In Vitro Susceptibility of Multidrug-Resistant Pseudomonas aeruginosa following Treatment-Emergent Resistance to Ceftolozane-Tazobactam

ABSTRACT

We compared the in vitro susceptibility of multidrug-resistant Pseudomonas aeruginosa isolates collected before and after treatment-emergent resistance to ceftolozane-tazobactam. Median baseline and postexposure ceftolozane-tazobactam MICs were 2 and 64 μg/ml, respectively. Whole-genome sequencing identified treatment-emergent mutations in ampC among 79% (11/14) of paired isolates. AmpC mutations were associated with cross-resistance to ceftazidime-avibactam but increased susceptibility to piperacillin-tazobactam and imipenem. A total of 81% (12/16) of ceftolozane-tazobactam-resistant isolates with ampC mutations were susceptible to imipenem-relebactam.

INTRODUCTION

Pseudomonas aeruginosa is a common nosocomial pathogen in U.S. hospitals. Multidrug-resistant (MDR) P. aeruginosa represents a major public health threat, accounting for 32,600 cases, 2,700 deaths, and $767 million in attributable health care costs annually (1). In clinical practice, treatment of MDR P. aeruginosa infection is associated with high rates of clinical failure and recurrent infections (2, 3). Ceftolozane-tazobactam was specifically designed to overcome classic MDR P. aeruginosa resistance mechanisms, leading to improved clinical outcomes compared with salvage therapy (2, 4). At our center, treatment of MDR P. aeruginosa infections with ceftolozane-tazobactam resulted in 30-day clinical cure and survival rates of 55% and 77%, respectively (4, 5). Despite these encouraging results, we have also shown that resistance emerged in 15% of patients after treatment courses ranging from 7 to 53 days (4). Mutations in chromosomal ampC and ampR genes are associated with treatment-emergent ceftolozane-tazobactam resistance (4, 6–11). It is unclear whether these specific mutations influence susceptibility to other β-lactam agents that might be viable treatment options, including ceftazidime-avibactam and imipenem-relebactam (8, 10–12). We hypothesized that treatment-emergent mutations in ampC and/or ampR that mediate resistance to ceftolozane-tazobactam impact the activity of other β-lactams by altering target binding sites and the subsequent hydrolysis spectrum of P. aeruginosa-derived AmpC β-lactamase (PDC). The objectives of this study were to measure the in vitro activity of β-lactam agents against MDR P. aeruginosa before and after emergence of ceftolozane-tazobactam resistance and to associate specific ampC mutations with antibiotic susceptibility changes in postexposure isolates.

We identified patients treated with ceftolozane-tazobactam for >72 h between August 2015 and May 2019, with microbiological failure defined as reisolation of MDR P. aeruginosa after ≥5 days of therapy (4, 5). All isolates collected after exposure to ceftolozane-tazobactam were initially screened for resistance by Etest (bioMérieux) or broth microdilution (BMD) and included if the ceftolozane-tazobactam MIC was ≥8 μg/ml by BMD. P. aeruginosa isolates were tested in triplicate by BMD, and results were interpreted according to 2020 CLSI criteria; modal MICs were used for analysis (13). Ceftolozane and ceftazidime were tested at a range of 0.25 to 256 μg/ml, whereas imipenem and piperacillin were tested at a range of 0.03 to 32 and 0.5 to 512 μg/ml, respectively. Avibactam, relebactam, and tazobactam were tested at a fixed concentration of 4 μg/ml. Quality control (QC) strains Escherichia coli ATCC 25922 and P. aeruginosa ATCC 27853 were used throughout, and all QC results were within accepted ranges. Statistical analysis was performed using GraphPad Prism 8 (San Diego, CA).

Twenty-three baseline and 32 postexposure MDR P. aeruginosa isolates collected from 23 patients were included; the median duration of ceftolozane-tazobactam therapy was 16 days (range, 3 to 60 days). Median ceftolozane-tazobactam MICs against baseline and postexposure isolates were 2 μg/ml (≤0.25 to 8 μg/ml) and 64 μg/ml (8 to >256 μg/ml), respectively (Table 1), representing a median 32-fold MIC increase. Corresponding median ceftazidime and ceftazidime-avibactam MICs were increased 4- and 8-fold, respectively. Postexposure median MICs against piperacillin-tazobactam and imipenem-relebactam were unchanged from baseline, whereas median imipenem MICs were decreased 2-fold.

