Evaluation of the Bactericidal Activity of Fosfomycin in Combination with Selected Antimicrobial Comparison Agents Tested against Gram-Negative Bacterial Strains by Using Time-Kill Curves

The effects of combining fosfomycin with various antimicrobial agents were evaluated in vitro by broth microdilution checkerboard and time-kill kinetic studies. Checkerboard analyses were used to evaluate the following 30 Gram-negative isolates: 5 Pseudomonas aeruginosa, 5 Acinetobacter baumannii-Acinetobacter calcoaceticus species complex, and 20 Enterobacteriaceae isolates.

ABSTRACT

The effects of combining fosfomycin with various antimicrobial agents were evaluated in vitro by broth microdilution checkerboard and time-kill kinetic studies. Checkerboard analyses were used to evaluate the following 30 Gram-negative isolates: 5 Pseudomonas aeruginosa, 5 Acinetobacter baumannii-Acinetobacter calcoaceticus species complex, and 20 Enterobacteriaceae isolates. No isolate exhibited antagonism when fosfomycin was tested in combination, and synergy was observed in more than 25% of the drug combinations tested. The most frequent instances of synergy occurred when testing fosfomycin with β-lactams. Two isolates of Pseudomonas aeruginosa, 2 of Klebsiella pneumoniae, and 1 of the A. baumannii-A. calcoaceticus species complex that exhibited synergy when fosfomycin was tested in combination were subjected to time-kill kinetic analyses for confirmation. Time-kill assays confirmed synergistic activity. These data indicated that combination therapy with fosfomycin may be beneficial.

INTRODUCTION

Fosfomycin is a broad-spectrum antimicrobial agent that exhibits a unique mechanism of action against an enzyme target and inhibits an earlier step in bacterial cell wall synthesis compared to those inhibited by other antibacterial agents (1, 2). Fosfomycin covalently binds to MurA, preventing the first committed step in peptidoglycan biosynthesis (2).

The in vitro activity of fosfomycin has been shown against a wide range of Gram-positive and Gram-negative bacteria, including multidrug-resistant (MDR) organisms (1, 3–9). The approved dose for fosfomycin to treat uncomplicated urinary tract infections is a single oral dose of 3 g (10). Based on enhancing its pharmacodynamic characteristics, an intravenous dosage of 6 g every 8 h of ZTI-01 (fosfomycin for injection) is currently in clinical development for treating complicated urinary tract infections (https://clinicaltrials.gov/ct2/show/NCT02753946).

Multidrug-resistant organisms, including those from deep-seated infections, are often treated with combination chemotherapy. Fosfomycin’s differentiated mechanism of action has been proposed to enhance killing when it is combined with other antibiotic agents (6, 7). A variety of studies have evaluated the in vitro effect that occurs when combining fosfomycin with other antibacterial agents (11–19). For example, synergy has been shown to occur when testing combinations against enterococci and Staphylococcus aureus (12, 13). In vitro synergy was also shown when testing the combination of colistin and fosfomycin against KPC-producing Klebsiella pneumoniae and against NDM-1-producing Enterobacteriaceae (14, 15). Temocillin and fosfomycin, when tested against KPC-3- or OXA-48-producing Escherichia coli, exhibited synergistic effects (16). In contrast, 12 K. pneumoniae isolates from Turkey that were positive by PCR for OXA-48 demonstrated antagonism when colistin and fosfomycin were tested, while synergy was noted with the combination of fosfomycin with imipenem, meropenem, or tigecycline (17). Synergy has also been shown with various combinations of fosfomycin when testing Pseudomonas aeruginosa (18) and Acinetobacter baumannii (19). In this study, we evaluated the antimicrobial activity of fosfomycin when combined with selected antimicrobial agents and tested against current clinical bacterial strains from the United States by broth microdilution checkerboard testing and time-kill kinetic analysis.

