Molecular Mechanism of Polymyxin Resistance in Multidrug-Resistant Klebsiella pneumoniae and Escherichia coli Isolates from Henan Province, China: A Multicenter Study.

PURPOSE: To evaluate polymyxin-resistant Klebsiella pneumoniae and Escherichia coli prevalence and characteristics in the Henan province, China.
MATERIALS AND METHODS: A total of 2301 bacterial isolates collected at six hospitals were assessed. Their response to polymyxin was evaluated by minimum inhibitory concentration (MIC) analysis, and the mobilized colistin resistance (mcr) and carbapenemase gene were explored. Mutations on mgrB, phoPQ, pmrAB, and crrAB in polymyxin-resistant K. pneumoniae were detected by PCR. phoP, phoQ, pmrK, pmrA, pmrB, and pmrC transcriptional levels were quantified by RT-qPCR. Pulsed-field gel electrophoresis (PFGE) and multi-locus sequence typing were performed to determine the phylogenetic relationship between the polymyxin-resistant isolates.
RESULTS: Of the E. coli and K. pneumoniae isolates identified, 0.3% and 1.4% were polymyxin-resistant, respectively, with MICs of 4-64 μg/mL. All polymyxin-resistant isolates were susceptible to tigecycline. Four E. coli isolates were mcr-1-positive and one was carbapenem-resistant, carrying bla NDM-5 and mcr-1. One K. pneumoniae isolate was mcr-1-positive and nine were carbapenem-resistant (PRCRKP), carrying bla KPC-2 but not mcr-1. The five E. coli isolates belonged to four sequence types (ST2, ST132, ST632, and ST983). All PRCRKP isolates belonged to ST11. However, all 16 isolates belonged to different PFGE types with <95% genetic similarity. Insertion sequences in mgrB were detected in nine (81.8%) polymyxin-resistant K. pneumoniae samples. Colistin resistance was linked with pmrHFIJKLM operon upregulation, with phoP, phoQ, and pmrK being overexpressed in all but one of the polymyxin-resistant K. pneumoniae samples. Furthermore, 33.3% of patients carrying polymyxin-resistant isolates had previously used polymyxin, and 66.7% patients displayed good clinical outcomes.
CONCLUSION: The K. pneumoniae polymyxin resistance rate was slightly higher than that of E. coli and mcr-1 was more common in E. coli than in K. pneumoniae. Moreover, the insertion of ISkpn14 into mgrB may be the main contributor to polymyxin-resistance in K. pneumoniae in Henan.

Introduction

In recent years, antibiotic resistance has become a global public health priority. Colistin, also known as polymyxin, is one of the few therapeutic options available for the treatment of infectious diseases caused by multidrug-resistant gram-negative bacteria.1 In China, polymyxin was approved in January 2017 as an injectable drug for the treatment of bacterial infections. However, because of the increased usage of polymyxin in the clinical setting, polymyxin-resistant strains, especially those carrying the plasmid-borne mobilized colistin resistance (mcr) gene, have appeared in China and various countries worldwide.2 Moreover, intraspecies transmission of resistant isolates has already been reported.1,3

Since its discovery in southern China in late 2015,4 mcr-1 has spread to over 40 countries and regions, implying that it plays a prevalent role in the transferability of polymyxin resistance. Mcr-1-positive strains have also appeared in the Henan province, and it has been reported that mcr exists in pig-derived Escherichia coli isolates.5 Clinical E. coli isolates coproducing blaNDM and mcr-1 were previously reported by our laboratory.6 A novel conjugative mcr-8.2-bearing plasmid was identified in an almost pan-resistant hypermucoviscous Klebsiella pneumoniae ST11 isolate in Henan.7 However, overall, the report about mcr in human-derived E. coli and K. pneumoniae isolates was primarily focused outside Henan.

Except for mcr, colistin resistance in K. pneumoniae can be mediated by chromosomal mutations in genes involved in lipopolysaccharide synthesis, namely phoPQ, pmrAB, and crrA/crrB, as well as the regulatory mgrB.8–10

To better understand the epidemiological trends and characteristics of polymyxin-resistant clinical strains, these strains were screened using isolates collected at six hospitals located in the Henan province between 2018 and 2019. A total of 16 polymyxin-resistant strains were collected and their molecular resistance characteristics were analyzed. To the best of our knowledge, this is the first multicenter study to screen and investigate the molecular mechanisms of polymyxin resistance among E. coli and K. pneumoniae strains in the Henan province.