TABLE 1.

Antimicrobial MICs against P. aeruginosa isolates before and after exposure to ceftolozane-tazobactam across all patients

Drug	Baseline (n = 23)		Postexposure (n = 32)		P valuea
	Median MIC (range) (μg/ml)	Susceptible (%)	Median MIC (range) (μg/ml)	Susceptible (%)
Ceftolozane-tazobactam	2 (≤0.25–8)	91	64 (8 to >256)	0	<0.0001
Ceftazidime	32 (1–256)	26	32 (32 to >512)	0	<0.0001
Ceftazidime-avibactam	4 (1–32)	74	64 (4 to >256)	28	<0.0001
Imipenem	16 (0.12–32)	17	4 (0.5 to >32)	50	0.0216
Imipenem-relebactam	2 (0.06–16)	65	2 (0.25–16)	63	0.6625
Piperacillin-tazobactam	128 (1–512)	17	128 (4 to >512)	16	0.6284

^aComparison of median MICs, Mann-Whitney U test.

To associate specific phenotypic changes to ampC mutations we selected representative serial isolates from 14 patients that were collected immediately before and after ceftolozane-tazobactam treatment for whole-genome sequencing (WGS). Isolates were prepared as previously described and sequenced on the Illumina NextSeq platform (4, 14). Genomes were assembled using SPAdes 3.14 (15) and annotated using Prokka (16). Nucleotide sequences for ampC, ampD, ampR, dacB, ftsI, and oprD were compared with sequences of wild-type (WT) P. aeruginosa strain PAO1. Single nucleotide polymorphisms (SNPs), insertions, and deletions were called using BWA and SAMtools (17, 18). Multilocus sequence typing was performed using MLST v2.19 (https://github.com/tseemann/mlst), and resistance genes were identified using AMRFinderPlus v3.8.4 (19).

A total of 15 baseline and 21 postexposure isolates from 14 patients underwent WGS (Table 2). All isolates encoded OXA-50-like chromosomal β-lactamases; none harbored metallo-β-lactamases or other serine β-lactamases known to confer resistance to ceftolozane-tazobactam (20, 21). Baseline isolates from 12 patients belonged to unique sequence types (STs); the remaining 2 patients were infected by novel STs that were not in the MLST database. Common PDC variants at baseline included PDC-5 (6 patients), PDC-3 (4 patients), and PDC-24 (2 patients). Postexposure isolates from 79% (11/14) of patients showed mutations in ampC, including amino acid point mutations (n = 7), insertions (n = 2), and deletions (n = 3). Sequential isolates from 1 patient (Table 2, patient 9) showed point mutations followed by a combination of point mutations and an amino acid insertion at position 243 with continued treatment. The most frequently identified treatment-emergent point mutations were F147L and G183D (3 patients each). Chromosomal ampC mutations were not identified among serial isolates from 3 patients, which were associated with modest 2- to 4-fold ceftolozane-tazobactam MIC increases. In each of those cases, mutations in ampC regulatory genes ampR, ampD, or dacB were present at baseline but did not change after ceftolozane-tazobactam exposure. AmpC promoter region mutations were not found in any isolates.

TABLE 2.

Molecular and phenotypic characterization of P. aeruginosa isolates from patients treated with ceftolozane-tazobactam

