In Vitro Activities of Colistin and Sitafloxacin Combinations against Multidrug-, Carbapenem-, and Colistin-Resistant Acinetobacter baumannii Using the Broth Microdilution Checkerboard and Time-Kill Methods

Drug-resistant Acinetobacter baumannii (A. baumannii) infections are a critical global problem, with limited treatment choices. This study aims to determine the in vitro activities of colistin–sitafloxacin combinations against multidrug-, carbapenem- and colistin-resistant A. baumannii (MDR-AB, CRAB, CoR-AB, respectively) clinical isolates from tertiary care hospitals. We used the broth microdilution checkerboard and time-kill methods in this study. Synergy was found using both methods. The colistin–sitafloxacin combination showed synergy in MDR-AB, CRAB, and CoR-AB isolates (3.4%, 3.1%, and 20.9%, respectively). No antagonism was found in any type of drug-resistant isolate. The majority of CoR-AB isolates became susceptible to colistin (95.4%). The time-kill method also showed that this combination could suppress regrowth back to the initial inocula of all representative isolates. Our results demonstrated that the colistin–sitafloxacin combination might be an interesting option for the treatment of drug-resistant A. baumannii. However, further in vivo and clinical studies are required.


Introduction
Acinetobacter baumannii (A. baumannii) is one of the most common causes of nosocomial pneumonia in Asia [1]. Currently, several types of resistance in Acinetobacter spp. are critical problems worldwide, including MDR-AB (multidrug-resistant A. baumannii), CRAB (carbapenem-resistant A. baumannii), and CoR-AB (colistin-resistant A. baumannii), which result in long hospitalizations and high mortality and morbidity rates [2,3]. This Gram-negative bacterium has various mechanisms of antibiotic resistance. For example, the mechanism behind its resistance to beta-lactams, including carbapenems, includes the production of beta-lactamases or changes in the outer membrane proteins and multidrug efflux pumps [4]. The mechanism of resistance to fluoroquinolones involves target mutation and efflux pump over-activity [4]. For aminoglycosides, drug-inactivating enzymes, target mutation, and efflux pump over-activity are the mechanisms of resistance [4]. Multidrug-resistant A. baumannii involves many complex mechanisms and more than 40 different genes [5].
The minimum inhibitory concentration (MIC) values of all isolates were tested by the broth microdilution method. The MIC range of colistin was 0.5-16 milligrams per liter (mg/L) in the MDR-AB and CRAB isolates and 4-16 mg/L in the CoR-AB isolates. The MIC range of sitafloxacin of the MDR-AB, CRAB, and CoR-AB isolates was 0.02-8 mg/L. We also determined the MIC of the 50th and 90th percentiles (MIC 50/90 ) of each type of isolate. The MIC 50/90 of colistin was 2/4 mg/L in the MDR-AB and CRAB isolates, and 8/8 mg/L in the CoR-AB isolates. The MIC 50/90 of sitafloxacin was 1/2 in the MDR-AB and CRAB isolates and 0.5/1 in the CoR-AB isolates. The percentages of susceptibility of colistin and sitafloxacin were calculated from the following formula: (the level of susceptibility of the isolate to each agent/the level of all isolates in each type) × 100. More than 80% were colistin-susceptible isolates among the MDR-AB and CRAB isolates, and more than 95% were sitafloxacin-susceptible isolates among all types. The MIC range, MIC 50/90 , and the colistin and sitafloxacin percentage susceptibility of each type of isolate are shown in Table 1. After combination, the MIC values of colistin and sitafloxacin were reduced. The MIC range of colistin was 0.06-8 mg/L in the MDR-AB and CRAB isolates and 0.05-8 mg/L in the CoR-AB isolates. The MIC range of sitafloxacin in the MDR-AB and CRAB isolates was 0.01-4 mg/L and 0.02-2 mg/L in the CoR-AB isolates. The MIC 50/90 values were also reduced. The MIC 50/90 of colistin was 0.5/1 mg/L in the MDR-AB and CRAB isolates and 1/2 mg/L in the CoR-AB isolates. The MIC 50/90 of sitafloxacin was 0.5/1 in the MDR-AB and CRAB isolates and 0.25/1 in the CoR-AB isolates. Moreover, some colistinor sitafloxacin-resistant isolates became susceptible isolate. Among the CoR-AB isolates, 95.4% were susceptible to colistin. The results are shown in Table 1. The fractional inhibitory concentration index (FICI) values that were used to determine the synergistic effect in this study were calculated from the following formula: FICI = (MIC of colistin in combination/MIC of colistin alone) + (MIC of sitafloxacin in combination/MIC of sitafloxacin alone). The interpretation of the results was as follows: synergy, if FICI ≤ 0.5; no interaction, if FICI >0.5 to 4; and antagonism, if FICI >4. The colistin-sitafloxacin combination showed synergy values of 3.4% and 3.1% for the MDR-AB and CRAB isolates, respectively. For CoR-AB isolates, the combination showed a synergy of 20.9%. No antagonism was found. The distribution of the FICI values is shown in Table 2.