Materials and Methods

Sample Collection

Non-duplicated E. coli and K. pneumoniae strains were obtained from routine microbiological cultures of clinical samples from patients, including blood, urine, sputum, bronchoalveolar lavage fluid (BAL), bile, hydrothorax, ascites, and various other specimens. A total of 2301 strains were collected from six hospitals in the Henan province. Identification at the species level was performed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Bruker Daltonik GmbH, Bremen, Germany).

Susceptibility Testing and Minimum Inhibitory Concentrations (MICs) Determination

Susceptibility to polymyxin was screened using polymyxin B susceptibility test Strip (Antu, Zhengzhou, China). MICs higher than 2 µg/mL were confirmed by the microbroth dilution method based on the clinical break points defined by the European Committee on Antimicrobial Susceptibility Testing.11

Susceptibility of polymyxin-resistant strains to ampicillin (AMP), meropenem (MEM), imipenem (IPM), ceftazidime (CAZ), cefotaxime (CTX), cefazolin (KZ), ampicillin/sulbactam (SAM), aztreonam (ATM), cefepime (FEP), piperacillin/tazobactam (TZP), levofloxacin (LEV), amikacin (AK), gentamicin (GN), trimethoprim/sulfamethoxazole (SXT), ceftazidime/avibactam (CZA), and tigecycline (TGC) were determined using the microbroth dilution method and interpreted according to the Clinical and Laboratory Standards Institute (CLSI) guidelines.12 Escherichia coli ATCC25922 and Pseudomonas aeruginosa ATCC27853 were used as quality controls.

Multi-Locus Sequence Typing (MLST)

Polymyxin-resistant K. pneumonia and E. coli isolates were typed using MLST following the scheme established by the Pasteur Institute ( and , respectively).

Characterization of Mcr, mgrB, phoPQ, pmrAB, crrAB, and Carbapenem Resistance Genes

The modified carbapenem inactivation method (mCIM) and ethylenediaminetetraacetic acid-modified carbapenem inactivation method (eCIM), which are recommended by the CLSI, were used for phenotypic detection to confirm carbapenemase production. The presence of carbapenem resistance genes (blaVIM, blaIMP, blaKPC, blaNDM, and blaoxa-48-like) and the polymyxin resistance gene mcr1–8 in polymyxin-resistant isolates were screened by polymerase chain reaction (PCR) using the methods described previously.13–15 The full-length mgrB, pmrA, pmrB, phoP, phoQ, crrA, and crrB genes in polymyxin-resistant K. pneumoniae were amplified and sequenced. Next, the translated amino acid sequences were analyzed and compared with those of polymyxin-susceptible K. pneumoniae.

Transcriptional Analysis Real-Time Quantitative PCR

RNA extraction and transcription were carried out as previously described.16 Real-time quantitative PCR (RT-qPCR) was used to measure the expression of the phoQ, phoP, pmrK, pmrA, pmrB, and pmrC genes using the primers as previously described.16–18 Normalization was performed against the rpsL gene using the ΔΔCT method (relative) with the rpsL gene as internal control. The obtained values were then normalized against those obtained with polymyxin-susceptible strains.

Pulsed-Field Gel Electrophoresis (PFGE)

The molecular epidemiology of all polymyxin-resistant strains was determined by PFGE after total chromosomal DNA digestion with XbaI, in accordance with a previous report.19 The PFGE patterns were analyzed with the BIONUMERICS software (Applied Maths NV, Sint-Martens-Latem, Belgium) using the dice similarity coefficient. Isolates were considered as the same strain (PFGE type) if they possessed a genetic similarity of ≥ 95%.

Results

Overall Prevalence of Polymyxin-Resistant Strains

Over the course of the study, 16 (0.7%) isolates were found to be polymyxin-resistant among the 2301 identified E. coli and K. pneumoniae isolates, of which 5 and 11 were E. coli and K. pneumoniae isolates, respectively. The prevalence of polymyxin resistance in E. coli and K. pneumoniae was of 0.3% and 1.4%, respectively (Table 1).

