Emergence of Ceftazidime-Avibactam Resistance Due to Plasmid-Borne bla_KPC-3 Mutations during Treatment of Carbapenem-Resistant Klebsiella pneumoniae Infections

ABSTRACT

Ceftazidime-avibactam is a novel β-lactam/β-lactamase inhibitor with activity against carbapenem-resistant Enterobacteriaceae (CRE) that produce Klebsiella pneumoniae carbapenemase (KPC). We report the first cases of ceftazidime-avibactam resistance to develop during treatment of CRE infections and identify resistance mechanisms. Ceftazidime-avibactam-resistant K. pneumoniae emerged in three patients after ceftazidime-avibactam treatment for 10 to 19 days. Whole-genome sequencing (WGS) of longitudinal ceftazidime-avibactam-susceptible and -resistant K. pneumoniae isolates was used to identify potential resistance mechanisms. WGS identified mutations in plasmid-borne bla_KPC-3, which were not present in baseline isolates. bla_KPC-3 mutations emerged independently in isolates of a novel sequence type 258 sublineage and resulted in variant KPC-3 enzymes. The mutations were validated as resistance determinants by measuring MICs of ceftazidime-avibactam and other agents following targeted gene disruption in K. pneumoniae, plasmid transfer, and bla_KPC cloning into competent Escherichia coli. In rank order, the impact of KPC-3 variants on ceftazidime-avibactam MICs was as follows: D179Y/T243M double substitution > D179Y > V240G. Remarkably, mutations reduced meropenem MICs ≥4-fold from baseline, restoring susceptibility in K. pneumoniae from two patients. Cefepime and ceftriaxone MICs were also reduced ≥4-fold against D179Y/T243M and D179Y variant isolates, but susceptibility was not restored. Reverse transcription-PCR revealed that expression of bla_KPC-3 encoding D179Y/T243M and D179Y variants was diminished compared to bla_KPC-3 expression in baseline isolates. In conclusion, the development of resistance-conferring bla_KPC-3 mutations in K. pneumoniae within 10 to 19 days of ceftazidime-avibactam exposure is troubling, but clinical impact may be ameliorated if carbapenem susceptibility is restored in certain isolates.

INTRODUCTION

Carbapenem-resistant Enterobacteriaceae (CRE) have emerged worldwide as causes of infections that are resistant to front-line antibiotics (1). Mortality rates among patients with serious CRE infections are as high as 70% (2–4). The most common determinants of carbapenem resistance among Enterobacteriaceae in the United States are Klebsiella pneumoniae carbapenemases (KPCs), especially KPC-2 and KPC-3. Treatment of CRE infections has relied heavily upon salvage antibiotics like colistin, aminoglycosides, and tigecycline, which are limited by suboptimal pharmacokinetics, toxicity, and/or emergence of resistance (5, 6).

Avibactam is a novel β-lactamase inhibitor that inactivates KPCs, OXA-48 carbapenemases, AmpC, and extended-spectrum β-lactamases (ESBLs), but not metallo-β-lactamases such as NDM, VIM, or IMP (7). The agent, by itself, is not active against KPC-producing Enterobacteriaceae (KPC-Enterobacteriaceae). In combination, however, avibactam restores ceftazidime activity against KPC-Enterobacteriaceae in vitro (8). Ceftazidime-avibactam was approved in 2015 by the U.S. Food and Drug Administration for treatment of complicated intra-abdominal infections (in conjunction with metronidazole) and urinary tract infections (9–11). Data against KPC-Enterobacteriaceae infections are limited, but ceftazidime-avibactam is endorsed as front-line therapy due to its activity in vitro, its safety profile, and the lack of effective alternative regimens (7).

In a recent study, we described our experience at the University of Pittsburgh Medical Center (UPMC) with ceftazidime-avibactam in treating 37 patients with CRE infections (12). Clinical success and survival at 30 days were 59% and 76%, respectively, rates that were comparable to those given in previous reports of CRE-infected patients treated with ≥2 in vitro-active agents (4). Ceftazidime-avibactam was well tolerated, with a 10% rate of acute kidney injury after ≥7 days of treatment, which was considerably lower than the ∼30% rate that we observed with colistin- or aminoglycoside-containing regimens (13, 14). As a counterpoint to these promising findings, ceftazidime-avibactam resistance emerged in three patients during treatment of KPC-producing K. pneumoniae infections. To our knowledge, these were the first descriptions of resistance developing during ceftazidime-avibactam treatment. To date, resistance mechanisms have not been described in KPC-producing clinical isolates.