Time-Kill Method
The two representatives of MDR-AB with CRAB isolates (H25, K21) were tested using the time-kill method. The first resisted sitafloxacin, whereas the second did not. No CoR-AB isolates were tested by this method. For the H25 isolate (sitafloxacin-resistant), the MICs of colistin and sitafloxacin were 1 and 8 mg/L, respectively. For the K21 isolate (sitafloxacin-susceptible), the MICs of both colistin and sitafloxacin were 2 mg/L. We aimed to find the differences between sitafloxacin-resistant and susceptible isolates with time-kill methods. Bactericidal activity was found in both isolates at 4 h when 2× the MIC of colistin was tested. In addition, 1× the MIC of sitafloxacin showed bactericidal activity in both isolates at 12 h (H25) and 4 h (K21). Colistin showed rapid bacterial regrowth at all tested concentrations.
Colistin-sitafloxacin combinations showed a bactericidal effect against both isolates at all tested concentrations. Bactericidal activity was found at 4 h. Synergy also occurred at all tested concentrations, except 1× the MIC of colistin and sitafloxacin. The lowest concentration that showed synergy was 0.25× the MIC of the colistin-sitafloxacin combination. The results of the time-kill method are shown in Figures 1 and 2.
Colistin-sitafloxacin combinations showed a bactericidal effect against both isolates at all tested concentrations. Bactericidal activity was found at 4 h. Synergy also occurred at all tested concentrations, except 1× the MIC of colistin and sitafloxacin. The lowest concentration that showed synergy was 0.25× the MIC of the colistin-sitafloxacin combination. The results of the time-kill method are shown in Figures 1 and 2.

Discussion
MDR-AB and CRAB are worldwide problems that negatively impact clinical outcomes. This is the first study of the MIC distribution of A. baumannii from tertiary care hospitals in Thailand. Our study showed MDR-AB at 87.7%, CRAB at 86%, and CoR-AB at 14.3%. Interestingly, 96.2% of MDR-

