Trends of Antimicrobial Resistance and Combination Susceptibility Testing of Multidrug-Resistant Pseudomonas aeruginosa Isolates from Cystic Fibrosis Patients: a 10-Year Update

ABSTRACT

Antimicrobial combination therapy is a time/resource-intensive procedure commonly employed in the treatment of cystic fibrosis (CF) pulmonary exacerbations caused by Pseudomonas aeruginosa. Ten years ago, the most promising antimicrobial combinations were proposed, but there has since been the introduction of new β-lactam plus β-lactamase inhibitor antimicrobial combinations. The aims of this study were to (i) compare in vitro activity of these new antimicrobials with other antipseudomonal agents and suggest their most synergistic antimicrobial combinations and (ii) determine antimicrobial resistance rates and study inherent trends of antimicrobials over 10 years. A total of 721 multidrug-resistant P. aeruginosa isolates from 183 patients were collated over the study period. Antimicrobial susceptibility and combination testing were carried out using the Etest method. The results were further assessed using the fractional inhibitory concentration index (FICI) and the susceptible breakpoint index (SBPI). Resistance to almost all antimicrobial agents maintained a similar level during the studied period. Colistin (P < 0.001) and tobramycin (P = 0.001) were the only antimicrobials with significant increasing isolate susceptibility, while an increasing resistance trend was observed for levofloxacin. The most active antimicrobials were colistin, ceftolozane-tazobactam, ceftazidime-avibactam, and gentamicin. All combinations with β-lactam plus β-lactamase inhibitors produced some synergistic results. Ciprofloxacin plus ceftolozane-tazobactam (40%) and amikacin plus ceftazidime (36.7%) were the most synergistic combinations, while colistin combinations gave the best median SBPI (50.11). This study suggests that effective fluoroquinolone stewardship should be employed for CF patients. It also presents in vitro data to support the efficacy of novel combinations for use in the treatment of chronic P. aeruginosa infections.

INTRODUCTION

In cystic fibrosis (CF) patients, Pseudomonas aeruginosa is the most commonly isolated pathogen, and more than 70% of CF patients are colonized with this bacterium by the age of 25 (1, 2). P. aeruginosa is the primary cause of acute respiratory exacerbations in CF patients, with persistent infections leading to a progressive decline in pulmonary function (3). It has been established that the presence of P. aeruginosa in respiratory cultures is a major predictor of mortality and morbidity (2). Therefore, in clinical practice, to improve life expectancy and the quality of life, especially for patients awaiting lung transplantation, aggressive antimicrobial treatment is employed (1–3). However, the cumulative lifetime treatment of CF patients with antibiotics leads to the development of multidrug-resistant (MDR) P. aeruginosa (4). For this reason, various treatment approaches are employed in patient management to delay the development of multidrug-resistant strains. These approaches include combination therapy and the use of modified dosing strategies to optimize antimicrobial pharmacokinetic/pharmacodynamics (PK/PD) parameters. To serve as a guide, 10 years ago, our lab published that the most promising in vitro antimicrobial combinations for use in the treatment of MDR P. aeruginosa infections were based on amikacin and ceftazidime combinations (5). However, recently, there has been the development of novel antipseudomonal agents, such as ceftolozane-tazobactam (C/T) and ceftazidime-avibactam (4). As single agents, ceftolozane and ceftazidime have been reported as the most active antipseudomonal agents. However, coupling these antimicrobials with tazobactam and avibactam extends the susceptibility pattern of these antimicrobials to include the extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae (4). In vitro investigations reported in British Society for Antimicrobial Chemotherapy (BSAC) data (UK report) state that ceftolozane-tazobactam is a potent antipseudomonal antibiotic with higher susceptibility rates than other β-lactam/β-lactamase inhibitor combinations, carbapenems, and fluoroquinolones. Susceptibility rates have been consistently high over the 9 years analyzed (2010 to 2018), with 100%, 99.5%, 99.4%, 99.4 to 100%, 99%, 100%, 90 to 100%, 100%, and 100% respiratory isolates susceptible to ceftolozane-tazobactam for each year (6). Similarly, susceptibility rates of ceftazidime-avibactam were 98.6% for the 2016 to 2017 period and 100% for the 2017 to 2018 period. As a result, these new β-lactam combinations are effective against many Gram-negative bacilli, including MDR P. aeruginosa associated with urinary tract infections, nosocomial pneumonia, and complicated intraabdominal infections as well as in the treatment of acute pulmonary exacerbations in cystic fibrosis (7–9). However, recent studies, including ours (10), have shown that there can be the development of resistance to these new antimicrobial agents.

