy than the CRKP isolates producing NDM carbapenemase. Coproduction of NDM-1 and aminoglycoside resistance genes (e.g., 16S Ribosomal RNA Methyltransferases F: RmtF) may be associated with mobile genetic elements in CRKP isolates, leading to a high level of resistance to aminoglycosides [51]. Recent studies also showed synergy of tigecycline and aminoglycosides in KPC-2-producing CRKP isolates ranging from 36.4% to 70% [19,50]. Therefore, these combinations may be used as a first choice of treatment in many regions and countries, based on epidemic spreading characteristics.
The combinations of these drugs have an activity against CRKP isolates. When two antibiotics were taken to the bacteria cells, the synergism may occur because of two reasons. Firstly, tigecycline binds to the bacterial ribosome and inhibits protein synthesis. Secondly, aminoglycosides may disrupt and increase permeability of the bacterial outer membrane [52,53]. In a comparison of two aminoglycosides, amikacin can increase the permeability of cell membrane more than gentamicin [54]. Aminoglycosides lead to mistranslated amino acids, in turn leading to damage to the bacterial cytoplasmic membrane and interfering with protein synthesis; these protein errors also lead to reduced β-lactamase expression from aminoglycoside activities. Therefore, aminoglycosides likely enhance combined antibiotics to the target sites, contributing to synergistic effects [52,53]. Moreover, the combination of antibiotics could also decrease mutational frequencies by a range of 1000-10,000-fold at low concentration (1× MIC), suppressing the emergence of resistance genes [55].
The synergistic or additive effects lead to a reduction in MICs. Our findings from Table 2 showed MIC50, MIC90, and maximum MICs of tigecycline, amikacin, and gentamicin in combination regimens decreased by 1-to 4-fold when compared to single antibiotic regimens. Consequently, the MICs moved from resistant to susceptible, leading to achievement of PTA and CFR targets with low-dose regimens [56]. These beneficial impacts are associated with low rates of adverse drug reactions, particularly nephrotoxicity and ototoxicity from aminoglycosides, and lead to high rates of successful treatment [10,11,48]. It is reasonable to suggest the combination of tigecycline with amikacin or gentamicin for the treatment of severe infections caused by CRE, especially in the polymyxin-resistance era.
This study has several limitations. Firstly, our study might not be generalizable to other settings because the collection was made from a single university hospital. Secondly, our results were obtained from in vitro experiments using stable concentrations of antibiotics, and the therapeutic efficacy from patients' immune response was not taken into account. Third, we did not perform other mechanisms (e.g., efflux pumps and porin loss). Whole-genome sequencing should be investigated to obtain further information on the genetic variants associated with antibiotic resistance genes. Lastly, tigecycline with aminoglycosides demonstrated higher activities in vitro and may be useful for treatment of infections caused by CRE in hospitals. Further studies in clinical patient-related outcomes are required to confirm their clinical use.

Conclusions
Tigecycline plus amikacin or gentamicin showed synergistic or additive effects against CRKP and CoRKP isolates. The benefit of such combinations might improve the PTA and CFR achievement, and also reduce the risk of nephrotoxicity due to a lower dose of aminoglycosides. A large sample of isolates in vitro and from clinical patient-related outcomes should be assessed to confirm the combinatory activities of these antibiotics against CRKP and CoRKP isolates.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/antibiotics10060736/s1, Supplementary Materials S1: Table S1: A set of parameters of tigecycline; Table S2: A set of parameters and an equation of amikacin; Table S3: A set of parameters  and an equation of gentamicin; Table S4: Antibiotic dosing regimens for simulation; Supplementary Materials S2: MIC distribution of single and combined antibiotics.