Carbapenem-Resistant Klebsiella pneumoniae in Southwest China: Molecular Characteristics and Risk Factors Caused by KPC and NDM Producers

Introduction

Klebsiella pneumoniae, belonging to the family Enterobacteriaceae, is an important pathogen that causes opportunistic infections in hospitalized patients, and plays a primary role in pneumonia and neonatal sepsis.¹ With the increase of prevalence of multidrug-resistant (MDR) strains that producing extended-spectrum β-lactamase (ESBL) and/or carbapenemase, K. pneumoniae has emerged as a major threat in clinical and public health. Carbapenem-resistant K. pneumoniae (CRKP) has become one of the particularly important problems worldwide due to the gradually higher morbidity and subsequently worrying mortality.^2,3 According to the 2020 CHINET Resistance Monitoring Network, K. pneumoniae resistance to imipenem has risen rapidly from 16.1% in 2016 to 23.3% in 2020.

The resistance mechanism of Enterobacteriaceae to carbapenem antibiotics is usually caused by two major mechanisms: (i) Nonenzymatic resistance mechanisms and (ii) Hydrolysis of carbapenems by carbapenemases production.⁴ Nonenzymatic resistance mechanisms mainly include overexpression of efflux pump-encoding genes and the mutation of porins.^5,6 As for later mechanism, carbapenemases are classified into three groups (ie, A, B, D) based on their molecular structure. Class A and D carbapenemases require an acyl enzyme that forms from an active serine site to hydrolyze their substrates, while class B metalloenzymes utilize the active zinc ion site to induce hydrolysis of carbapenem.⁷ Klebsiella pneumoniae carbapenemase (KPC) is the most common class A enzyme in Enterobacteriaceae and have widely disseminated due to the encoding genes located on plasmid, especially for KPC-2.⁸ In addition, New Delhi metallo-β-lactamase (NDM) is one of the important member in class B carbapenemases, and the NDM-1 is the most prevalent type globally. NDM-1-producing clinical strains have caused several outbreaks in China since it was first reported in 2012.^9–11 In other regions, such as Colombia, Italy, and Brazil, these strains also lead to health crisis.^12–14 Furthermore, a more pathogenic and virulent phenotype, also known as hypervirulent K. pneumoniae (HVKP), has been well-separated quickly. Importantly, the emergence of multidrug-resistant HVKP (MDR-HVKP) has been gradually reported due to the horizontal transfer of mobile genetic elements, bringing a remarkable challenge to healthcare system.^15–17

The objectives of this study were to systematically analyze the clinical characteristics, molecular epidemiology, and risk factors for CRKP infection in a tertiary hospital. These findings will provide valuable information for further monitoring and controlling the dissemination of KPC and NDM K. pneumoniae strains in Southwest China.

Materials and Methods

Data Collection

The isolates were defined as CRKP if they were resistant to at least one of the carbapenem agents, including imipenem, ertapenem, and meropenem (www.cdc.gov/HAI/organisms/cre). A total of 66 non-repetitive clinical CRKP isolates were collected from September 2016 to August 2019 in the Affiliated Hospital of Southwest Medical University (Luzhou, China), and all the CRKP isolates were confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS, Bruker Daltonics, Bremen, Germany).

In order to investigate the risk factors for CRKP infection, a stepwise matching method at a ratio of 1:1 has been used to identify appropriate control cases from patients who are infected with carbapenem-susceptible K. pneumoniae (CSKP). The same site of infection, the same gender, age ± 2 years, and the same year of hospital admission were considered as the matching criteria. The relative clinical data were retrospectively collected from medical records of each patient, including basic demographic information, underlying diseases and comorbidities, invasive procedures, antibiotic treatment, and clinical outcomes.