The purpose of the current study was to compare the in vitro activity of ceftolozane-tazobactam and ceftazidime-avibactam with other antimicrobials on CF MDR P. aeruginosa and propose an up-to-date most promising antimicrobial combination for the treatment of CF MDR P. aeruginosa infections. A secondary objective was to determine the antimicrobial resistance rates of CF MDR P. aeruginosa and study inherent trends of these antimicrobials over 10 years. This would provide empirical evidence in the treatment of pulmonary exacerbations.

RESULTS

Study isolates.

During the study period, 721 MDR P. aeruginosa isolates from 104 female and 79 male CF patients were referred for extended susceptibility testing from 8 Scottish hospitals, while others were from York and Belfast. The median age at first referral was 27 years (range, 7 to 69 years), and with a median of 3 samples, between 1 and 20 isolates were submitted per patient during the study period.

Figure 1 shows that 69% (496/721 isolates) of the submitted isolates were resistant to the three groups of antimicrobials (MDR3) tested, while 22% (158/721) of submitted isolates were resistant to only two groups (MDR2). Of the latter, 81% (129/158) of MDR2 isolates showed resistance to the fluoroquinolones and β-lactams.

FIG 1.

Resistance profile of study isolates to antimicrobial groups. Antimicrobial agents in the aminoglycoside group are amikacin, gentamicin, and tobramycin. Levofloxacin and ciprofloxacin are grouped as fluoroquinolones, while aztreonam, ceftazidime, meropenem, and imipenem are grouped as the β-lactams. Also included in this group are β-lactam combinations, piperacillin-tazobactam, ceftazidime-avibactam, ticarcillin-clavulanate, and ceftolozane-tazobactam.

Antimicrobial susceptibility profile.

The results of MIC tests (Fig. 2) carried out on 721 isolates showed that the most active antimicrobial agent was colistin (R = 7%), followed by the new β-lactam combinations, ceftolozane-tazobactam (R = 37%) and ceftazidime-avibactam (R = 47%). Interestingly, P. aeruginosa isolates were resistant to the β-lactam combinations, piperacillin-tazobactam (TZP) (67%) and ticarcillin-clavulanate (86%). Most of the P. aeruginosa isolates were resistant to the fluoroquinolones, ciprofloxacin (89%) and levofloxacin (93%), while <70% resistance was observed for the aminoglycosides with lower resistance rates in gentamicin (36%). In summary, 20% of isolates were susceptible while 63.9% were resistant to all tested antimicrobials. The fluoroquinolones had the most resistant isolates (90.83%), followed by β-lactams (67.88%) and aminoglycosides (56.68%).

FIG 2.

Pseudomonas aeruginosa MIC susceptibility patterns to tested antimicrobials. Percentage of susceptible isolates are represented by green bars while orange and blue bars represent Intermediate and resistant isolates. Pip/Tazo, piperacillin-tazobactam; Tic/Clav, ticarcillin-clavulanate; Ceft/Tazo, ceftolozane-tazobactam; Cef/Avi, ceftazidime-avibactam.

Antimicrobial resistance trend.

When the annual mean MIC values for each antimicrobial agent were analyzed (Table 1), colistin was the only antimicrobial that showed a downward trend (R² = 0.48), while upward trends were observed for the fluoroquinolones, especially for levofloxacin (R² = 0.44). Similarly, an upward trend was observed in the β-lactam group, of which meropenem (R² = 0.4967) and piperacillin-tazobactam (R² = 0.3007) demonstrated the greatest increase. The trends for the aminoglycosides during the study period were level (R² ≤ 0.005).

TABLE 1.