RESULTS AND DISCUSSION

The reference agar dilution MIC method was not used in this study because using reference agar dilution in a checkerboard analysis for a large number of isolates and drugs is prohibitively difficult. For that practical reason, broth microdilution with glucose-6-phosphate (G-6-P) was chosen. In this study, quality control results for fosfomycin and comparator agents when tested against E. coli ATCC 25922 and P. aeruginosa ATCC 27853 were within the Clinical and Laboratory Standards Institute (CLSI)-established quality control ranges for reference method testing, even though all testing was done in broth microdilution with glucose-6-phosphate. This outcome indicated that the results for the comparator agents were not affected by the presence of G-6-P and that fosfomycin in broth, at least for the 2 quality control strains, performed adequately enough, as it remained in range. Also, broth microdilution with G-6-P and reference agar dilution with G-6-P have been shown to correlate reasonably well, but not at the required 90% to 95% of MIC values within ± one 2-fold dilution that is expected for comparable methods (20). For example, Flamm et al. showed that MIC values for 81.1% of 106 isolates were within ± one 2-fold dilution when tested in broth microdilution with G-6-P compared to those tested in agar dilution with G-6-P (21). The degree of comparability varied among species, and the differences tended to be one 2-fold dilution higher in broth, although there were examples of multiple fold increases. Ballestero-Tellez showed 86.4% categorical agreement when testing 81 E. coli isolates in broth microdilution with G-6-P compared to reference agar dilution and 51.1% categorical agreement for 139 K. pneumoniae isolates (22).

All isolates used in this study were recovered from clinical respiratory tract (n = 9), urinary tract (n = 7), skin and skin structure (n = 6), or bloodstream (n = 5) infections from U.S. hospitals in 2015; except for the 3 ATCC strains (E. coli ATCC 25922, P. aeruginosa ATCC 27853, and A. baumannii ATCC 19606). Within the recovered isolates, 7 MDR-phenotype strains included 2 extended-spectrum β-lactamase (ESBL)-producing and 2 KPC-producing Enterobacteriaceae isolates. Each strain is cataloged in the JMI Laboratories organism bank with a unique 6-digit identification collection number.

Checkerboard methodology was used to evaluate the following 30 Gram-negative bacteria: 5 Pseudomonas aeruginosa, 5 A. baumannii-A. calcoaceticus species complex, and 20 Enterobacteriaceae isolates.

Fosfomycin tested in checkerboard synergy combinations with 9 antimicrobials against P. aeruginosa showed synergy in 16% of the 45 total combinations. The most common synergistic combinations for fosfomycin when tested against P. aeruginosa were with ceftazidime (2/5 isolates) and minocycline (2/5; Table 1). A minimum fractional inhibitory concentration (FIC_min) of >0.5 to <1 (partial synergy) was observed in 24/45 (53%) combinations, including 5/5 isolates for piperacillin-tazobactam (Table 1). No antagonism was observed among any combinations tested. When fosfomycin antimicrobial combinations were tested against Acinetobacter baumannii-Acinetobacter calcoaceticus species complex isolates, a total of 30% of the 40 antimicrobial/isolate combinations tested showed synergy, and 33% showed partial synergy. Synergy was observed when testing fosfomycin in combination with piperacillin-tazobactam for 3/5 isolates and with amikacin, gentamicin, ceftazidime, and colistin for 2/5 isolates (Table 1). Average FIC_min values for piperacillin-tazobactam, meropenem, and levofloxacin for the 5 A. baumannii-A. calcoaceticus species complex isolates were 0.51, 0.61, and 0.78, respectively (data not shown). A total of 20 isolates of Enterobacteriaceae from 4 species groups were tested in checkerboard configurations with 10 antimicrobials. A total of 29% of combinations showed synergistic interactions with fosfomycin, and 36% showed partial synergy. The most frequent synergy interaction occurred with fosfomycin and piperacillin-tazobactam (12/20 isolates), followed by fosfomycin and meropenem or ceftazidime (8/20 isolates), and then fosfomycin and levofloxacin or gentamicin (6/20 isolates; Table 1). No antagonism was observed among combinations tested against Enterobacteriaceae.

TABLE 1.

Categorical interactions for antimicrobial agents tested in checkerboard combinations with fosfomycin for Gram-negative isolates

Antimicrobial agent	Interactions fora:
	P. aeruginosa						A. baumannii-calcoaceticus species complex						Enterobacteriaceae
	SYN	INDIF			ANTAG	INDET	SYN	INDIF			ANTAG	INDET	SYN	INDIF			ANTAG	INDET
		PS	ADD	INDIF				PS	ADD	INDIF				PS	ADD	INDIF
Amikacin		3	1	1			2	1	2		—	—	—	—	—	—	—	—
Gentamicin		1	3	1			2	3					6	11	2	1
Aztreonam	1	3				1	—	—	—	—	—	—	5	5	1			9
Ceftazidime	2	3					2	1		1		1	8	9	1			2
Piperacillin-tazobactam		5					3		1			1	12	5	2			1
Meropenem	1	2	1	1			1	2	1			1	8	9	1			2
Levofloxacin	1	3	1					3	1	1			6	8	4	2
Minocycline	2	3					—	—	—	—	—	—	4	7	6	1		2
Tigecycline	—	—	—	—	—	—		2	2	1			5	9	1	5
Colistin		1		4			2	1		1		1	1	5		8		6
Trimethoprim-sulfamethoxazole	—	—	—	—	—	—	—	—	—	—	—	—	2	4	2	7		5
All agents	7	24	6	7	0	1	12	13	7	4	0	4	57	72	20	24	0	27
% of total interactions	16	53	13	16	0	2	30	33	18	10	0	10	29	36	10	12	0	14