Patient no./infection and isolate typea	Days of C/T before collection	SNPb	ST	PDC variant	Mutationc in:						MIC (μg/ml)
					ampC	ampR	ampD	dacB	ftsI (PBP3)	oprDd
											C/T	CAZ	CZA	TZP	IPM	I-R
1/BSI, VAP
Baseline	0	209	PDC-24	T21A; G391A, T105A	WT	R11L, G148A, D183Y	P88L, A394P	G216S	W277*	2	64	2	256	32	4
Postexposure	14	29	209	PDC-24	T21A, G391A, T105A	WT	R11L, G148A, D183Y	P88L, A394P	G216S	E384*	8	128	32	>512	32	16
3/VAP
Baseline	0	3118	PDC-5	R79Q, T105A	R244W, G283E, disrupted	G148A	S326P	WT	E426*	4	128	16	256	16	2
Postexposure	10	0	3118	PDC-244	R79Q,234–241 del, T105A	R244W, G283E, disrupted	G148A	S326P	WT	E426*	256	512	256	32	1	1
4/HAP
Baseline	0	645	L176H, E335K, T105A	WT	WT	D176N	WT	8	256	8	512	16	4
Postexposure	16	57	645	F147L, L176H, G183D, E335K, T105A	WT	WT	D176N	WT	64	128	64	16	0.5	1
7/CFE
Baseline	0	274	PDC-24	T21A, G391A, T105A	WT	R11L, G148A, D183Y	A394P	WT	13nt del fs, 225*	1	16	4	128	32	8
Postexposure	31	26	274	T21A, G391A, D245 del, T105A	WT	R11L, G148A, D183Y	A394P	WT	13nt del fs, 225*	16	32	8	32	1	1
8/VAP
Baseline	0	111	PDC-3	T105A	G283E, disrupted	A85 duplicated, G148A	WT	WT	1nt del fs, 345*	1	8	4	512	32	2
Postexposure	3	3	111	PDC-3	T105A	G283E, disrupted	A85 duplicated, G148A	WT	WT	1nt del fs, 345*	1	8	4	512	32	2
Postexposure	20	8, 9	111	G183R, T105A	G283E, disrupted	A85 duplicated, G148A	WT	WT	1nt del fs, 345*	64	128	64	512	2	1
9/CFE
Baseline	0	PDC-5	R79Q, T105A	WT	13nt del fs, 177*	WT	V471G	W65*	1	8	1	64	16	2
Baseline	0	81	PDC-5	R79Q, T105A	WT	G148A, P168T, S175L	WT	WT	1nt del fs, 473*	1	32	4	128	16	2
Postexposure	26	83, 2	R79Q, P243S, T105A	WT	G148A, P168T, S175L	WT	WT	1nt del fs, 473*	16	32	4	16	2	1
Postexposure	45	80, 5	R79Q, D234G, P243S, T105A	WT	G148A, P168T, S175L	WT	WT	1nt del fs, 473*	128	256	128	32	2	1
Postexposure	45	316, 291	R79Q, G183D, 243G ins, T105A	WT	G148A, S175L	S257L	M460T	W65*	128	128	128	32	1	2
10/IAI
Baseline	0	155	PDC-5	R79Q, T105A	WT	Q44H, G148A, D183Y	Disrupted	WT	E424*	2	128	32	256	32	8
Postexposure	8	PR	155	R79Q, T105A,236–242del	WT	Q44H, G148A, D183Y	Disrupted	WT	E424*	128	512	256	32	2	2
Postexposure	40	PR	155	R79Q, T105A,236–242del	WT	Q44H, G148A, disrupted, D183Y	Disrupted	WT	E424*	>256	>512	>256	128	2	1
12/CFE
Baseline	1	1207	PDC-3	T105A	1nt ins fs, 292*	WT	WT	R504C	4nt del fs, 431*	2	64	4	128	16	0.5
Postexposure	19	PR	1207	T96I, T105A	G273R, 1nt ins fs, 292*	WT	H261R	R504C	2nt del fs, 193*	256	32	8	8	2	2
Postexposure	42	PR	1207	T96I, T105A	G273R, 1nt ins fs, 292*	WT	H261R	R504C	2nt del fs, 193*	128	128	64	32	2	2
13/CF, VAP
Baseline	0	883	PDC-5	R79Q, T105A	R244W, G283E, disrupted	A136V, 5nt ins fs, stop lost	WT	WT	R310S	1	16	8	128	2	0.5
Postexposure	52	6	883	PDC-264	R79Q, F147L, T105A	R244W, G283E, disrupted	A136V, 5nt ins fs, stop lost	N316K	R504C	R310S, W339*	256	512	128	256	16	8
16/UTI w/ stent
Baseline	0	2167	PDC-3	T105A	D135G	G148A	WT	WT	WT	4	64	4	512	2	0.5
Postexposure	7	3	2167	PDC-3	T105A	D135G	G148A	WT	WT	WT	16	512	64	>512	8	2
18/HAP
Baseline	0	699	PDC-1	WT	WT	WT	WT	WT	≤0.25	4	4	16	0.12	0.06
Postexposure	7	31	699	PDC-335	F147L	WT	WT	1nt ins fs, 316*	WT	WT	16	64	4	128	0.5	0.25
19/CFE
Baseline	0	796	PDC-3	T105A	WT	G148A, S175L	WT	WT	W415*	2	16	2	64	32	4
Postexposure	30	16	796	241R ins, T105A	WT	G148A, G156R, S175L	WT	R504C	E394K	128	>512	128	512	32	4
Postexposure	30	16, 1	796	241R ins, T105A	WT	G148A, G156R, S175L	WT	R504C	E394K	128	>512	128	>512	32	4
20/VAP
Baseline	0	1074	PDC-5	R79Q, T105A	D135N	WT	WT	WT	1nt del fs, 473*	4	128	8	512	16	1
Postexposure	6	33	1074	PDC-5	R79Q, T105A	D135N	WT	WT	WT	1nt del fs, 473*	8	128	16	>512	32	2
25/VAP
Baseline	0	PDC-5	R79Q, T105A	WT	G148A, S175L	Disrupted	WT	Start lost	4	64	16	256	32	2
Postexposure	3	4	PDC-346	R79Q, G183D, T105A	WT	G148A, S175L	Disrupted	WT	Start lost	>64	128	128	32	1	1
Postexposure	7	4, 6	PDC-5	R79Q, T105A	WT	G148A, S175L	Disrupted	WT	Start lost	4	128	16	256	32	2