Our objectives in this study were to provide detailed case reports of the emergence of ceftazidime-avibactam resistance, identify and validate resistance mechanisms using genomic and molecular genetic approaches, and study the impact of ceftazidime-avibactam resistance mechanisms on susceptibility to meropenem and other β-lactam agents.

(These data were presented at ID Week 2016, October 2016, New Orleans, LA.)

RESULTS

Case reports.

Clinical and microbiologic details, timelines, and antimicrobial therapy are summarized in Fig. 1 and Table 1. Baseline isolates from each patient (isolates 1-A, 2-A, and 3-A) were meropenem resistant and ceftazidime-avibactam susceptible. Ceftazidime-avibactam was dosed at 2.5 g intravenously every 8 h (adjusted based on renal function) (9).

FIG 1.

Time courses of infection and treatment among patients in whom ceftazidime-avibactam-resistant K. pneumoniae emerged. Black and shaded bars represent duration of treatment with respective antibiotics. *, interpretation of K. pneumoniae phenotype, as reported by the clinical microbiology laboratory. Abbreviations: CRE, carbapenem-resistant Enterobacteriaceae; CTB, ceased to breathe; ESBL, extended-spectrum β-lactamase; IAI, intra-abdominal infection; PNA, pneumonia; UTI, urinary tract infection.

TABLE 1.

K. pneumoniae isolates recovered from patients treated with ceftazidime-avibactam^a

Patient-isolate ID	Days from admission to culture (source)	Ceftazidime-avibactam exposure at time of culture (days)	Ceftazidime-avibactam	MIC (μg/ml)b							KPC-3 variantc	Relative fold change in blaKPC expression by qRT-PCRd
				Ceftazidime	Meropenem	Ertapenem	Piperacillin-tazobactam	Cefepime	Ceftriaxone	Aztreonam
1-A	6 (sputum)	0	2 (S)	512	128	256	>512	>128	>128	>256	Wild-type	Reference
1-B	20 (sputum)	10	256	>512	0.5 (S)	1	>512	16	32	8	D179Y, T243 M	↓ 2.64-fold
1-C	42 (sputum)	24	256	>512	0.25 (S)	2	>512	16	32	16	D179Y, T243 M	NPe
2-A	7 (abscess)	0	4 (S)	256	32	8	>512	>128	>128	>256	Wild-type	Reference
2-B	48 (urine)	19	32	>512	8	8	>512	>128	>128	>256	V240G	<2-fold
2-C	48 (urine)	19	>256	>512	4	16	>512	16	4	8	D179Y	↓ 10.45-fold
2-D	78 (urine)	19	4 (S)	256	4	8	>512	>128	>128	>256	T243A	<2-fold
3-A	6 (BALF)	0	2 (S)	256	32	8	>512	>128	>128	>256	Wild-type	Reference
3-B	64 (BALF)	15	128	512	0.25 (S)	1	256	4	16	4 (S)	D179Y	↓ 2.05-fold
3-C	82 (BALF)	15	64	512	0.125 (S)	0.5	256	8	16	4 (S)	D179Y	NP

^aBALF, bronchoalveolar lavage fluid; S, susceptible, based on CLSI interpretive criteria (susceptibility breakpoints: ceftazidime-avibactam, ≤8 μg/ml; ceftazidime, ≤4 μg/ml; meropenem, ≤1 μg/ml; piperacillin-tazobactam, ≤16 μg/ml; cefepime, ≤2 μg/ml; ceftriaxone, ≤1 μg/ml; aztreonam, ≤4 μg/ml). Boldface rows represent baseline isolate from each patient.

^bMICs within the susceptible range are indicated by “(S)”; all other MICs indicated resistance.

^cWhen relevant, amino acid substitutions within KPC-3 are listed.

^dRT-PCR was performed for a representative isolate from each patient that expressed a given KPC-3 variant. Data show differences in gene expression relative to the corresponding baseline isolate (reference).

^eNP, RT-PCR not performed.