Temporal variations in MIC values for CF-derived P. aeruginosa isolates^a

	Antimicrobial(s)b
Yr or parameter	Amikacin	Gentamicin	Tobramycin	Ciprofloxacin	Levofloxacin	Aztreonam	Ceftazidime	TZP	Imipenem	Meropenem	Colistin	TIM	C/T	CZA
2009	131.60	25.95	8.22	10.07	18.40	169.67	159.60	171.05	32.62	20.62	17.81	229.98
2010	76.79	38.18	58.97	10.02	16.06	133.71	152.52	189.76	28.53	18.20	20.01	206.70
2011	131.68	41.71	30.13	16.10	21.24	165.16	144.21	161.02	25.00	20.27	30.23	187.95
2012	142.28	36.88	48.99	9.74	16.02	116.48	134.09	138.38	27.32	19.14	0.73	210.01
2013	119.52	51.96	7.31	10.35	15.46	145.00	150.62	144.64	28.65	25.10	6.61	182.49
2014	195.39	103.96	41.02	18.44	26.68	227.46	192.45	192.67	30.35	32.47	6.89	272.44
2015	120.46	33.54	6.40	8.02	19.04	145.57	137.62	144.05	26.29	24.12	6.77	233.15
2016	137.84	67.30	73.53	9.72	20.85	160.54	169.03	160.38	30.04	23.80	3.67	235.37	10.20
2017	118.37	59.40	68.81	11.70	19.36	153.16	134.91	150.33	26.80	22.95	0.61	256.44
2018	106.35	40.79	30.64	13.63	20.95	158.30	161.34	245.52	39.10	29.71	7.41	35.95	93.91
2019	147.68	54.13	28.38	16.87	26.98	152.20	162.69	190.22	28.95	25.55	3.99	34.18	78.73
2020	116.80	14.20	5.60	23.60	38.40	260.00	273.20	300.00	32.90	33.50	0.39	152.30	157.10
Mean	129.14	51.38	40.36	12.28	20.64	158.17	156.71	176.00	29.93	24.25	7.66	228.75	40.89	90.41
Avg MIC50	48.00	12.00	4.00	4.00	12.00	192.00	128.00	>256	>32	>32	0.38	>256	3.00	8.00
Avg MIC90	>256	>256	32.00	>32	>32	>256	>256	>256	>32	>32	1.50	>256	>256	>256
P value	0.004	0.001	0.52	<0.001	<0.001	0.011	0.185	<0.001	0.023	<0.001	0.032	0.001	0.002	0.192
Remark	Sig	Sig	NS	Sig	Sig	Sig	NS	Sig	Sig	Sig	Sig	Sig	Sig	NS
R2	0.0050	0.0028	0.0004	0.2662	0.4438	0.1593	0.2358	0.3007	0.1171	0.4967	0.4782	0.2764	ND	ND

^aP. aeruginosa isolates (n = 721).

^bTZP, piperacillin-tazobactam; TIM, ticarcillin-clavulanate; C/T, ceftolozane-tazobactam; CZA, ceftazidime-avibactam; ND, not determined; NS, nonsignificant; Sig, significant.

We analyzed our data to determine if there were any statistically significant differences in the annual means for each antimicrobial. Analysis using the one-way analysis of variance (ANOVA) showed that there was a statistically significant difference in the annual mean MICs of all tested antimicrobials except tobramycin (P = 0.52), ceftazidime (P = 0.19), and ceftazidime-avibactam (P = 0.19).

Therefore, we investigated whether observed increases in annual antimicrobial MICs corresponded to temporal increases in annual resistant strains by assessing time-based differences in resistance to each tested antimicrobial. Table 2 shows that among the aminoglycosides, there were statistically significant differences (P = 0.001 to 0.041) in the decrease of resistant isolates, with tobramycin exhibiting the sharpest decrease (R² = 0.5633). In contrast, levofloxacin (R² = 0.472) showed an upward trend, but this was not statistically significant. For the β-lactam group (except imipenem), a statistically significant resistance increase to meropenem (P = 0.01), piperacillin-tazobactam (P < 0.001), and ticarcillin-clavulanate (P = 0.024) was observed, while a statistically significant resistance decrease to ceftazidime (P = 0.017) and aztreonam (P = 0.024) in resistance rates (R² ≤ 0.1) was observed. Interestingly, longitudinal analyses of isolates for colistin resistance showed that there was a statistically significant continuous decrease (R² = 0.6881; P < 0.001) in resistant isolates during the study period.