^aAbbreviations: BSI, bloodstream infection; CAZ, ceftazidime; CF, cystic fibrosis; CFE, cystic fibrosis exacerbation; C/T, ceftolozane-tazobactam; CZA, ceftazidime-avibactam; del, deletion; fs, frameshift; HAP, hospital-acquired pneumonia; IAI, intra-abdominal infection; IMP, imipenem; ins, insertion; I-R, imipenem-relebactam; nt, nucleotide; PDC, Pseudomonas-derived cephalosporinase; PR, previously reported (4); TZP, piperacillin-tazobactam; UTI, urinary tract infection; VAP, ventilator-associated pneumonia; *, stop codon.

^bNumber of SNPs in postexposure isolate compared with genetically related baseline.

^cTreatment-emergent mutations are in boldface.

^dOnly mutations predicted to result in partial or complete OprD inactivation are reported (start lost, stop lost, early termination, and indel). Individual SNPs present in both baseline and postexposure isolates or those occurring after the stop codon are not reported.

Treatment-emergent mutations in ampR, ampD, and/or dacB were identified in postexposure isolates from 6 patients (Table 2). New mutations were also identified in PBP3 (ftsI) among serial isolates in 3 patients. Compared to P. aeruginosa PAO1, in-frame insertions and other mutations in DNA polymerase subunits gamma and tau (dnaX) were observed for all isolates; however, compared to respective baseline isolates, only two postexposure isolates harbored unique in-frame insertions of various lengths (Table 2, patients 9 and 25). Across all postexposure isolates, treatment-emergent mutations in ampR, ampD, dacB, ftsI, and dnaX only occurred in the presence of ampC mutations, suggesting a complementary rather than independent role in ceftolozane-tazobactam resistance.