Patient 1, a double-lung transplant recipient in her 40s, was diagnosed with carbapenem-resistant K. pneumoniae pneumonia and urinary tract infection (isolate 1-A, recovered from sputum; ceftazidime-avibactam MIC, 2 μg/ml; meropenem MIC, 128 μg/ml) in November 2015. She was treated with 10 days of ceftazidime-avibactam. Four days after completing treatment, ceftazidime-avibactam and gentamicin were initiated for recurrent pneumonia. Sputum culture (isolate 1-B) grew meropenem-susceptible K. pneumoniae (MIC = 0.5 μg/ml), which was reported as ESBL producing. Ceftazidime-avibactam was discontinued after 14 days, and meropenem was administered because meropenem-susceptible (MIC = 0.25 μg/ml), ESBL K. pneumoniae was still identified in sputum culture (isolate 1-C). Both meropenem-susceptible isolates were ceftazidime-avibactam resistant (MICs = 256 μg/ml). Meropenem was given for 17 days (alone and then combined with gentamicin), but the patient died of respiratory failure. There were no epidemiological links between patient 1 and patient 2 or 3.

Patient 2, a woman in her 50s, was diagnosed with a subphrenic abscess due to carbapenem-resistant K. pneumoniae (isolate 2-A; ceftazidime-avibactam MIC, 4 μg/ml; meropenem MIC, 32 μg/ml) in January 2016. She was successfully treated with abscess drainage and 19 days of ceftazidime-avibactam. Subsequent hospitalization was complicated by recurrent fevers and worsening of comorbid conditions. Work-up during these febrile episodes revealed urinary tract colonization with three different carbapenem-resistant K. pneumoniae isolates, which was not treated (isolates 2-B, -C, and -D; meropenem MICs, 4 to 8 μg/ml). Two colonizing isolates (isolates 2-B and -C) were ceftazidime-avibactam resistant (MICs, 32 and >256 μg/ml). The patient was ultimately discharged.

Patient 3, a man in his 70s who had undergone esophagectomy for cancer, was diagnosed with pneumonia due to carbapenem-resistant K. pneumoniae (isolate 3-A; ceftazidime-avibactam MIC, 2 μg/ml; meropenem MIC, 32 μg/ml) in January 2016. He was treated with ceftazidime-avibactam for 15 days. Thereafter, meropenem-susceptible, ESBL-producing K. pneumoniae isolates were identified as a respiratory tract colonizer or cause of recurrent pneumonia (isolates 3-B, 3-C; MICs, 0.25 and 0.125 μg/ml, respectively). Both isolates were ceftazidime-avibactam resistant (MICs, 64 and 128 μg/ml). Recurrent pneumonia was treated successfully with meropenem plus colistin for 14 days.

Characterization of K. pneumoniae isolates.

All 10 K. pneumoniae isolates were sequence type (ST) 258, clade II (cps-2, wzi allele 154) clones (Table 2) (15). Whole-genome sequences of these isolates were compared to those of 23 ST258, clade II isolates from our center (2010 to 2011) and others (Fig. 2). A total of 892 core single nucleotide polymorphisms (SNPs) were identified across the 33 isolates; 152 SNPs that were located at repetitive or prophage regions were excluded from phylogenetic analyses. An average of 91 SNPs (range, 1 to 214) were identified in pairwise comparisons of isolates, which is in agreement with our previous data that ST258, clade II isolates differed by ∼82 SNPs (range, 2 to 231) (15). The 10 isolates from this study clustered tightly with, but were divergent from, the other ST258, clade II K. pneumoniae isolates (Fig. 2 and 3; see also Table S1 in the supplemental material).

TABLE 2.