^aNumber of interactions unless otherwise indicated. SYN, synergy (<0.5); INDIF (>0.5 to ≤4) category broken down into PS, partial synergy (>0.5 to <1), ADD, additive (1), and INDIF, indifferent (>1 to ≤4); ANTAG, antagonism (>4); INDET, indeterminant, recorded when off-scale MIC test results occurred; —, not tested.

A total of 5 isolates in the checkerboard testing were selected based on their demonstration of synergy when testing fosfomycin in combination with at least 1 other agent. The isolates comprised 2 Pseudomonas aeruginosa, 1 Acinetobacter baumannii-A. calcoaceticus species complex (MDR phenotype), and 2 Klebsiella pneumoniae (1 carbapenem-resistant Enterobacteriaceae [CRE] phenotype and 1 ESBL phenotype). Fosfomycin was tested in combination with meropenem against 1 P. aeruginosa isolate and with ceftazidime against another P. aeruginosa isolate (Fig. 1A and B). By 8 h, fosfomycin at 4× MIC, when tested against P. aeruginosa (isolate 889839), reduced the bacterial count (relative to that of the starting inoculum) by 1.8 log₁₀ CFU/ml, followed by a rebound in growth at 24 h (a 3.1-log₁₀ CFU/ml increase from starting inoculum concentration) (Fig. 1A). The combination of fosfomycin and meropenem reduced bacterial count by >2 log₁₀ CFU/ml compared to that of the starting inoculum and was >2 log₁₀ CFU/ml lower than fosfomycin alone at 24 h. Against P. aeruginosa (isolate 893949), fosfomycin at 4× MIC reduced the bacterial count (relative to that of the starting inoculum) by 1.8 log₁₀ CFU/ml by 8 h, followed by a rebound in growth at 24 h (a 2.7-log₁₀ CFU/ml increase from starting inoculum concentration) (Fig. 1B). The combination of fosfomycin at 4× MIC and ceftazidime at 1× MIC reduced the bacterial count by >2 log₁₀ CFU/ml compared to that of the starting inoculum and was >2 log₁₀ CFU/ml lower than that with fosfomycin alone at 24 h.

FIG 1.

Time-kill curves with fosfomycin and comparator agents alone and in combination: (A) time-kill curve for P. aeruginosa isolate 889839 for fosfomycin (with 25 mg/liter glucose-6-phosphate) at 4× MIC both alone and combined with meropenem at 0.25× and 1× MIC; (B) time-kill curve for P. aeruginosa isolate 893949 for fosfomycin (with 25 mg/liter glucose-6-phosphate) at 4× MIC both alone and combined with ceftazidime at 0.25× and 1× MIC; (C) time-kill curve for A. baumannii-A. calcoaceticus species complex isolate 920549 (MDR phenotype) for fosfomycin (with 25 mg/liter glucose-6-phosphate) at 2× MIC both alone and combined with meropenem at 0.25× and 1× MIC; (D) time-kill curve for K. pneumoniae isolate 875100 (CRE phenotype) for fosfomycin (with 25 mg/liter glucose-6-phosphate) at 2× MIC both alone and combined with piperacillin-tazobactam at 0.25× and 1× MIC; and (E) time-kill curve for K. pneumoniae isolate 885542 (ESBL phenotype) for fosfomycin (with 25 mg/liter glucose-6-phosphate) at 4× MIC both alone and combined with ceftazidime at 0.25× and 1× MIC.

By 8 h, fosfomycin at 2× MIC tested against A. baumannii-A. calcoaceticus species complex (isolate 920549; MDR phenotype) reduced the bacterial count (relative to that of the starting inoculum) by 2.5 log₁₀ CFU/ml, followed by an increase in growth of 1.9 log₁₀ CFU/ml by 24 h (0.6 log₁₀ CFU/ml lower than that of the starting inoculum, Fig. 1C). The combination of fosfomycin at 2× MIC and meropenem (1× MIC) by 8 h exhibited a 4.6 log₁₀ CFU/ml decrease in growth compared to the starting inoculum, and by 24 h, growth for the combination decreased to below detectable levels.