TABLE 2.

Temporal differences of antimicrobial resistance of CF-derived MDR Pseudomonas aeruginosa

Yr or parameter	Antimicrobial agent(s)a (%)b
	AMK	GEN	TOB	CIP	LVX	ATM	CAZ	TZP	IPM	MEM	COL	TIM	C/T	CZA
2009	46 (77)	23 (38)	45 (75)	50 (83)	52 (87)	43 (72)	39 (65)	40 (67)	50 (83)	38 (63)	15 (25)	51 (86)
2010	37 (77)	14 (29)	38 (79)	44 (92)	43 (90)	32 (67)	39 (81)	38 (79)	41 (85)	28 (58)	6 (13)	41 (87)
2011	27 (79)	11 (32)	27 (79)	29 (85)	29 (85)	23 (68)	25 (74)	20 (59)	21 (62)	19 (56)	8 (24)	28 (82)
2012	31 (62)	17 (34)	30 (60)	44 (88)	44 (88)	25 (50)	33 (66)	26 (52)	36 (72)	28 (56)	4 (8)	38 (76)
2013	35 (70)	15 (30)	32 (64)	48 (96)	48 (96)	30 (60)	30 (60)	28 (56)	37 (74)	32 (64)	3 (6)	38 (76)
2014	41 (84)	31 (63)	38 (78)	44 (90)	47 (96)	42 (86)	37 (84)	37 (76)	42 (86)	40 (83)	2 (4)	48 (98)
2015	32 (65)	15 (31)	25 (51)	44 (92)	46 (94)	27 (55)	26 (53)	30 (61)	34 (69)	32 (65)	4 (8)	45 (92)
2016	81 (72)	43 (38)	72 (64)	101 (90)	106 (95)	71 (63)	73 (65)	54 (61)	83 (74)	70 (63)	3 (3)	95 (85)
2017	56 (60)	32 (34)	53 (57)	76 (83)	84 (90)	55 (61)	52 (56)	53 (57)	72 (77)	58 (62)	3 (3)	84 (90)
2018	51 (61)	28 (33)	42 (50)	77 (92)	80 (95)	55 (65)	57 (68)	73 (87)	71 (85)	63 (75)	2 (2)	30 (36)	39 (47)
2019	61 (74)	29 (35)	47 (57)	74 (90)	78 (95)	46 (56)	50 (61)	57 (70)	61 (74)	53 (65)	1 (1)	28 (34)	36 (44)
Resistant	506 (70)	260 (36)	457 (64)	640 (89)	666 (93)	457 (64)	470 (66)	465 (67)	556 (77)	469 (65)	50 (7)	467 (86)	67 (37)	82 (47)
R2	0.0015	0.0031	0.5633	0.0396	0.472	0.0641	0.1751	0.0295	5E-05	0.1574	0.6881	0.098	ND	ND
P value	0.041	0.007	0.001	0.402	0.265	0.024	0.017	<0.001	0.101	0.010	<0.001	0.024	0.505	0.692
Remarks	Sig	Sig	Sig	NS	NS	Sig	Sig	Sig	NS	Sig	Sig	Sig	NS	NS

^aAMK, amikacin; GEN, gentamicin; TOB, tobramycin; CIP, ciprofloxacin; LVX, levofloxacin; ATM, aztreonam; CAZ, ceftazidime; TZP, piperacillin-tazobactam; IPM, imipenem; MEM, meropenem; COL, colistin; TIM, ticarcillin-clavulanate; CZA, ceftazidime-avibactam; C/T, ceftolozane-tazobactam; ND, not determined; NS, nonsignificant; Sig, significant.

^bPercentage of resistant isolates.

Antimicrobial synergy testing.

