In Vitro Selection of Meropenem Resistance among Ceftazidime-Avibactam-Resistant, Meropenem-Susceptible Klebsiella pneumoniae Isolates with Variant KPC-3 Carbapenemases

ABSTRACT

Ceftazidime-avibactam resistance is mediated by bla_KPC-3 mutations, which restore carbapenem susceptibility. We subjected Klebsiella pneumoniae isolates with different bla_KPC-3 mutations (n = 5) or wild-type bla_KPC-3 (n = 2) to serial passages with meropenem. The meropenem MIC against each isolate increased. Mutations in the ompK36 porin gene evolved in 5 isolates. Among isolates with D179Y substitutions in KPC-3, bla_KPC-3 mutations reverted to wild type, were replaced by new mutations, or were retained. Carbapenem treatment of ceftazidime-avibactam-resistant K. pneumoniae infections may select for carbapenem resistance.

TEXT

Ceftazidime-avibactam is a novel β-lactam/β-lactamase inhibitor that is active against carbapenem-resistant Enterobacteriaceae (CRE) expressing Klebsiella pneumoniae carbapenemase (KPC). The drug is endorsed as front-line therapy for infection caused by KPC-producing Enterobacteriaceae due to its activity in vitro, its favorable safety profile, and the paucity of other active agents. We recently reported the first 3 cases of ceftazidime-avibactam resistance to emerge among KPC-producing K. pneumoniae isolates during treatment (1). We demonstrated that ceftazidime-avibactam resistance was caused by plasmid-borne bla_KPC-3 mutations (2), which also restored carbapenem susceptibility in certain isolates. In this study, we determined the stability of restored meropenem susceptibility among ceftazidime-avibactam-resistant isolates during in vitro passage experiments.

Representative K. pneumoniae isolates with wild-type bla_KPC-3 or with different bla_KPC-3 mutations were selected from 4 patients in whom ceftazidime-avibactam resistance had emerged. Genomic DNA was prepared from overnight cultures using a Wizard genomic DNA purification kit (Promega, Madison, WI). Genome libraries were prepared using a Nextera DNA library preparation kit and sequenced using the Illumina NextSeq platform (with paired-end reads of 150 bp). Analysis was performed as previously reported (2). In total, 7 isolates were tested (2 isolates with wild-type bla_KPC-3 and 5 isolates with different bla_KPC-3 mutations) (Table 1). bla_KPC genes were carried on an IncFIA pBK30683-like plasmid. All isolates harbored genes encoding SHV-11, TEM-1, and OXA-9 β-lactamases, a mutated ompK35 gene encoding a truncated porin (premature stop codon at amino acid position 89), and a wild-type ompK36 porin gene. Phylogenetic analysis of whole-genome sequences (WGS) demonstrated that isolates clustered tightly within a distinct sequence type 258 (ST258) sublineage, which was previously shown to include longitudinal K. pneumoniae isolates from patients in whom ceftazidime-avibactam resistance had emerged (2).

TABLE 1.

Characteristics of selected K. pneumoniae isolates harboring wild-type bla_KPC-3 or bla_KPC-3 mutations

Patient and isolate	KPC variant(s)	MIC (μg/ml) for:
		Ceftazidime-avibactam		Meropenem
		Baseline	After serial passagea	Baseline	After serial passagea
3-A	KPC-3	2	1	32	16
4-A	KPC-3	1	1	16	16
2-B	V240G	32	16	8	4
2-D	T243A	4	4	4	4
3-B	D179Y	128	128	0.25	0.125
4-C	A177E, D179Y	128	128	0.25	0.25
1-B	D179Y, T243M	256	128	0.5	0.25

^aIn the absence of antibiotics.