Median ceftolozane-tazobactam MICs against postexposure isolates with and without treatment-emergent ampC mutations were 128 and 8 μg/ml, respectively (P = 0.0002). Corresponding ceftolozane-tazobactam MIC fold changes compared to paired baseline isolates were 64- and 2-fold, respectively (P < 0.0001). Median ceftazidime and ceftazidime-avibactam MICs against postexposure isolates with ampC mutations were increased by 6- and 16-fold, respectively. In contrast, median postexposure isolates were more susceptible by 2-, 4-, and 8-fold against imipenem-relebactam, piperacillin-tazobactam, and imipenem, respectively. Seventy-five percent (12/16) of postexposure isolates with new ampC mutations displayed cross-resistance to ceftazidime-avibactam (MIC range, 4 to >256 μg/ml); 81% of such isolates were susceptible to both imipenem and imipenem-relebactam. Against all postexposure P. aeruginosa isolates demonstrating nonsusceptibility to ceftolozane-tazobactam, 68% (13/19) and 79% (15/19) were susceptible to imipenem and imipenem-relebactam, respectively. Median imipenem and imipenem-relebactam MICs were 2 and 1 μg/ml, respectively. Two isolates demonstrated susceptibility to imipenem-relebactam but not to imipenem; both harbored mutations in ampR and/or ampD but not ampC. Each of the 4 postexposure isolates demonstrating nonsusceptibility to imipenem-relebactam harbored oprD mutations, including early terminations (n = 2) or substitutions (n = 2) (Table 2). Interestingly, 3 of 4 imipenem-relebactam-resistant postexposure isolates harbored an R504C mutation in PBP3 that was not present at baseline. New mutations in efflux genes mexA, mexB, and mexR or penicillin binding protein ponA (PBP1a) that could be associated with imipenem-relebactam resistance were not identified (10).

Our data highlight the important role ampC mutations play during treatment against MDR P. aeruginosa infection. Our findings strengthen and corroborate two prior reports describing ampC mutations in 40% (2/5) and 80% (4/5) of isolate pairs from patients treated with ceftolozane-tazobactam (8, 11). In our experience, ampC mutations were identified in 79% (11/14) of paired isolates collected before and after ceftolozane-tazobactam exposure. Mutations within the AmpC Ω-loop were common across all three studies. The Ω-loop of AmpC is well known to impact the catalytic efficiency and spectrum of cephalosporin hydrolysis (22, 23). Mutations in this region widen the enzyme’s binding pocket to accommodate the bulkier R2 side chain of ceftolozane, resulting in increased catalysis against both ceftazidime and ceftolozane (24, 25). Interestingly, mutations within the Ω-loop also result in a reduced affinity for avibactam (22, 24), likely explaining the high rates of cross-resistance reported here and in previous studies (8, 11, 26). Equally important, we identified other key residues likely implicated in ceftolozane-tazobactam resistance, including previously reported mutations at positions T96, F147, and G183, which interact with the Ω-loop, resulting in structural modifications (4, 23, 24, 27). Finally, we identified ampC mutations not previously reported, including a G183R point mutation and single amino acid insertions (241Rins and 243Gins) or deletions (D245del) within the Ω-loop, that merit further validation.

A key finding from our study is the increased susceptibility to piperacillin-tazobactam, imipenem, and imipenem-relebactam against selected P. aeruginosa isolates following the development of ceftolozane-tazobactam resistance, which is consistent with prior reports (9, 12). Whereas piperacillin-tazobactam MICs were not lowered below the susceptibility breakpoint in most cases, imipenem MICs were. These manifestations are likely a consequence of AmpC structural modifications leading to a widened binding pocket, which in turn allows for carbapenems to rotate their bulky 6α-hydroxyethyl side chain away from key binding residues, thereby evading hydrolysis (22). The comparative MICs of imipenem and imipenem-relebactam support this mechanism in that relebactam did not significantly reduce imipenem MICs against isolates harboring ampC mutations (Table 2). At the same time, treatment of ceftolozane-tazobactam-resistant P. aeruginosa with imipenem alone may select for, or induce, PDC variants capable of hydrolyzing imipenem. It is not clear whether relebactam plays a role in preserving imipenem susceptibility. Postexposure isolates resistant to both ceftolozane-tazobactam and imipenem were recovered from 5 patients, including 2 patients initially infected by imipenem-susceptible strains. Isolates from all 5 patients harbored highly polymorphic OprD protein sequences, including a shortened loop 7 that was described previously (28, 29) and various other amino acid substitutions (Table 2). In 3 of 5 cases, postexposure isolates contained mutations causing a frameshift or loss of the start or stop codon, predicted to result in complete porin loss. Isolates from another patient contained an E394K mutation not present at baseline. Isolates from the remaining patient showed WT oprD genes present in baseline and postexposure paired isolates. It is unlikely that any of the oprD mutations referenced here alter the activity of ceftolozane (10, 12). Thus, the interplay between ceftolozane, imipenem, and relebactam merits future investigation as a combination strategy to suppress the emergence of resistance in MDR P. aeruginosa strains.