Table 1

Prevalence of Polymyxin-Resistant Isolates in the Six Participating Hospitals

Isolates	Hospitals	No. of Isolates	No. of Polymyxin-Resistant Isolates (%)
Escherichia coli
	Hospital 1	326	3 (0.9)
	Hospital 2	231	1 (0.4)
	Hospital 6	942	1 (0.1)
Total		1499	5 (0.3)
Klebsiella pneumoniae
	Hospital 3	141	1 (0.7)
	Hospital 4	133	1 (0.8)
	Hospital 5	78	2 (2.6)
	Hospital 6	450	7 (1.6)
Total		802	11 (1.4)
Overall total		2,301	16 (0.7)

Antimicrobial Susceptibility Testing for Polymyxin-Resistant Isolates

Antimicrobial susceptibility testing for the 16 polymyxin-resistant isolates showed that all of the isolates were resistant to AMP, KZ, and CTX; 93.3% were resistant to LEV; 86.6% were resistant to CAZ, FEP, and ATM; 80% were resistant to SAM and TZP; 66% were resistant to GN and AK; 62.5% were resistant to IPM and MEM; and 60% were resistant to SXT. Only one isolate was resistant to CZA (6.3%), and all of them were susceptible to TGC (Figure 1).

Figure 1

Antimicrobial susceptibility profiles of the 16 polymyxin-resistant isolates. (A) Heatmap showing the resistance phenotypes of each of the isolates. (B) Percentage of strains resistant to the tested antibiotics.

The MICs of these 16 strains to polymyxin ranged from 4–64 μg/mL. K. pneumoniae isolates had MICs in the range of 4–64 μg/mL (median: 64 μg/mL), whereas the MICs for polymyxin-resistant E. coli isolates were all of 4 μg/mL (Table 2).

Table 2

Phenotypic and Genotypic Characteristics of the Polymyxin-Resistant Strains

Site	Isolate	Polymyxin MIC(μg/mL)	mCIM	eCIM	KPC	NDM	mcr–1
Hospital 5	Kpn1	4	+	–	KPC-2	–	–
Hospital 4	Kpn2	32	+	–	KPC-2	–	–
Hospital 6	Kpn3	64	+	–	KPC-2	–	–
Hospital 5	Kpn4	8	+	–	KPC-2	–	–
Hospital 6	Kpn5	64	ND	ND	–	–	+
Hospital 3	Kpn6	64	ND	ND	–	–	–
Hospital 6	Kpn7	16	+	–	KPC-2	–	–
Hospital 6	Kpn8	32	+	–	KPC-2	–	–
Hospital 6	Kpn9	64	+	–	KPC-2	–	–
Hospital 6	Kpn10	64	+	–	KPC-2	–	–
Hospital 6	Kpn11	64	+	–	KPC-2	–	–
Hospital 1	Eco1	4	+	+	–	NDM–5	+
Hospital 1	Eco2	4	ND	ND	–	–	+
Hospital 1	Eco3	4	ND	ND	–	–	+
Hospital 2	Eco4	4	ND	ND	–	–	+
Hospital 6	Eco5	4	ND	ND	–	–	–

Abbreviations: eCIM, ethylenediaminetetraacetic acid-modified carbapenem inactivation method; Eco, Escherichia coli; KPC-2, K. pneumoniae carbapenemase-2; Kpn, Klebsiella pneumoniae; mCIM, modified carbapenem inactivation method; mcr-1, mobilized colistin resistance-1; MIC, minimal inhibitory concentration; ND, data were not collected; NDM-5, New Delhi metallo-enzyme-5.

Detection of Antimicrobial Resistance Genes

Among the 16 polymyxin-resistant isolates, five carried mcr-1, including 1 K. pneumoniae and four E. coli isolates. No other mcr genes were detected in the 16 polymyxin-resistant isolates. In addition, nine K. pneumoniae and one E. coli isolates were carbapenemase-positive. The mCIM and eCIM results showed that nine K. pneumoniae isolates were serine carbapenemase-positive and one E. coli isolate was metallo-carbapenemase-positive. The PCR results showed that nine K. pneumoniae isolates were blaKPC-2-positive, but none of them carried mcr-1. Furthermore, one E. coli isolate was both blaNDM-5- and mcr-1-positive. No other carbapenemase genes, such as blaIMP, blaVIM, or blaOXA48-like, were detected (Table 2).