Molecular characteristics of K. pneumoniae isolates from the present study^a

Patient no.	Strain ID	OmpK36	KPC	blaKPC	Other resistance genes	Plasmid InC
1	1-A	IS5 promoter ins	KPC-3	blaKPC-3	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib,strAB, sul2, dfrA14	ColpVC,Col(MG828), pBK30683 repA, repB, pNJST258N3
1	1-B	WT	AA179D→Y, 243T→M	532G→T, 725C→T	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	ColpVC,Col(MG828), pBK30683 repA, repB, pNJST258N3, L/M
1	1-C	AA288stop	AA179D→Y, 243T→M	532G→T, 725C→T	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	ColpVC,Col(MG828), pBK30683 repA, repB, pNJST258N3, L/M
2	2-A	WT	KPC-3	blaKPC-3	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	pBK30683 repB, pNJST258N3
2	2-B	WT	AA240V→G (KPC-8)	716T→G	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	pBK30683 repB, pNJST258N3
2	2-C	IS5 ins 827 bp	AA179D→Y	532G→T	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	pBK30683 repB, pNJST258N3
2	2-D	WT	AA243T→A	724G→A	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	pBK30683 repB, pNJST258N3
3	3-A	WT	KPC-3	blaKPC-3	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	pBK30683 repB, pNJST258N3
3	3-B	WT	AA179D→Y	532G→T	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	pBK30683 repB, pNJST258N3
3	3-C	WT	AA179D→Y	532G→T	blaTEM-1, blaSHV-11, blaOXA-9, aadA1, aac(6′)-Ib, strAB, sul2, dfrA14	pBK30683 repB, pNJST258N3

^aAbbreviations: InC, incompatibility group; KPC, Klebsiella pneumoniae carbapenemase; WT, wild type. All isolates had the following features: sequence type (ST) 258; wzi allele 154; clade cps-2; OmpK35, AA89stop.

FIG 2.

Phylogenetic comparison of ST258, clade II K. pneumoniae isolates from our center and others in the United States. The phylogenetic tree, based on whole-genome SNP analysis, was generated with the use of the maximum-likelihood optimality criterion. Branch lengths are proportional to the number of evolutionary changes, and all nodes were supported by 100% bootstrap. Ten K. pneumoniae isolates from the present study and 22 isolates collected from hospitals in New York, New Jersey, Pennsylvania, Maryland, and Michigan in previous studies were included (15, 37, 39). Isolates from our medical center are presented in yellow ovals. Isolates in cluster a are from the present study, and isolates in cluster b were collected from our center in 2010 to 2011.

FIG 3.

SNP matrix of the 10 K. pneumoniae isolates from the present study. Numbers of SNPs for each pairwise comparison of isolates are shown.

All isolates had genes encoding TEM-1, SHV-11, and OXA-9 β-lactamases; none had genes encoding NDM, other metallo-β-lactamases, or OXA-48-like carbapenemases. Baseline isolates carried bla_KPC-3. Subsequent isolates carried mutant bla_KPC-3 encoding variant KPC-3 enzymes (Table 2). The most common KPC-3 variant featured a tyrosine for aspartic acid substitution at Ambler amino acid position 179 (D179Y), either alone (isolates 2-C, 3-B, and 3-C) or in combination with a methionine-for-threonine substitution at position 243 (D179Y/T243M; isolates 1-B and 1-C). Alanine-for-threonine and glycine-for-valine substitutions were observed at positions 243 (T243A) and 240 (V240G), respectively, in other isolates from patient 2; the V240G variant corresponds to KPC-8 (16). Each wild-type and mutant bla_KPC-3 gene was in a ΔTn1331-Tn4401d nested transposon element, which was found within a common ∼40-kb contig. This contig is virtually identical to the corresponding region on IncFIA plasmid pBK30683, a predominant bla_KPC-3-bearing plasmid in New York and New Jersey hospitals (17).

K. pneumoniae isolates with wild-type bla_KPC-3 were ceftazidime-avibactam susceptible and meropenem resistant. Ceftazidime-avibactam MICs were increased ≥4-fold against isolates with mutant bla_KPC-3 encoding D179Y/T243M, D179Y, and V240G variants compared to corresponding baseline isolates (Table 1); increased MICs conferred clinical drug resistance. Meropenem MICs against ceftazidime-avibactam-resistant isolates were reduced ≥4-fold, which restored meropenem susceptibility to isolates from patients 1 and 3. Cefepime, ceftriaxone, and aztreonam MICs were reduced ≥4-fold against D179Y/T243M and D179Y variant isolates. All isolates remained resistant to cefepime and ceftriaxone. The expression of bla_KPC-3 encoding D179Y/T243M and D179Y variants was reduced compared to bla_KPC-3 expression in corresponding baseline isolates. No mutations were evident within promoter sequences of mutant bla_KPC-3 compared to wild-type bla_KPC-3.