Synergy occurred when testing fosfomycin combined with piperacillin-tazobactam or ceftazidime against 2 Klebsiella pneumoniae isolates (Fig. 1D and E). When fosfomycin was tested against K. pneumoniae (isolate 875100; CRE phenotype), at 4 h the viable count was reduced by >3 log₁₀ CFU/ml. Growth rebounded with a 1.1 log₁₀ CFU/ml increase in growth from the 4-h time point by 8 h and with a 2.3 log₁₀ CFU/ml increase in growth compared to the starting inoculum at 24 h (Fig. 1D). By 8 h, the combination of fosfomycin (2×) and piperacillin-tazobactam (0.25× or 1× MIC) showed a reduced viable count (≥4 log₁₀ CFU/ml reduction) compared to that of the starting inoculum, which remained at least >2 log₁₀ CFU/ml below the level of the starting inoculum at 24 h (Fig. 1D). Against K. pneumoniae (isolate 885542; ESBL phenotype), fosfomycin at 4× MIC showed a decrease in growth at 4 h (a 2.3-log₁₀ CFU/ml decrease) followed by a rebound in growth at 8 h and 24 h (a 2.8-log₁₀ CFU/ml increase from initial inoculum) (Fig. 1E). In combination with 0.25× and 1× MIC ceftazidime, growth decreased by 3.9 to 4.2 log₁₀ CFU/ml at 8 h and decreased to below detectable levels at 24 h (ceftazidime, 1× MIC). Ceftazidime alone at 0.25× and 1× MIC showed increased growth at 24 h (a 2.2- to 2.5-log₁₀ CFU/ml increase from initial inoculum).

Overall, a high percentage of synergistic or partially synergistic/additive combinations were observed when fosfomycin was tested in combination with a variety of antimicrobials against Gram-negative bacteria. This synergy (bactericidal activity) occurred in combination with various β-lactam antibiotics (ceftazidime, meropenem, and piperacillin-tazobactam) when tested against K. pneumoniae (including CRE and ESBL phenotypes), P. aeruginosa, and A. baumannii-A. calcoaceticus species complex (including MDR isolates). These in vitro results demonstrate the potential beneficial effects of using fosfomycin in combination with other classes of antimicrobial agents against current MDR organisms of concern.

Despite a number of published studies having demonstrated in vitro synergy, limited information is available about how fosfomycin tested in drug combinations translates to in vivo synergy. (11–19, 23). For example, in 94 patients infected with carbapenem-resistant Acinetobacter strains, those who received colistin plus fosfomycin had a more favorable microbiology response and a trend toward more favorable outcomes than those receiving colistin alone (24). The data were less compelling in a mouse thigh model study of combinations of colistin and various agents, including fosfomycin (25). In that mouse study, rifampin and fusidic acid showed synergy with colistin, whereas fosfomycin did not show obvious synergy with colistin. However, with fosfomycin monotherapy, some strains showed a statistical difference from the untreated control (4 of 12 strains at 48 h showed a reduction of 0.73 to 1.06 log₁₀ recovered CFU). Fosfomycin combined with colistin demonstrated a statistical difference versus colistin alone in 8 of 12 strains (a reduction of 0.34 to 0.75 log₁₀ recovered CFU). Further investigation is needed to clarify the relationship of the in vitro synergies with in vivo outcomes.

MATERIALS AND METHODS

Checkerboard broth microdilution.