Among the 11 polymyxin-resistant K. pneumoniae samples, crrA was detected in 72.7% (8/11) isolates, and crrB was detected in 54.5% (6/11) isolates. Nine (81.8%) strains carried sequence insertions at the coding region for the mgrB gene. Due to the generation of amplicons larger than the normal size of the mgrB gene, one strain carried the normal mgrB gene and one strain had a W47R mutation in mgrB. ISKpn14 was the most common insertion sequence found in mgrB and was identified in seven (63.6%) isolates, followed by IS5 (two, 18.2%). No mutations were identified in phoP, phoQ, crrA, or pmrB. Only one strain carried M66I in the pmrA gene and another carried a frameshift in the crrB gene (Table 3).

Table 3

Chromosomal Mutations and Insertion Sequences (IS) in Polymyxin-Resistant K. pneumoniae Strains

Site	Isolate	mgrB	phoP	PhoQ	pmrA	pmrB	crrA	crrB
Hospital 5	Kpn1	ISkpn14	+	+	+	+	+	Fr at16L
Hospital 4	Kpn2	ISkpn14	+	+	+	+	+	–
Hospital 6	Kpn3	ISkpn14	+	+	+	+	–	–
Hospital 5	Kpn4	ISkpn14	+	+	+	+	+	+
Hospital 6	Kpn5	+	+	+	M66I	+	+	+
Hospital 3	Kpn6	W47R	+	+	+	+	+	+
Hospital 6	Kpn7	ISkpn14	+	+	+	+	+	+
Hospital 6	Kpn8	ISkpn5-like	+	+	+	+	-	-
Hospital 6	Kpn9	ISkpn5-like	+	+	+	+	+	-
Hospital 6	Kpn10	ISkpn14	+	+	+	+	-	-
Hospital 6	Kpn11	ISkpn14	+	+	+	+	+	+

Abbreviations: Fr, frameshift, +, presence of PCR product and no change in the nucleotide/amino acid sequences; −, absence of PCR product.

Overexpression of phoPQ and pmrK Contribute to Polymyxin-Resistance in K. pneumoniae

In general, transcriptional upregulation of the phoQ, phoP, and pmrK genes was observed in all strains with mgrB alterations, except for KPN5 with a normal mgrB gene. Compared to the levels obtained for our laboratory polymyxin-susceptible wild-type strains, the transcription level of phoP was elevated from 8.4- to 24.9-fold, that of phoQ was elevated from 6.9- to 28.9-fold, and that of pmrK was elevated from 3.4- to 21.3-fold. Moreover, three polymyxin-resistant K. pneumoniae isolates were found to have increased expression of the pmrA gene from 4.2- to 10.1-fold, of the pmrB gene from 5.0- to 10.7-fold, and of the pmrC gene from 13.3- to 21.6-fold (Figure 2).

Figure 2

Transcriptional levels of phoP, phoQ, pmrK, pmrA, pmrB, and pmrC in polymyxin-resistant K. pneumoniae.

Epidemiological Characterization

MLST analysis showed that nine K. pneumoniae isolates belonged to the sequence type (ST) 11, and the other two isolates belonged to ST37 and ST364, respectively. Among the five E. coli isolates, two belonged to ST132, and the other three isolates belonged to ST2, ST983, and ST632 (Figure 3).

Figure 3

Pulsed-field gel electrophoresis (PFGE)-based dendrogram of polymyxin-resistant Escherichia coli (A) and Klebsiella pneumoniae (B) strains.

Dendrogram analysis of PFGE at ≥ 95% similarity revealed that homology among the five E. coli and 11 K. pneumoniae isolates was low and sporadic, suggesting a very low likelihood of clonal spread (Figure 3).

Clinical Characteristics of Patients Harboring Polymyxin-Resistant Isolates

The 16 polymyxin-resistant isolates were collected within 1 year from 15 patients aged 2 months to 93 years. One patient had two isolates collected separately from blood and urine, whereas from all other 14 patients each isolate had been collected from urine (n = 4), BAL (n = 3), blood (n = 3), secretion (n = 2), peritoneal puncture fluid (n = 1), and sputum (n = 1) samples. The diagnosed diseases for these patients included cerebrovascular disease (n = 3), urinary tract disease (n = 3), pneumonia (n = 2), sepsis (n = 1), fever (n = 1), acute coronary syndrome (n=1), pregnancy-induced hypertension (n = 1), premature baby (n = 1), infection around the prosthesis (n = 1), and Guillain-Barre syndrome (n = 1). Five patients received polymyxin treatment before the isolation of polymyxin-resistant strains. Ten patients displayed positive clinical outcomes (Table 4).