Validation of bla_KPC-3 mutations.

Genes encoding D179Y/T243M, D179Y, and V240G KPC-3 variants were validated as causes of ceftazidime-avibactam resistance through three lines of evidence (Table 3). First, deletion of mutant bla_KPC-3 in ceftazidime-avibactam-resistant K. pneumoniae restored drug susceptibility. Second, laboratory transfer of mutant bla_KPC-3-containing plasmids from K. pneumoniae to Escherichia coli DH10B resulted in ≥4-fold increases in ceftazidime-avibactam MICs, compared to control, wild-type bla_KPC-3 transformants. Third, ceftazidime-avibactam MICs also increased against E. coli DH5α harboring plasmid constructs with cloned mutant bla_KPC-3, compared to E. coli DH5α with wild-type bla_KPC-3 constructs. For mutant bla_KPC-3 encoding D179/T243M, D179Y, and V240G variants, ceftazidime-avibactam MICs were increased ≥128-fold, ≥16-fold, and ≥4-fold, respectively. Mutant bla_KPC-3 encoding D179/T243M and D179Y variants conferred ≥4-fold reductions in meropenem, cefepime, ceftriaxone, and aztreonam MICs against E. coli DH5α compared to wild-type bla_KPC-3.

TABLE 3.

MICs against bla_KPC-3 deletion mutants and E. coli transformants^a

KPC-3 variant	Description	MIC range (μg/ml)
		Ceftazidime-avibactam	Ceftazidime	Meropenem	Piperacillin-tazobactam	Cefepime	Ceftriaxone	Aztreonam
Wild-type	Clinical isolates 1-A, 2-A, 3-A	2–4	256–512	32–128	>512	>128	>128	>256
	Clinical isolate with blaKPC deletionb	0.5	256	≤0.125	16–32	8	32	>256
	E. coli with blaKPC-containing plasmid from clinical isolatec	1	128	2	256	32	>128	>256
	E. coli with pET30a plasmid into which blaKPC was clonedd	≤0.25–0.5	64–128	2–8	>512	>128	>128	>256
D179Y, T243 M	Clinical isolates 1-B, 1-C	256	>512	0.25–0.5	>512	16	32	8
	Clinical isolate with blaKPC deletionb	≤0.25	128	≤0.125	128	8	4	2
	E. coli with blaKPC-containing plasmid from clinical isolatec	64	256	≤0.125	64	8	32	64
	E. coli with pET30a plasmid into which blaKPC was clonedd	64	≥512	≤0.125	2–4	2–8	4–8	2–4
D179Y	Clinical isolates 2-C, 3-B, 3-C	64 to >256	≥512	0.125–4	256 to >512	4–16	4–16	4–8
	Clinical isolate with blaKPC deletionb	≤0.25	256	≤0.125	32–128	16	32	>256
	E. coli with blaKPC-containing plasmid from clinical isolatec	16	128	≤0.125	16	4	4	2
	E. coli with pET30a plasmid into which blaKPC was clonedd	8	64	≤0.125	1	1	4	2
V240G	Clinical isolate 2-B	32	>512	8	>512	>128	>128	>256
	Clinical isolate with blaKPC deletionb	0.5	128	≤0.125	32	8	32	>256
	E. coli with blaKPC-containing plasmid from clinical isolatec	4	512	1	256	32	>128	>256
	E. coli with pET30a plasmid into which blaKPC was clonedd	2	256	1	128	128	>128	>256

^aBoldface rows represent baseline carbapenem-resistant K. pneumoniae isolates.

^bbla_KPC deletions were performed for a representative clinical isolate with each KPC-3 variant.

^cIncFIA pBK30683-like plasmids were extracted from representative clinical isolates and transferred into E. coli DH10B.

^dbla_KPC genes from representative clinical isolates were cloned into a pET30a expression vector plasmid, which was transferred into E. coli DH5α. Ceftazidime-avibactam, ceftazidime, and meropenem MICs against E. coli DH5α carrying pET30a without bla_KPC genes were ≤0.25, ≤0.5, and ≤0.125 μg/ml, respectively.