Discussion
MDR-AB and CRAB are worldwide problems that negatively impact clinical outcomes. This is the first study of the MIC distribution of A. baumannii from tertiary care hospitals in Thailand. Our study showed MDR-AB at 87.7%, CRAB at 86%, and CoR-AB at 14.3%. Interestingly, 96.2% of MDR-AB isolates were also CRAB.
The lack of new antimicrobial agents for use against drug-resistant organisms can lead to inadequate therapy and limited treatment options [27]. Therefore, combination regimens of antibiotics might be considered. Colistin is commonly the last resort, both singly and in combination, for the treatment of resistant Gram-negative bacteria. There have been several studies on colistin-based therapy against A. baumannii which showed strong and synergistic effect. Sitafloxacin has been reported to show synergy against XDR-AB when combined with rifampin, sulbactam, and colistin [24,26]. There is no prior report on the synergy of the colistin-sitafloxacin combination against MDR-AB, CRAB, and CoR-AB. Therefore, this study focused on assessing its potential synergistic activity.
The synergy of the MDR-AB and CRAB isolates with the broth microdilution checkerboard method were 3.4% and 3.1%, respectively ( Table 2). A prior in vitro study on the use of colistin-sitafloxacin against XDR-A. baumannii by checkerboard method was carried out by Dong et al. [24]. They found a synergy of 12.5%, which is higher than our results. Because our isolates were susceptible to both colistin (86.7% and 87.2% for MDR-AB and CRAB, respectively) and sitafloxacin (96.6% and 99.5% for MDR-AB and CRAB, respectively), to a greater degree than the isolates used by Dong et al. (62.5% for colistin and 91.7% for sitafloxacin), our synergy was lower than that found by Dong et al. A study by Guelfi et al. [28], which tested polymyxin B combined with gatifloxacin against A. baumannii isolates, did not find any synergy. The study did not include polymyxin B-resistant isolates. Therefore, the synergy of the study could be less than in our study. We observed a reduction in the MIC values of each agent after combination (Table 1). Although the amount of MIC reduction was insufficient to show synergy in some cases, drug-resistant strains became susceptible. Our study found that more than 95% of the isolates were susceptible to sitafloxacin or colistin when tested in combination. Our study found more isolates susceptible to sitafloxacin when tested before combination compared with Xu et al. (96% and 92%, respectively) [24], who tested the sitafloxacin-sulbactam combination against XDR-AB. On the other hand, the sitafloxacin-sulbactam combination showed synergy that was higher than ours (3.4% and 16%, respectively). It was to be expected that most of our isolates that were susceptible to sitafloxacin did not affect the synergy when combined with colistin. For CoR-AB isolates, the synergy was 20.9%. Remarkably, we found that most of the CoR-AB isolates (95.4%) became colistin-susceptible isolates after testing in combination. These results showed that colistin enhanced the bactericidal activity of sitafloxacin, and vice versa. The proposed mechanism by which colistin enhanced the activity of sitafloxacin is the disruptive effect on membrane integrity, which makes the isolates more susceptible to a hydrophobic agent [29]. The logP (the partition coefficient), a particular ratio of the concentrations of a solute between a biphasic system of lipids and water, of sitafloxacin is 3.4 [30]. A positive value for logP denotes a higher concentration in the lipid phase [31]. Therefore, the logP value of sitafloxacin indicates that sitafloxacin is hydrophobic. This is the probable mechanism by which colistin enhances sitafloxacin activity. However, the mechanism by which sitafloxacin enhances the activity of colistin has not been elucidated.
The results from the time-kill method also showed synergy in representative isolates. Moreover, the results showed that colistin accelerated the bactericidal activity of sitafloxacin in sitafloxacin-resistant strains. The mechanism for this has not been fully clarified. Regarding the bactericidal effect of colistin, colistin alone showed rapid regrowth of inocula even though they were susceptible to colistin (Figure 1). We found that the combination could maintain the bactericidal effect of colistin for 24 h or could slow the regrowth so that inocula did not reach initial levels ( Figure 2). This result showed that sitafloxacin supported colistin to maintain bactericidal activity. This was similar to Safarika et al.'s [22] study of the colistin-levofloxacin combination against MDR-AB using the time-kill method, whose combination suppressed the regrowth of inocula. The mechanism was unclear. A study Antibiotics 2020, 9, 516 7 of 10 conducted by Zhanel et al. [32] demonstrated a reduction of the MPC/MIC ratio (mutant prevention concentrations/minimal inhibitory concentration ratio) of colistin when combined with levofloxacin against Pseudomonas aeruginosa. This might explain our result, in which fluoroquinolones helped to suppress the mutation of the isolates.
The first limitation of this study was the lack of genotypic data of the isolates, which might have predicted the mechanism of resistance of the isolates to sitafloxacin and colistin. This could have led us to precisely propose the mechanism of the synergistic effect and to specify the type of drug-resistance that is targeted by this combination. Another limitation of this study was that we did not test the synergistic effect by the time-kill method using CoR-AB isolates. Our results indicated that synergism was rarely observed in MDR-AB and CRAB isolates but was found for CoR-AB isolates. The drug-resistant isolates became susceptible, especially CoR-AB isolates. Further studies are needed to determine if such observations are correlated with improved clinical outcomes.