Table 4

Clinical Characteristics of Patients Carrying Polymyxin-Resistant Bacterial Strains

Pt	Sex/Age (Years)	Isolate	Source	Clinical Diagnosis	Underlying Disease	Indwelling Devices	Antimicrobial Use Prior to Culture Within 30 Days	Outcome
1	M/87	Kpn	Blood	Cerebral infarction	Diabetes	Tracheal cannula	TGC, Carbapenems	Discharge
2	F/33	Kpn	Secretion	Pneumonia, AFE	No	No	Clindamycin, Quinolones	Discharge
3	M/38	Kpn	Sputum	Septic shock	No	CVC, Tracheal cannula	β-lactam, Quinolones	Death
4	F/69	Eco/Eco	Blood/urine	Acute coronary syndrome	Diabetes, CHD	No	β-lactam, Quinolones	Discharge
5	M/81	Eco	Blood	Fever	Hemodialysis	No	β-lactam, Quinolones,	Discharge
6	M/93	Eco	BAL	Cerebral hemorrhage	Diabetes	CVC, Tracheal cannula	Carbapenems	Discharge
7	F/69	Kpn	Urine	Urinary retention	No	CVC	β-lactam	Worsening
8	M/62	Kpn	BAL	Cerebral hemorrhage	No	Tracheotomy	TGC, Polymyxin	Worsening
9	F/89	Eco	Urethral secretions	cUTI	Hypertension, CHD	Urethral catheter	β-lactam, Quinolones	Death
10	M/56	Kpn	Urine	Guillain-Barre syndrome	Hypertension	Tracheal cannula	TGC, Polymyxin, Carbapenems	Discharge
11	M/67	Kpn	Urine	Urethral injury	Hypertension	Urethral catheter	β-lactam	Discharge
12	F/2 months	Kpn	BAL	Premature baby	No	Tracheal cannula	Polymyxin, Fosfomycin, fluconazole	Discharge
13	F/28	Kpn	Peritoneal puncture fluid	Pregnancy-induced hypertension	SLE	Peritoneal drainage tube	TGC, Polymyxin, Carbapenems	Discharge
14	M/89	Kpn	Blood	Severe pneumonia	No	Tracheal cannula	Polymyxin, Carbapenems, fluconazole	Death
15	F/54	Kpn	Secretion	Infection around the prosthesis	No	No	Carbapenems, Quinolones	Discharge

Abbreviations: AFE, amniotic fluid embolism; BAL, bronchoalveolar lavage fluid; Kpn, Klebsiella pneumoniae; CVC, central venous catheter; CHD, coronary heart disease; cUTI, complicated urinary tract infection; Eco, Escherichia coli; SLE, systemic lupus erythematosus; TGC, tigecycline.

Discussion

Polymyxin has been used against aggressive infections caused by multidrug-resistant bacteria, but its use has been severely compromised by the emergence of plasmid-mediated polymyxin resistance in Enterobacteriaceae. Hence, we surveyed polymyxin resistance rates and mechanisms in E. coli and K. pneumoniae isolates from hospitalized patients at six local hospitals in the Henan province.

Among the total 2301 E. coli and K. pneumoniae isolates, 16 (0.7%) were polymyxin-resistant, five of which carried mcr-1. In addition, 0.88% (34/3854) of the E. coli isolates and 0.21% (5/2410) of the K. pneumoniae isolates carrying mcr-1 were previously reported in the China Antimicrobial Resistance Surveillance Trial.20 Of the 1499 E. coli isolates in this study, 5 (0.3%) were polymyxin-resistant and 4 were mcr-1-positive. Previous research found that 1% (20/1495) of the E. coli isolates and 0.18% (1/571) of the K. pneumoniae isolates recovered from bloodstream infections in China were mcr-1-positive.21 Herein, of the 802 K. pneumoniae isolates, 11 (1.4%) were polymyxin-resistant, one of which carried mcr-1. Although mcr-1 was more common in E. coli than in K. pneumoniae isolates, the polymyxin resistance rate of K. pneumoniae was slightly higher than that of E. coli in our study, which is possibly due to antibiotic selection driven by the high detection rate (32.8%) of carbapenem-resistant K. pneumoniae (PRCRKP) in Henan among all Chinese provinces in 2019.22