A total of 4,062 antimicrobial combination tests were performed using different antimicrobial pairs. Overall, 0.01% antagonism and 9.97% synergy were observed for all of the tested combinations. In the antimicrobial groups, 10.31% synergy was observed for aminoglycosides (n = 1,290), 9.30% for fluoroquinolones (n = 774), and 10.20% for β-lactams (n = 2,196), while low synergy rates (3.84%) were observed for colistin (n = 964). Of these, the β-lactam (cephalosporin) with aminoglycoside (n = 281) as well as β-lactam plus β-lactamase inhibitor antimicrobials (n = 19) with another β-lactam (carbapenems) gave the highest synergy values of 20.64 and 26.32%, respectively. Table 3 shows that the highest synergy was observed with antimicrobial combinations of ciprofloxacin and ceftolozane-tazobactam (n = 15; 40% synergy) followed by amikacin and ceftazidime (n = 60; 36.7% synergy). Similarly, combinations with ceftazidime were synergistic in 6/7 tested combinations. No synergy was observed when antimicrobial combinations of colistin with levofloxacin-ceftazidime or imipenem with tobramycin-ciprofloxacin were tested. In addition, Table 3 shows that synergy was observed in all of the tested combinations, with the β-lactam plus β-lactamase inhibitor antimicrobials (n = 12) with ceftolozane-tazobactam combinations as the most synergistic. Indeed, this antimicrobial combination gave the highest synergy rate (n = 82; 23.17% synergy). Synergy rates for ceftazidime-avibactam were not analyzed, as only one combination was synergistic.

TABLE 3.

Summary of results for combinations tested ≥10 times for CF-derived MDR Pseudomonas aeruginosa

First antimicrobiala (%Sb)	Second antimicrobiala (%Sb)	No.c	Synergy		Antagonistic		No interaction		SBPI
			%	Rank	%	Rank	%	Rank	Median	Rank
CIP (0)	C/T (63)	15	40.0	1	60.0	39	2.00	36
AMK (30)	CAZ (0)	60	36.7	2	63.3	38	5.00	26
LVX (0)	MEM (26)	11	36.4	3	63.6	37	0.63	39
GEN (31)	CAZ (0)	48	33.3	4	66.7	36	4.00	31
TOB (37)	CZA (53)	10	30.0	5	70.0	34	5.67	23
LVX (0)	CAZ (0)	14	28.6	6	71.4	33	5.33	25
TOB (37)	C/T (63)	25	28.0	7	72.0	32	7.33	21
AMK (30)	MEM (26)	22	27.3	8	72.7	31	3.83	32
C/T (63)	MEM (26)	19	26.3	9	5.3	2	68.4	35	5.00	26
GEN (31)	TIM (0)	14	21.4	10	78.6	28	5.75	22
CIP (0)	TZP (0)	91	20.9	11	79.1	27	7.96	19
ATM (0)	CIP (0)	23	17.4	12	4.3	3	78.3	29	3.67	33
GEN (31)	CIP (0)	18	16.7	13	83.3	26	1.83	37
CAZ (0)	MEM (26)	31	16.1	14	6.5	1	77.4	30	4.67	30
GEN (31)	ATM (0)	20	15.0	15	85.0	25	7.50	20
CIP (0)	MEM (26)	138	14.5	16	85.5	23	2.33	35
CIP (0)	CAZ (0)	48	12.5	17	2.1	5	85.4	24	5.67	23
TOB (37)	CAZ (0)	173	11.6	18	88.4	22	8.20	18
TOB (37)	ATM (0)	156	11.5	19	88.5	21	9.33	16
COL (93)	IPM (0)	44	9.1	20	90.9	20	38.17	2
GEN (31)	COL (93)	13	7.7	21	92.3	19	37.05	4
COL (93)	ATM (0)	54	7.4	22	92.6	18	26.32	8
GEN (31)	MEM (26)	28	7.1	23	92.9	17	5.00	26
TOB (37)	TZP (0)	107	6.5	24	93.5	16	32.50	6
TOB (37)	MEM (26)	287	5.2	25	94.8	15	16.67	14
TOB (37)	LVX (0)	21	4.8	26	95.2	14	8.67	17
COL (93)	MEM (26)	251	4.4	27	95.6	13	36.51	5
COL (93)	TZP (0)	138	4.3	28	95.7	11	50.11	1
COL (93)	C/T (63)	23	4.3	28	95.7	11	29.05	7
COL (93)	TIM (0)	90	3.3	30	96.7	8	37.26	3
TOB (37)	CIP (0)	60	3.3	30	96.7	8	5.00	26
TOB (37)	COL (93)	41	2.4	32	97.6	5	15.79	15
COL (93)	CIP (0)	249	2.4	32	97.6	5	23.63	9
TOB (37)	TIM (0)	127	2.4	32	97.6	5	21.83	10
CIP (0)	TIM (0)	58	1.7	35	98.3	4	2.92	34
COL (93)	LVX (0)	17	100.0	1	21.26	11
COL (93)	CAZ (0)	44	100.0	1	20.00	12
TOB (37)	IPM (0)	60	100.0	1	16.83	13
CIP (0)	IPM (0)	25	4.0	4	96.0	10	1.67	38