DISCUSSION

Ceftazidime-avibactam resistance due to plasmid-borne bla_KPC-3 mutations appeared within 10 to 19 days of treatment in three patients with carbapenem-resistant K. pneumoniae infections. These findings are particularly concerning because K. pneumoniae isolates were all ST258, the most successful CRE worldwide (18), and ceftazidime-avibactam resistance arose through a variety of bla_KPC-3 mutations. For some ceftazidime-avibactam-resistant isolates, bla_KPC-3 mutations restored meropenem susceptibility, suggesting that their clinical impact may be ameliorated. However, carbapenem susceptibility was not restored against all ceftazidime-avibactam-resistant isolates, and the stability of the phenotype is unknown. MICs of other β-lactams were also reduced against bla_KPC-3 mutant K. pneumoniae. Nevertheless, isolates generally remained resistant to these drugs, highlighting that other plasmid-mediated β-lactamases contribute to overall resistance phenotypes. Taken together, the data raise questions of whether bla_KPC mutations that confer ceftazidime-avibactam resistance will emerge in CRE at other centers, mutant bla_KPC will disseminate through horizontal gene transfer, and carbapenems will serve as effective salvage agents in treating at least some infections by such pathogens.

In rank order, the impact of KPC-3 variants on ceftazidime-avibactam MICs was as follows: D179Y/T243M > D179 > V240G. D179Y/T243M and D179Y variants caused comparable reductions in meropenem MICs, which were greater than with V240G. The results are in keeping with published laboratory data. Ceftazidime-avibactam resistance and KPC-3 amino acid substitutions arose at low frequency (∼10⁻⁹) in K. pneumoniae and Enterobacter cloacae during in vitro selection (19). D179Y emerged most frequently; a proline-for-threonine substitution at position 243 (T243P) was less common (19). D179Y and T243P mutants reverted to carbapenem susceptibility. In another study, introduction of substitutions at KPC-2 amino acid positions 179 and 243 by site-directed mutagenesis conferred ceftazidime-avibactam resistance in E. coli (20). E. coli recombinants carrying KPC-8 also exhibited significant increases in ceftazidime-avibactam MICs compared to control E. coli (16).

The mechanisms by which bla_KPC-3 mutations mediate ceftazidime-avibactam resistance and reversion to carbapenem susceptibility are not defined. The KPC Ω-loop encompasses amino acid positions 165 through 179. KPC variants with Ω-loop substitutions demonstrate enhanced ceftazidime affinity, which is postulated to prevent binding of avibactam (20). Therefore, Ω-loop substitutions exert ceftazidime-related effects, but they are not thought to confer avibactam resistance per se (19, 20). KPC-3 has 9- to 30-fold-greater catalytic efficiency than KPC-2 against ceftazidime (21, 22), and we previously demonstrated that ceftazidime-avibactam MICs were significantly higher against bla_KPC-3-producing than bla_KPC-2-producing K. pneumoniae (8). These factors may predispose KPC-3-K. pneumoniae to ceftazidime-avibactam resistance (8). Stepwise mutations are well recognized to result in KPC variants with reduced hydrolytic activity against carbapenems (18). Kinetic studies are needed to determine if the KPC-3 substitutions described here attenuate carbapenemase activity. Finally, expression of mutant bla_KPC-3 encoding D179Y/T243M or D179Y variants was significantly diminished; these preliminary data suggest that transcriptional regulation may modulate meropenem and ceftazidime-avibactam susceptibility patterns.

These are the first reports of ceftazidime-avibactam resistance emerging during treatment of CRE infections. Preexisting resistance is reported in up to 2 to 3% of Enterobacteriaceae in clinical trials and large biorepositories (8, 10, 11, 23, 24). In a recent report from a large U.S. cancer center, ceftazidime-avibactam resistance was detected in 6 CRE isolates that were shown subsequently to carry NDM-1 (25). Since NDM-1 is not inhibited by avibactam, the study described baseline ceftazidime-avibactam resistance rather than the emergence of resistance as documented in the present study. Two case reports, including one from the same cancer center, described ceftazidime-avibactam resistance in non-NDM-carrying K. pneumoniae isolates recovered from untreated patients (25, 26). One isolate harbored KPC-3, and the other did not carry a carbapenemase, but mechanisms of resistance were not identified. The three cases of drug resistance presented here accounted for 30% of microbiologic failures in our early experience with ceftazidime-avibactam against CRE infections (12).