Ceftazidime-avibactam and meropenem MICs were determined by broth microdilution according to CLSI guidelines (3). Avibactam (kindly provided by AstraZeneca) was tested at a fixed concentration of 4 μg/ml in combination with ceftazidime (4). Baseline MICs of ceftazidime-avibactam and meropenem ranged from 1 to 256 and 0.25 to 32 μg/ml, respectively (Table 1). In our first passage experiments, 3 colonies from each isolate (i.e., replicate isolates) were inoculated into separate 3-ml volumes of cation-adjusted Mueller-Hinton broth and incubated at 37°C with shaking for 24 h. The resulting suspension was diluted 1:200 into fresh medium and incubated for 24 h. The procedure was repeated daily for 42 consecutive days. MICs were determined weekly. Following serial passage, MICs for each isolate were not significantly changed compared to the MIC for the respective parent isolate (Table 1), and the bla_KPC-3 genotypes remained unchanged.

Next, we repeated serial passage experiments in the presence of meropenem at a concentration of 0.25× MIC for each isolate. Meropenem concentrations were adjusted according to the results of weekly MIC determinations. Experiments were continued until high-level meropenem resistance (MIC ≥ 64 μg/ml) was documented in consecutive weeks or 42 days of passage were completed. In contrast to passage results in the absence of antibiotics, meropenem MICs against all isolates increased significantly following multistep passage in the presence of the drug (Table 2). At the end of meropenem passage, bla_KPC-3 and porin genes ompK35 and ompK36 from replicates were characterized by PCR and DNA sequencing (5). Nucleotide sequences of replicates were compared with those of the parent isolates. Among K. pneumoniae isolates harboring wild-type bla_KPC-3 (parent isolates 3-A and 4-A), meropenem MICs increased from 16 to 32 μg/ml at baseline to >64 μg/ml within 1 week of passage, and new insertions arose within the ompK36 promoter (IS5). For parent isolates with V240G (isolate 2-B) and T243A (isolate 2-D) KPC-3 variants, MICs increased >8-fold from baseline within 1 and 3 weeks, respectively; however, ompK36 remained wild type. Ceftazidime-avibactam MICs and bla_KPC genes encoding wild-type KPC-3, or V240G and T243A variants were unchanged following passage with meropenem.

TABLE 2.

Results of multistep serial passage with meropenem against K. pneumoniae isolates harboring wild-type bla_KPC-3 or bla_KPC-3 mutations

Isolate	Starting KPC variant	Replicate no.	Meropenem MIC (μg/ml) ata:							Ending KPC variant	ompK36 variantb	Ceftazidime-avibactam MIC (μg/ml) at end of protocol (change from baseline)c
			Baseline	Wk 1	Wk 2	Wk 3	Wk 4	Wk 5	Wk 6
3-A	KPC-3	1	32	>64	>64	—	—	—	—	KPC-3	IS5 ins	4 (↔)
		2	32	>64	>64	—	—	—	—	KPC-3	—	4 (↔)
		3	32	>64	>64	—	—	—	—	KPC-3	—	4 (↔)
4-A	KPC-3	1	16	>64	>64	—	—	—	—	KPC-3	IS5 ins	4 (↔)
		2	16	>64	>64	—	—	—	—	KPC-3	—	2 (↔)
		3	16	>64	>64	—	—	—	—	KPC-3	—	4 (↔)
2-B	V240G	1	8	>64	>64	—	—	—	—	V240G	WT	32 (↔)
		2	8	>64	>64	—	—	—	—	V240G	—	16 (↔)
		3	8	>64	>64	—	—	—	—	V240G	—	16 (↔)
2-D	T243A	1	4	16	16	>64	>64	—	—	T243A	WT	8 (↔)
		2	4	64	32	>64	>64	—	—	T243A	WT	16 (↔)
		3	4	64	32	>64	>64	—	—	T243A	WT	8 (↔)
3-B	D179Y	1	0.25	8	16	>64	>64	—	—	KPC-3	IS5 ins	128 (↔)
		2	0.25	4	16	>64	>64	—	—	T243A	F362D	16 (↓ 8-fold)
		3	0.25	8	16	>64	>64	—	—	T243A	WT	128 (↔)
4-C	A177E, D179Y	1	0.25	16	16	>64	>64	—	—	KPC-3	IS5 ins	2 (↓ 64-fold)
		2	0.25	0.25	16	>64	>64	—	—	T243A	WT	64 (↔)
		3	0.25	0.5	8	32	>64	—	—	A177E	WT	2 (↓ 64-fold)
1-B	D179Y, T243 M	1	0.5	8	8	16	16	32	64	D179Y, T243 M	Deletion	512 (↔)
		2	0.5	8	8	16	64	64	>64	T243A	IS5 ins	128 (↔)
		3	0.5	8	16	16	32	32	32	D179Y, T243 M	IS5 ins	512 (↔)

^aMICs were determined weekly and compared to those at baseline. Serial passage was stopped once the median MIC was ≥64 μg/ml for 2 consecutive weeks or after 42 days.