^aAMK, amikacin; GEN, gentamicin; TOB, tobramycin; CIP, ciprofloxacin; LVX, levofloxacin; ATM, aztreonam; CAZ, ceftazidime; TZP, piperacillin-tazobactam; IPM, imipenem; MEM, meropenem; COL, colistin; TIM, ticarcillin-clavulanate; CZA, ceftazidime-avibactam; C/T, ceftolozane-tazobactam.

^bPercentage susceptible when used as a single agent.

^cNumber of times the combinations were tested.

DISCUSSION

The use of antimicrobials has been demonstrated to greatly improve the life expectancy of CF patients (11). However, a major drawback of this management approach is the development of antimicrobial resistance due to exposure to several multiple antimicrobial cocktails (1–4, 12). To manage infective pulmonary exacerbations, CF patients are treated with antimicrobial combinations, of which one/both is generally effective as a single agent and there is a lack of evidence guiding the clinician to decide the best antimicrobial combination that would give a positive treatment outcome (5). Our study focused on P. aeruginosa. Bullington et al. (13) reported that 62% of health care providers and 56% of people living with CF are concerned about antimicrobial-resistant infections from P. aeruginosa and Burkholderia spp. This study analyzed the multi and extensively drug-resistant isolates received by our CF antimicrobial reference laboratory and, hence, does not provide a representative picture of the general CF population. Nonetheless, as previously reported by studies sampling CF patients (5), we observed that colistin (93% susceptible) was the most active antimicrobial. These results should be interpreted with care because, for colistin susceptibility testing, it is advised that the use of broth microdilution should be employed (14), but our lab used the Etest method. In keeping with the same study (5), ciprofloxacin was the most active fluoroquinolone. However, we show that a steady upward trend in annual MIC values was observed for the quinolone antimicrobial class. This predominance of fluoroquinolone-resistant isolates in our study population may be linked to the use of ciprofloxacin for first isolates or patients chronically infected with P. aeruginosa as per European guidelines (15). Fluoroquinolones are used in the treatment of a range of infections due to its safety, oral bioavailability, and broad-spectrum activity (16, 17). Despite several guidelines to limit the use of fluoroquinolones in human and veterinary medicine, quinolone resistance in all species targeted by this antimicrobial class has been growing steadily (16–20). Also, our data suggest that for the aminoglycosides (especially tobramycin) and colistin, there were increases in P. aeruginosa susceptibility rates; however, in contrast, for the fluoroquinolones, we observed that there was an ∼50% upward trend in the resistance to levofloxacin. Therefore, we agree with Cogen et al. (11), who reported that although antimicrobial stewardship in this patient population is challenging, its role and impact would enrich patient management and care.