Our WGS data indicate that ceftazidime-avibactam resistance arose within a novel K. pneumoniae ST258 sublineage. It is unclear at present if this sublineage is uniquely predisposed to the development of bla_KPC-3 mutations or if mutations will be encountered at similar frequency in other ST258 populations, different STs, and diverse Enterobacteriaceae. Clearly, the 10 isolates here share a recent common ancestry. Nevertheless, resistance-conferring bla_KPC-3 mutations were more likely to have evolved independently in different isolates (and in different patients) than to have emerged in a single parent isolate that subsequently disseminated. In support of this conclusion, patient 1 did not have any epidemiological links with patients 2 and 3. In addition, KPC variants had mutations at three distinct amino acid positions; even isolates from patients 2 and 3, who resided in the same unit, exhibited distinct amino acid substitutions. Longitudinal isolates from each of the patients were also heterogeneous in their core genomes, carrying noncumulative SNPs that varied from isolate to isolate; isolates from patients 1 and 2 demonstrated a variety of ompK36 porin genotypes. Therefore, patients appeared to harbor subpopulations of K. pneumoniae that evolved through different pathways from a common ancestor rather than in a direct lineage. Taken together, the findings highlight the well-recognized genetic plasticity and speed of evolutionary changes manifested by carbapenem-resistant K. pneumoniae (27).

Due to the potential for rapid KPC spread and evolution, it is imperative that centers detect CRE isolates and infected patients in a timely fashion. Our cases illustrate the difficulties faced by clinical laboratories in this regard. Indeed, several ceftazidime-avibactam-resistant isolates were reported as meropenem-susceptible, ESBL-producing K. pneumoniae and were not identified as KPC producing. Such misidentification was inevitable since carbapenemase screening by our laboratory, as at many centers, is triggered by elevated carbapenem MICs. Routine molecular screening for bla_KPC has been advocated for rapid detection of carbapenem resistance (28). However, such assays will need to detect specific genes, since KPC variants have different antibiotic affinities and hydrolytic properties. As shown here, sequencing the bla_KPC gene directly or by next-generation genomics technologies may be necessary to identify and interpret the loss or gain of carbapenem and ceftazidime-avibactam susceptibility.

In conclusion, ceftazidime-avibactam is one of the few new antibiotics at an advanced stage of development that have anti-CRE activity. In order to preserve ceftazidime-avibactam, the medical community must define optimal dosing strategies (including the role of antibiotic combination regimens), adhere to rigorous stewardship and infection control practices, develop assays for detecting resistance and resistance determinants expeditiously, and understand resistance mechanisms. Together with reports of the recent emergence of MCR-1, the first plasmid-mediated colistin resistance determinant in Enterobacteriaceae (29), our findings attest to the urgent need for new drugs and treatment strategies against Gram-negative bacterial infections.

MATERIALS AND METHODS

Susceptibility testing.

MICs were determined by broth microdilution according to Clinical Laboratory Standard Institutes (CLSI) guidelines (30). Antimicrobial agents were purchased from the UPMC pharmacy. Avibactam (kindly provided by AztraZeneca) was tested at a fixed concentration of 4 μg/ml in combination with ceftazidime (8). Susceptibility was defined according to CLSI interpretive criteria (31). Screening for ESBL production by the UPMC clinical microbiology laboratory used standard disc diffusion methods (31). In brief, isolates that were nonsusceptible to aztreonam and/or ceftriaxone were tested against ceftazidime and cefotaxime with and without clavulanic acid. Isolates demonstrating a ≥5-mm increase in zone diameter for either agent in combination with clavulanic acid were classified as ESBL.

Whole-genome sequencing (WGS) and analysis.

Genomic DNA was prepared from overnight cultures using a Wizard Genomic DNA purification kit (Promega, Madison, WI). Genome libraries were prepared using the Nextera DNA Library Preparation kit and sequenced using the Illumina NextSeq platform (paired end reads of 150 bp). Sequences were trimmed with sickle (GitHub), followed by de novo genome assembly using SPAdes 3.8 (32). In silico multilocus sequence typing and capsular wzi typing were performed using the BIGSdb online server (http://bigsdb.pasteur.fr/klebsiella/klebsiella.html). Antimicrobial resistance genes and plasmid replicons were identified using SRST2 (33). Novel plasmid replicons from our previous studies also were included in the analysis (15, 17).