^bins, insertion; WT, wild type.

^cMICs within two 2-fold dilutions were considered unchanged (↔). —, not tested.

Among K. pneumoniae isolates with D179Y (parent isolate 3-B) and A177E/D179Y (parent isolate 4-C) KPC-3 variants, a stepwise increase in meropenem MICs was noted. Baseline meropenem MICs (0.25 μg/ml) were increased by ≥32-fold within 2 weeks and by ≥256-fold within 4 weeks. ompK36 promoter insertions (IS5, n = 2) or a mutation (glycine to aspartic acid at position 362 [G362D], n = 1) within ompK36 were detected in 50% (3/6) of the isolates. Following passage with meropenem, bla_KPC genes encoding D179Y and A177E/D179Y variants were no longer evident. Rather, genes encoding wild-type KPC-3, a T243A variant, or a variant with an A177E substitution alone were detected (Table 2). Ceftazidime-avibactam MICs were unchanged against 50% (3/6) of the isolates and reduced by ≥8-fold against the other 50% (3/6).

Finally, among K. pneumoniae isolates with bla_KPC-3 encoding the D179Y/T243M variant (parent isolate 1-B), meropenem MICs increased ≥16-fold within 1 week. In the subsequent 5 weeks, MICs increased 2- to 4-fold. ompK36 was deleted in 1 replicate and IS5 insertions were detected in the other 2 replicates. In 2 of 3 replicates, D179Y/T243M substitutions were still present following meropenem passage. In the other replicate, the D179Y/T243M variant was replaced by a T243A variant. Ceftazidime-avibactam MICs against each replicate were unchanged from baseline.

In this study, we determined that serial passage of ceftazidime-avibactam-resistant, meropenem-susceptible K. pneumoniae isolates in medium containing a sublethal meropenem concentration selected for high-level meropenem resistance and generated new bla_KPC and ompK36 porin gene mutations. Moreover, the vast majority of isolates remained resistant to ceftazidime-avibactam following passage. These data attest to the genomic plasticity of ST258 K. pneumoniae strains (2, 6) and bring into question the use of carbapenem monotherapy against ceftazidime-avibactam-resistant, KPC-producing Enterobacteriaceae infections. Follow-up studies are urgently needed to establish the precise role of carbapenems, alone or in combination, in treating such infections and the durability of restored carbapenem susceptibility in the clinical setting.

Several patterns of evolution were noted following meropenem passage, and these patterns suggest possible resistance mechanisms. Among K. pneumoniae isolates with wild-type bla_KPC-3, higher-level meropenem resistance was associated with ompK36 mutations, including an IS5 promoter insertion. We previously reported that IS5 promoter insertions and other major ompK36 mutations in KPC-producing ST258 K. pneumoniae strains result in extremely high carbapenem MICs and attenuate responses to carbapenem-containing combinations in vitro and in patients (5, 7, 8). ompK36 mutations were also found among K. pneumoniae isolates with different bla_KPC-3 mutations, which exhibited high-level resistance to both meropenem and ceftazidime-avibactam after passage. Such isolates, if they were to emerge clinically, would be effectively untreatable with currently available β-lactams. The clinical significance of ompK36 mutations observed in this study requires further investigation. Porin-deficient K. pneumoniae isolates were attenuated for virulence in murine peritonitis models (9, 10), suggesting that the clinical impact of enhanced resistance may be somewhat mitigated by loss of fitness in vivo.