Susceptibility testing for fosfomycin and combination agents was performed using broth microdilution checkerboard synergy methods as described in the Clinical Microbiology Procedure Handbook, 4th edition (26). CLSI recommends that susceptibility testing for fosfomycin be done by the agar dilution method, and it recommends against broth dilution testing for this agent (27, 28). However, the most practical way to conduct checkerboard and time-kill studies is by broth microdilution; thus, broth methods were used in this study. Cation-adjusted Mueller-Hinton broth was supplemented with 25 mg/liter glucose-6-phosphate for testing fosfomycin alone and in combinations (27, 28). Checkerboard synergy panels were produced by JMI Laboratories (North Liberty, IA, USA) and stored at less than −70°C until use. Each panel design contained fosfomycin alone (MIC ranges: 4,096 to 8, 256 to 0.5, or 64 to 0.12 mg/liter) and combined with the following antimicrobial agents: amikacin (128 to 2 or 16 to 0.25 mg/liter), aztreonam (128 to 2, 32 to 0.5, or 2 to 0.03 mg/liter), ceftazidime (128 to 2, 16 to 0.25, or 1 to 0.015 mg/liter), colistin (512 to 8, 8 to 0.12, or 2 to 0.03 mg/liter), gentamicin (256 to 4 or 4 to 0.06 mg/liter), levofloxacin (256 to 4, 8 to 0.12, or 0.5 to 0.008 mg/liter), meropenem (2,048 to 32, 512 to 8, 32 to 0.5, 2 to 0.03, or 0.06 to 0.001 mg/liter), minocycline (64 to 1, 2 to 0.03, or 0.12 to 0.002 mg/liter), piperacillin-tazobactam (fixed 4 mg/liter tazobactam: 4,096 to 64, 1,024 to 16, 32 to 0.5, or 8 to 0.12 mg/liter), tigecycline (8- to 0.12 or 1 to 0.015 mg/liter), trimethoprim-sulfamethoxazole (1:19 ratio: 64 to 1 or 1 to 0.015 mg/liter). Antimicrobial powders were acquired from various sources; ceftazidime, colistin, levofloxacin, minocycline, piperacillin, sulfamethoxazole, tigecycline and trimethoprim were from Sigma-Aldrich (St. Louis, MO, USA); aztreonam, gentamicin, meropenem, and tazobactam were from USP Pharmacopeia (North Bethesda, MD, USA); amikacin was from Ercros (Aranjuez, Spain); and fosfomycin was from Zavante Therapeutics (San Diego, CA, USA; now part of Nabriva Therapeutics, King of Prussia, PA, USA).

Quality control for fosfomycin and all comparator agents for all panel designs was ensured by testing E. coli ATCC 25922 and P. aeruginosa ATCC 27853. CLSI-established quality control ranges were applied. Quality control organisms were used to test all original plate designs that were produced before they were frozen and then tested in parallel each day that checkerboard testing was performed to ensure run compliance.

The ∑FIC value for each well adjacent to growth in the checkerboard synergy panel was calculated and used to identify the ∑FIC minimum, ∑FIC maximum, and ∑FIC mean for each antimicrobial combination (26, 29). Even if synergism or antagonism occurred at only 1 FIC within the combination, that occurrence was used to categorize results in Table 1 (26). The FIC index categorical interpretations were defined as synergy when the value was ≤0.5, defined as indifferent when the FIC index was >0.5 to ≤4.0, and defined as antagonistic when the FIC index was >4.0 (26, 29, 30). In addition to the categorical interpretations, we provided alternate criteria for the indifferent category, although the inherent variability of the 2-fold dilution scheme makes it unreliable to interpret any individual result in the indifferent category when many isolates are tested. We felt that any trend that occurred based on multiple results would most likely be suggestive of a real mechanistic response and not just due to random method variation. The breakdown of results in the indifferent category is clearly identified in Table 1. The alternative criteria applied were partial synergy when the FIC index was >0.5 to <1, additive when the FIC index was 1, and indifferent when the FIC index was >1 to ≤4.0 (31–33). Some combination interactions could not be determined due to the presence of off-scale MIC test results, and a categorical result of indeterminate was recorded in these cases.

Time-kill kinetics.

Time-kill kinetics were performed as described in the Clinical Microbiology Procedure Handbook, 4th edition (26). Broth used for the time-kill kinetic studies to test fosfomycin and the various combinations was cation-adjusted Mueller-Hinton broth supplemented with 25 mg/liter glucose-6-phosphate. The starting inoculum for time-kill kinetics was targeted at 1 × 10⁶ CFU/ml. Time-kill concentration tubes were sampled for fosfomycin and comparators at time 0 hours (T0), T2, T4, T8, and T24. The sample volume was 0.1 ml that was plated directly to tryptic-soy agar with 5% sheep blood and serially diluted up to 8 times in tubes with 0.9 ml of a 0.85% saline solution before plating to additional agar plates. Dilution agar plate series were incubated for 24 to 48 h at 35°C in a CO₂ environment before quantifying the viable cell count for each tube at the specified times. Bacterial cell counts were plotted over time to create the time-kill curves. The limit of detection was 10 CFU/ml. Synergy was defined as a ≥2-log₁₀ decrease in CFU/ml between the combination and its most active constituent after 24 h, and the number of surviving organisms in the presence of the drug combination was ≥2 log₁₀ CFU/ml below that of the starting inoculum (26, 29, 30).