Single nucleotide polymorphisms (SNPs) and insertion-deletions (indels) were called using BWA (34) and Samtools (35). SNP and indel annotation was conducted with SnpEff (36). K. pneumoniae isolate DMC1097 (37), an ST258 clade II (cps-2) strain from Detroit, MI, was used as reference. For core SNP analysis, SNPs located at repetitive (identified by nucmer from MUMmer) or prophage (identified by PHAST) regions of the genome were excluded. SNP phylogenetic analysis was performed using RAxML 8.0.0 (38). In addition, 22 ST258 clade II isolates from New Jersey (NJST258_1, NJST258_2, 1748, 1750, 1756, 1757) (15), New York (1765, 1790, 1789, 1814, 1801, 1780, 1764, 1824) (15), Maryland (KPNIH1, accession number CP008827) (39), Michigan (DMC1097, CP011976) (37), and UPMC (1818, 1808, 1809, 1816, 1811, 1817) (15) were included in the core SNPs analysis.

qRT-PCR.

DNase-treated RNA was obtained from late-exponential-phase cultures (Mueller-Hinton broth, 37°C; RiboPure-Bacteria kit, ThermoFisher Scientific, MA). cDNA was made using qScript cDNAMix (Quanta Biosciences, MD). Quantitative reverse transcription (qRT)-PCR was performed using the Applied Biosystems 7900 system, with established bla_KPC primers and the SYBR green kit (Quanta Biosciences, MD) (40). bla_KPC expression was normalized to housekeeping gene rpoB. In each case, relative expression was calibrated against corresponding baseline isolates that were susceptible to ceftazidime-avibactam.

bla_KPC gene deletion.

bla_KPC genes in K. pneumoniae isolates were deleted using the Red/ET system (Gene Bridges, Heidelberg, Germany), according to the manufacturer's protocol with slight modifications. In brief, K. pneumoniae isolates were transformed with the pRED/ET plasmid, which contains all genes necessary for homologous recombination. A linear DNA fragment, comprising the FRT-PGK-gb2-arr3-FRT cassette (in which a kanamycin resistance gene was replaced by an arr3 rifampin resistance gene) with 50-bp homology arms targeting the upstream and downstream sequence of bla_KPC, was PCR amplified. The amplicon was electroporated into K. pneumoniae cells, resulting in replacement of bla_KPC through recombination. Transformants, selected on lysogeny broth (LB) agar containing 100 μg/ml of rifampin, were confirmed by PCR and sequencing.

Plasmid transfer.

Plasmid DNA from K. pneumoniae isolates, extracted using a Qiagen Plasmid Midi kit (Qiagen, Valencia, CA), was electroporated into Escherichia coli DH10B (Invitrogen) using a Gene Pulser II instrument (Bio-Rad Laboratories). E. coli DH10B transformants were selected on LB agar plates with 20 μg/ml ceftazidime. Transformants were screened by multiplex real-time PCR for the presence of bla_KPC genes (41) and checked by S1-PFGE (S1 nuclease digestion of plasmid DNA followed by pulsed-field gel electrophoresis) for the presence of a single bla_KPC-harboring plasmid. Mutations within bla_KPC genes were confirmed using standard PCR and sequencing.

bla_KPC-3 cloning.

bla_KPC and flanking regions (600 bp upstream, 240 bp downstream) were amplified from plasmid DNA by PCR using primers KPC3_For_KpnI (CGGGGTACCCTGCTGGTGATCGATGAGAT) and KPC3_Rev_EcoRI (CCGGAATTCTTAAATGGGGCGGCTCAAAT), which contained introduced KpnI and EcoRI restriction sites, respectively (underlined). Amplicons were digested and ligated in the antisense direction into pET30a (Novagen, Madison, WI); inserts were sequenced on both strands. Plasmids were transformed into E. coli DH5α by electroporation. Successful transfer of bla_KPC genes was confirmed by PCR and sequencing.

Accession number(s).

Sequence data have been deposited in NCBI under BioProject number PRJNA326665.