Compared to polymyxin-resistant E. coli, polymyxin-resistant K. pneumoniae had 8–64 times higher MICs, and insertion sequences in the mgrB were detected in 81.8% of polymyxin-resistant K. pneumoniae isolates, which suggested that chromosomal mutations by insertional inactivation mediated by mobile insertion sequences may play important roles in polymyxin resistance in K. pneumoniae in the Henan province. Many studies found that mgrB alterations, including insertion elements, non-silent point mutations, and small intragenic deletions, mediate colistin resistance.16,18,23 ISkpn14 was the most common insertion observed in our study. However, an ISKpn26-like element was predominant in Taiwan23 and Greece.10 One strain carried a W47R mutation in the mgrB gene, which has been previously reported.24 M66I mutation in the pmrA gene and a frameshift at 16L in the crrB gene were newly found in this study, but further research is needed to clarify the mechanism.

The mgrB gene was a negative feedback regulator of the PhoQ-PhoP signaling system; therefore, phoP, phoQ, and pmrK were all upregulated in polymyxin-resistant K. pneumoniae with mgrB alterations in this study, as previously reported.16,25 pmrA, pmrB, and pmrC overexpression was found only in three polymyxin-resistant K. pneumoniae isolates, which implies that mgrB alterations might be the main reason for polymyxin resistance in K. pneumoniae.

Two carbapenemase genes, blaKPC and blaNDM, are responsible for the phenotypic resistance of 90% of carbapenem-resistant Enterobacteriaceae strains in China.26 The co-existence of mcr and carbapenemase genes, such as blaNDM-5,27 blaNDM-4,28 blaKPC,29 and blaOXA,30 has been sporadically reported in different countries. In the national monitoring data from China, one report showed that the mcr-1 was detected in 4.6% (13/282) of carbapenem-resistant E. coli isolates and coexisted with New Delhi metallo-enzyme (NDM)-5 in one strain.31 In another study, mcr-1 prevalence among carbapenem-resistant E. coli and PRCRKP isolates was 3.7% (14/376) and 0% (0/1134), respectively, and 14 carbapenem-resistant E. coli isolates coproduced blaNDM4/5/9 with mcr-1.32 In this study, only one E. coli isolate coproduced mcr-1 and blaNDM-5.

An E. coli isolate belonging to ST167 that coproduced blaNDM and mcr-1 was previously reported in Henan,6,33 but in our study, the aforementioned coproducing E. coli isolate belonged to ST2. The other E. coli strains in this study belonged to ST132, ST983, and ST632. Nine PRCRKP isolates belonged to ST11, but PFGE showed that they belonged to different types and were not clonal, even when isolated from the same hospital. These results demonstrated that polymyxin-resistant isolates were non-clonal and had different resistance potentials. It was reported that ST11 KPC-Kp was clonally heterogeneous and could be further classified into eleven mobile genetic element (MGE) types and fourteen PFGE subtypes.34

The patients carrying polymyxin-resistant isolates had varying severities of illness, and 33.3% of them had a history of polymyxin use. Moreover, 66.7% of them were cured, which could be explained by the retained susceptibility to other antimicrobials, such as CZA, SXT, and TGC, that most polymyxin-resistant isolates showed.

The limitation of this study is the use of the polymyxin B susceptibility test strip to screen polymyxin-resistant isolates, which is not the gold standard for identifying resistance to polymyxin. Indeed, it has been reported that results obtained by this method has high very major errors (VMEs), leading to an underestimation of the resistance.35–37

Conclusions

In conclusion, the polymyxin resistance rate of K. pneumoniae was slightly higher than that of E. coli, and mcr-1 occurrence was lower in polymyxin-resistant K. pneumoniae than in polymyxin-resistant E. coli in the Henan province of China. Further molecular investigations showed that insertion sequences in the mgrB gene might be the main mechanism contributing to polymyxin resistance in K. pneumoniae in Henan. In addition, continuous and close monitoring is required to prevent the dissemination of polymyxin-resistant K. pneumoniae and E. coli strains.