Clinical Isolates
A total of 412 clinical isolates of A. baumannii were collected from patients who attended 13 tertiary care hospitals in Thailand from 2016 to 2017. The isolates were collected from non-duplicated patients. Three hundred clinical isolates of A. baumannii were randomly selected by counting from the first to the twenty-fifth isolate from each hospital. Regardless of drug-resistant isolates, the selected isolates included MDR-AB, CRAB, and CoR-AB isolates. MDR-AB was defined as A. baumannii non-susceptible to at least one agent in ≥3 antimicrobial categories, as follows; aminoglycosides, antipseudomonal fluoroquinolones, antipseudomonal carbapenems, antipseudomonal penicillin plus beta-lactamase inhibitors, folate pathway inhibitors, penicillin plus beta-lactamase inhibitors, extended-spectrum cephalosporins, polymyxins, and tetracyclines [33]. CRAB was defined as A. baumannii resistant to imipenem, doripenem, or meropenem, following the Clinical and Laboratory Standard Institute (CLSI) 2016 criteria for screening carbapenem-resistant strains. This study mentioned MDR in relation to CRAB isolates, meaning the MDR definition, which includes carbapenem resistance. CoR-AB was defined as A. baumannii resistant to colistin conforming to CLSI 2016 (the resistant breakpoint was ≥ 4 mg/L) [34]. Genotypic testing was not performed in this study.
This study was approved by The Ethics Committee of Faculty of Dentistry/Faculty of Pharmacy, Mahidol University (COA.No.MU-DT/PY-IRB 2017/040.2607).

Antimicrobial Agents
Colistin sulfate standard powder was purchased from Tokyo Chemical Industry, Japan. Sitafloxacin standard powder was provided by Daichi Sankyo, Thailand.

Broth Microdilution Checkerboard Method
Susceptibility and MIC testing of single agents were carried out by the broth microdilution method, according to CLSI 2016. Both agents were tested and ranged from 0.015625 to 128 mg/L. The triplicate susceptibility test was incubated at 37 • C for 18 h. Escherichia coli ATCC25922 was used as a quality control isolate. According to CLSI 2016, susceptible and resistant breakpoints for colistin against Acinetobacter spp. are ≤ 2 mg/L and ≥ 4 mg/L, respectively. Breakpoints for sitafloxacin against Acinetobacter spp. were unavailable from CLSI 2016. Isolates with MICs of sitafloxacin ≤ 2 mg/L were tentatively considered to be susceptible to sitafloxacin, based on molecular surveillance with automated ribotyping [35].
After MIC determination of single agents for each isolate, the antibiotics for testing were prepared as 8×, 4×, 2×, 1×, 0.5×, 0.25×, and 0.125× of the MIC of each agent and the growth control.
Antibiotics 2020, 9, 516 8 of 10 A checkerboard microdilution method was triply tested [36]. The results of the synergistic effect were determined after incubating plates at 37 • C for 18 h. The fraction inhibitory concentration index (FICI) was calculated from the sum of the fractional inhibitory concentrations according to the following formula: FICI = (MIC of colistin in combination/MIC of colistin alone) + (MIC of sitafloxacin in combination/MIC of sitafloxacin alone). The interpretation of the results was as follows: synergy, if FICI ≤ 0.5; no interaction, if FICI > 0.5 to 4; and antagonism, if FICI > 4 [37].

Time-Kill Method
The time-kill method was used for two representative isolates with different susceptibilities to single agents. The MICs of each isolate, which were used for testing, were determined by the broth microdilution method. Colistin and sitafloxacin were tested singly and in combination and were incubated for 0, 4, 8, 12, and 24 h at 37 • C. The concentrations of colistin were 0.25×, 0.5×, 1×, and 2× MIC, whereas for sitafloxacin, the concentrations were 0.25×, 0.5×, and 1× MIC. We tested sitafloxacin at 1× MIC, as referred to in the Xu et al. study [26]. The time-kill study showed that sitafloxacin had good activity against the isolates at the concentration of 1× MIC, regardless of the susceptibility status of the isolates. The tested concentrations of this combination were concentrations of 0.25× MIC of colistin with 0.25× and 0.5× MIC of sitafloxacin, 0.5× MIC of colistin with 0.25× and 0.5× MIC of sitafloxacin, and 1× MIC of colistin with 1× MIC of sitafloxacin. Bactericidal activity was assessed as a ≥ 3 log10 reduction in colony forming unit (cfu)/mL over the time period measured, while synergy was a ≥ 2 log10 reduction in cfu/mL for the antibiotic combination in comparison with its more active constituent [38].

Conclusions
In conclusion, the combination of colistin and sitafloxacin is an interesting alternative option for the treatment of A. baumannii infections, including MDR-AB, CRAB, and CoR-AB infections. However, the results from this study should be further tested, which might lead to clinical studies.