From a clinical standpoint, imipenem-relebactam demonstrates in vitro activity against most MDR P. aeruginosa clinical isolates (12). The data reported here further support a potential role against MDR P. aeruginosa isolates that develop resistance to ceftolozane-tazobactam; however, MICs were not always within the susceptible range. Indeed, 63% of all postexposure isolates were susceptible to imipenem-relebactam, including 81% of isolates with treatment-emergent mutations in ampC. These observations are consistent with the preserved activity of both imipenem and imipenem-relebactam against a collection of PAO1 P. aeruginosa isogenic mutants that included strains with ampC mutations T96I, F147L, and E247K (12). Notably, imipenem-relebactam MICs were 4- to 16-fold lower than those of imipenem alone when other mechanisms of resistance were considered, such as AmpC hyperproduction, OprD inactivation, and efflux pump overexpression (12). Therefore, imipenem-relebactam may be preferred over imipenem when treatment is indicated. Other therapeutic options include cefiderocol, which has been reported to be highly active against ceftolozane-tazobactam-resistant P. aeruginosa strains in surveillance studies (30). The cefiderocol MIC₅₀ and MIC₉₀ against 199 ceftolozane-tazobactam-nonsusceptible P. aeruginosa strains were 0.25 and 2 μg/ml, respectively; however, none of the strains were collected after exposure to ceftolozane-tazobactam (31). For both, imipenem-relebactam and cefiderocol, the clinical data to support use against MDR P. aeruginosa infections are limited but encouraging compared with polymyxin-based regimens (32, 33).

Finally, it should be acknowledged that observations reported here require further validation. Most notably, we did not measure changes in ampC expression for isolates with treatment-emergent mutations; however, we have previously shown a 4- to 30-fold increase in expression levels among strains harboring a T96I mutation or a 21-bp deletion within the AmpC Ω-loop (4). We also identified new mutations in AmpC regulator genes ampR, ampD, and dacB, suggesting that ampC overexpression plays a complementary role in ceftolozane-tazobactam resistance in some isolates. In future studies, it will be important to further define the roles of other mechanisms of ceftolozane-tazobactam resistance, including mutations in DNA polymerase subunits gamma and tau (dnaX) and PBP3 (ftsI) that were previously reported (4, 8). The available data for DNA polymerase subunits gamma and tau in particular require continued investigation. A recent study by Tamma and colleagues (8) identified new dnaX mutations, V406G and V477I, after treatment with ceftolozane-tazobactam; however, the same mutations were common in both baseline and postexposure isolates in our experience. At this point, it is unknown whether such mutations will contribute to cross-resistance with ceftazidime-avibactam or reverted susceptibility to imipenem as reported with ampC mutations in an isogenic background (10). Moreover, the frequency of ceftolozane-tazobactam resistance among patients with MDR P. aeruginosa infections is not well defined but likely varies by infection type, treatment duration, dosing strategy, and the molecular characteristics of infecting strains. To date, the prevalence of ceftolozane-tazobactam resistance across all treated patients ranges from 11% to 15% (4, 5, 12) but is higher in the setting of recurrent infections (8). Thus, strategies to suppress or overcome resistance are ongoing research priorities.

In conclusion, our data underscore that ampC mutations are the predominant mechanism of treatment-emergent ceftolozane-tazobactam resistance and most often involve point mutations, insertions, or deletions within the AmpC Ω-loop. Mutations at these sites are associated with cross-resistance to ceftazidime-avibactam but restored susceptibility to imipenem. Imipenem-relebactam demonstrates in vitro activity against most ampC variants and may be a reasonable therapeutic option in the setting of treatment-emergent ceftolozane-tazobactam resistance. Ultimately, additional clinical experience is needed to define the efficacy and durability of imipenem-relebactam treatment for ceftolozane-tazobactam-resistant P. aeruginosa infections.

Data availability. Sequences are available in BioProject under number PRJNA715186.