In this study, ceftolozane-tazobactam and ceftazidime-avibactam were observed as the most susceptible β-lactam antimicrobials tested. However, our susceptibility rates were lower in contrast with previous studies, which reported in vitro activity of ceftolozane-tazobactam (85.1%) against P. aeruginosa as comparable with the activity of colistin (89.4%) (21). Gramegna et al. (22), working on 120 CF-derived P. aeruginosa isolates, demonstrated that the lowest percentage of in vitro drug resistance was observed using ceftolozane-tazobactam with an 84.2% susceptibility rate. A plausible explanation of the difference in susceptibility rates might be the study isolate population; their study was composed of 55% susceptible strains, therefore increasing the susceptibility rates. Indeed, Zamudio et al. (10) reported lower susceptibility values (50%) and Finklea et al. (23) agreed that lower susceptibility values (30%) were observed if the isolate population differed. Similarly, Mirza et al. proposed that previous studies had reported a susceptibility rate of 65.4% to 94% for ceftolozane-tazobactam and 51.8% to 92% for ceftazidime-avibactam in meropenem-nonsusceptible isolates (24). Several resistance mechanisms have been proposed; for example, our laboratory, characterizing resistance mechanisms in P. aeruginosa, showed that it is due to mutation in the AmpC β-lactamase and loss of outer membrane porin D (OprD), while ceftolozane-tazobactam and ceftazidime-avibactam double resistance is associated with AmpD β-lactamase variations (10). However, more research is important to determine other resistance mechanisms that would help develop effective strategies to cope with drug resistance and for epidemiological studies.

To improve efficacy while preventing the emergence of drug resistance, antimicrobial combinations are often prescribed in the management of CF patients (5). However, the selection of an optimal combination remains a continual clinical challenge. In a previous work published by our laboratory (5), the antimicrobial combination of amikacin-ceftazidime was stated as the most synergistic combination. This present study reiterates the dominance of this combination as one of the most synergistic combinations. Interestingly, Nazli et al. (25) demonstrated a 15% synergy using amikacin-ceftazidime antimicrobial combinations. Furthermore, our analysis demonstrates that combinations with β-lactam combinations were synergistic. Indeed, newer β-lactam combinations with ciprofloxacin, tobramycin, and meropenem showed promising results (>25% synergy). The most promising antimicrobial combination in the present study was ciprofloxacin plus ceftolozane-tazobactam. On the basis of our data suggesting in vitro effectiveness of ciprofloxacin antimicrobial combinations with ceftolozane-tazobactam, we propose that this combination is explored in clinical care, particularly on the backdrop of restrictions in fluoroquinolone usage. The use of this combination therapy may reduce the likelihood of the emergence of antimicrobial resistance and achieve multitarget engagements through inhibition of DNA replication and cell wall biosynthesis. The use of the susceptible breakpoint index (SBPI) was proposed earlier (5) as an index for ranking in vitro effectiveness of combinations. Our results suggest that combinations of colistin with several antimicrobials can give high SBPI values while not predicting synergism as measured by fractional inhibitory concentration index (FICI). Though the reason for this is unclear, we hypothesize that while both indices use the combination MIC, SBPI compares it with the organism’s susceptible breakpoint while FICI employs the single-agent MIC.

We acknowledge several limitations to this study. The study population consisting of mainly a multidrug-resistant isolate population might have impacted our observations. Also, the choice of antimicrobials and the combination cutoff (≥10 times) might have impacted our results. For example, it would have made our data richer if other newer combinations, such as cefiderocol, which has low affinity for AmpC β-lactamases and is active against carbapenem-nonsusceptible isolates, were used in susceptibility/synergy testing.

In summary, this research reiterates the upward trend in fluoroquinolone resistance and the increase in susceptibility to colistin and aminoglycosides in CF isolates, suggesting effective antimicrobial stewardship for these antimicrobial agents. It also gives empirical in vitro evidence that antimicrobial combinations with β-lactam plus β-lactamase inhibitors may be the best synergistic antimicrobial combinations to use in the treatment of chronic P. aeruginosa infections.

MATERIALS AND METHODS

Study isolates.

Between 13 January 2009 and 2 April 2020, 721 CF-MDR Pseudomonas aeruginosa isolates identified by 10 British laboratories were collected over 10 years when they were sent for extended antimicrobial susceptibility testing. Isolates were stored in the bacterial preservation system Microbank (Pro-Lab Diagnostics, ON, Canada) at −80°C and were plated upon receipt onto Mueller-Hinton (MH) agar, MacConkey agar, Pseudomonas cetrimide agar, and Burkholderia cepacia selective agar plates (all agar plates were manufactured by Oxoid Ltd., Basingstoke, UK). After 18 to 24 h incubation in ambient air at 35°C, plates were verified for culture purity. As a confirmatory test, oxidase testing (Oxoid Ltd., Basingstoke, UK) was performed on 18- to 24-h colonies. Isolates were accepted as Pseudomonas aeruginosa when they were oxidase-positive and non-lactose fermenting. In this study, multidrug resistance was defined as acquired nonsusceptibility to at least one agent in ≥3 antimicrobial groups (26). These isolates were referred to as MDR3 while MDR2 and MDR1 referred to isolates with resistance to two and one antimicrobial groups, respectively.