Several isolates that carried bla_KPC-3 mutations but no ompK36 mutations likewise developed carbapenem resistance or exhibited more pronounced resistance. Therefore, other carbapenem resistance mechanisms must have evolved. Among isolates with D179Y substitutions in KPC-3, meropenem selection resulted in a reversion to wild-type KPC-3 or, in some cases, replacement of baseline mutations with an A177E or T243A substitution. We previously demonstrated that substitutions at T243A had less impact on carbapenem or ceftazidime-avibactam MICs than did D179Y (2). As such, the high-level ceftazidime-avibactam MICs exhibited here by isolates expressing KPC-3 or an A177E or T243A variant were likely caused by as yet unidentified resistance mechanisms. Future studies to identify previously unrecognized determinants of carbapenem and ceftazidime-avibactam resistance are warranted.

ompK36 and bla_KPC adaption are well-recognized evolutionary consequences of β-lactam exposure (9, 11–13). Our finding that ompK36 mutations emerged under meropenem selection in vitro is consistent with a previous report of ompK36 loss in KPC-producing K. pneumoniae within 4 to 8 h of exposure to bactericidal imipenem concentrations (11). Porin loss in this earlier study was associated with induction of high-level imipenem resistance (11). Likewise, our bla_KPC data are in keeping with results from numerous studies demonstrating that KPC mutations alter enzyme activity (12, 13). The D179Y KPC-3 substitution alone or in combination with other substitutions, such as T243M, is the most common mutation among clinical ceftazidime-avibactam-resistant K. pneumoniae isolates at our center (1, 2) and in resistant Enterobacteriaceae isolates selected by in vitro drug exposure (14). Substitutions in the KPC Ω-loop (amino acid positions 165–179) enhance affinity for ceftazidime, which is postulated to prevent subsequent binding of avibactam (15). It is plausible that the D179Y variant has reduced hydrolytic activity against carbapenems (12), a hypothesis we will test in future enzyme kinetics studies. This mechanism of restored carbapenem susceptibility would differ from that previously reported for a KPC-3-producing K. pneumoniae isolate in which plasmid rearrangements resulted in deletion of the Tn4401::bla_KPC-3 transposon (16).

Thus far, ceftazidime-avibactam resistance has emerged in ∼10% of patients with KPC-producing K. pneumoniae infection at our center who were treated with the drug; bla_KPC-3 mutations developed within 10 to 19 days of ceftazidime-avibactam exposure (1). Here, we confirm that ceftazidime-avibactam resistance, meropenem susceptibility, and bla_KPC-3 mutation-bearing plasmids were maintained during serial passage in the absence of selection pressure in vitro. Therefore, clinicians should be aware that ceftazidime-avibactam resistance may be detected after the drug is discontinued in patients. Likewise, clinicians should understand that ceftazidime-avibactam-resistant K. pneumoniae isolates bearing bla_KPC mutations may be identified as meropenem-susceptible, extended spectrum β-lactamase (ESBL)-producing K. pneumoniae by clinical microbiology laboratories, if carbapenemase screening is triggered by elevated carbapenem MICs. As technologies evolve, targeted sequencing of bla_KPC and WGS of isolates may prove to be powerful tools for rapidly identifying loss or restored activity of ceftazidime-avibactam and carbapenems, respectively.

In conclusion, our data suggest that carbapenem monotherapy in patients with ceftazidime-avibactam-resistant, KPC-3-variant K. pneumoniae infections may select for strains with high-level resistance to these two agents. It should be noted that meropenem resistance here was selected in vitro by sublethal meropenem concentrations. The impact of clinically relevant meropenem concentrations over a full treatment course merits further investigation. Since KPC Ω-loop substitutions exert ceftazidime-related effects, they are not thought to confer avibactam resistance, per se (14, 15). Therefore, avibactam and other diazabicyclooctanes may be useful against ceftazidime-avibactam-resistant KPC-producing Enterobacteriaceae in combination with β-lactams other than ceftazidime. Combinations currently in development include aztreonam-avibactam, ceftaroline-avibactam, imipenem-relebactam, and meropenem-vaborbactam. It is imperative that resistance mechanisms to new agents be identified expeditiously, rational strategies for employing these agents in the clinical setting be devised, and dosing regimens that optimize pharmacokinetic-pharmacodynamic target attainment be validated.