MIC testing.

MIC testing was performed on MH agar using the Etest methodology according to the manufacturer’s instructions (Liofilchem, Abruzzi, Italy; bioMérieux, Basingstoke, UK). The antimicrobials tested were the aminoglycosides (amikacin, gentamicin, and tobramycin), fluoroquinolones (ciprofloxacin and levofloxacin), lipopeptides (colistin), and the β-lactams. Of the β-lactams, mono-agents tested were monobactams (aztreonam), cephalosporins (ceftazidime), and carbapenems (imipenem and meropenem), while combinations tested were piperacillin-tazobactam, ticarcillin-clavulanate, ceftazidime-avibactam, and ceftolozane-tazobactam. Susceptibility of ticarcillin-clavulanate included in the analyses was up to its stop date (2017), while ceftazidime-avibactam and ceftolozane-tazobactam were included in the analyses from the time of introduction (January 2018).

In this study, MIC values between the standard doubling dilution scale were rounded up to the next doubling dilution. The MICs for all tested antimicrobials were interpreted as susceptible (S), intermediate (I), or resistant (R) according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) approved interpretive standards for P. aeruginosa (14). Due to changes in EUCAST breakpoints during the studied period, isolate susceptibility patterns were determined according to the year of submission.

Combination testing.

Antimicrobial combination testing for each isolate was performed using a minimum of six pairs of antimicrobials as previously described (5). Briefly, a saline suspension of 0.5 McFarland standard (1.0 for mucoid strains) from 24-h cultures was inoculated onto MH agar plates according to the EUCAST guidelines for the disk diffusion plate inoculation. Two Etest strips (A and B) were placed top-to-tail according to the manufacturer’s instructions. After 1 h to allow antimicrobial diffusion into the agar, each strip was removed and replaced with a fresh Etest (i.e., Etest A strip replaced with fresh Etest B strip and vice versa). Plates were further incubated for 18 ± 2 h in ambient air at 35 ± 1°C.

(i) Fractional inhibitory concentration index.

Synergy MIC was expressed using the FICI and calculated as described below.

$$\text{FICI}=\left({\text{MIC\hspace{0.17em}A}}_{\text{combination}}/{\text{MIC\hspace{0.17em}A}}_{\text{single}}\right)\hspace{0.17em}+\hspace{0.17em}\left({\text{MIC\hspace{0.17em}B}}_{\text{combination}}/{\text{MIC\hspace{0.17em}B}}_{\text{single}}\right)$$

If an MIC value was greater than the antimicrobial range tested, the next doubling dilution above this value of the range tested was used to calculate the FICI (e.g., if an MIC of >32 mg/liter was found, then the FICI was calculated using 64 mg/liter) (27). These indices were interpreted as synergy, FICI ≤ 0.5; no interaction, FICI > 0.5 and FICI ≤ 4.0; and antagonism, FICI > 4.0 (28).

Analyses of species susceptibility to synergy combinations (≥10 replicates) of tested antimicrobials were carried out when EUCAST breakpoints for P. aeruginosa were known.

(ii) Susceptible breakpoint index.

The SBPI was used to describe synergy analysis and calculated as described below.

SBPI = (susceptible breakpoint of antimicrobial A/MIC of antimicrobial A_combination) + (susceptible breakpoint of antimicrobial B/MIC of antimicrobial B_combination) (5). These combination results were categorized in rank order of their decreasing SBPI results. All antagonistic (FICI > 4.0) combinations irrespective of the SBPI result were not ranked or recommended for therapy.

Statistical methods.

Statistical analyses of categorical and continuous variables were carried out using Microsoft Office Excel 2013 and IBM SPSS statistics for Windows, version 24 (IBM Corp., Armonk, NY, USA). The One-way ANOVA with Duncan post hoc test was used for continuous data while the Kruskal-Wallis test was used for comparing categorical data.