Antimicrobial Susceptibility Testing (AST)

The antimicrobial susceptibilities of all the 66 clinical CRKP isolates to 16 antimicrobials were tested by VITEK 2 Compact system (bioMérieux, Marcy l’Etoile, Lyon, France), including cefepime, cefotaxime, cefazolin, cefuroxime, ceftazidime, cefoxitin, piperacillin/tazobactam, ampicillin/sulbactam, amikacin, tobramycin, gentamicin, levofloxacin, ciprofloxacin, sulperazone, compound sulfamethoxazole, and aztreonam. The minimum inhibitory concentrations (MICs) of meropenem, imipenem, ertapenem, tigecycline, and polymyxin B were determined by broth microdilution method, and the results were interpreted according to the standards of the Clinical and Laboratory Standards Institute (CLSI) 2020-M100.¹⁸ Pseudomonas aeruginosa ATCC27853 and Escherichia coli ATCC25922 were used as quality control strains (purchased from China National Health Inspection Center).

Screening of CRKP and Phenotypic Detection of Carbapenemase

The metallo-β-lactamase-producing isolates were differentiated from all the 66 CRKP by using the imipenem-EDTA double-disk synergy method and the carbapenem inactivation method (CIM). Briefly, the bacterial suspension (0.5 McFarland standard) was diluted from the overnight culture, and then smeared on Mueller-Hinton (MH) agar plate (Haibo, Qingdao, China). A disk containing 10 μg imipenem (OXOID, ThermoFisher Scientific, Massachusetts, USA) and a blank disk with 1.5mg/mL EDTA were placed 10 mm between each disk on the plate. After 18 h incubation, an enlarged zone of inhibition was considered as EDTA-synergy test positive.^18,19

The CIM was performed as described previously.^18,20 A single colony of CRKP isolates was inoculated into a tube containing 2 mL trypticase soy broth (TSB) (Haibo, Qingdao, China) and a tube containing 2 mL TSB with 5mM EDTA, respectively. A 10 μg meropenem disk was placed in each tube. After 4 h incubation, these disks were took out and placed on a MH agar plate that was inoculated with a lawn of the meropenem-susceptible Escherichia coli ATCC25922 (0.5 McFarland standard). The results were interpreted according to CLSI.¹⁸

The Detection of Resistance Genes, Virulence Genes, and Integrase-Associated Genes

All of template DNAs were extracted by bacterial DNA Kit (Tiangen, Beijing,China) and used to detect carbapenemase genes, ESBL genes, AmpC β-lactamase genes, virulence genes, and integrase-associated genes by polymerase chain reaction (PCR). The primers were shown in Tables S1 and S2. The PCR conditions were described previously.^21–23 Positive products were sequenced by Shanghai Jieli Biotechnology, and the variable region of the class I integron (intl1) and the insertion sequence common region I (ISCR I) were analyzed by BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Plasmid Conjugation and Analysis

In order to evaluate whether the carbapenemase genes are located on the plasmid and whether these genes can be horizontally transferred, a plasmid conjugation transfer experiment was performed. Four hundred μL donor strain (CRKP strains) and 200 μL receptor strain (sodium azide-resistant E. coli strain J53) at logarithmic phase were added into a glass tube containing 800 μL Luria-Bertani (LB) broth and cultured at 37°C for 18 h. Transconjugants were selected by using LB plates that contains sodium azide (180 μg/mL) and imipenem (0.5 μg/mL).²¹ The transconjugants were verified by MALDI-TOF MS, and the conjugative carbapenemase genes were confirmed by PCR. In addition, the successful conjugative plasmids were analyzed by PCR-based replicon typing according to previous study.²²

Genetic Environments of KPC-2-Carrying Plasmids and NDM-1-Carrying Plasmids

The overlapping PCR was applied to investigate the genetic environments of the bla_KPC-2 and bla_NDM-1 (Tables S3 and S4).^23,24 The PCR products were sequenced by Shanghai Jieli Biotechnology, and then analyzed by NCBI GenBank database.²⁵

Molecular Epidemiological Study

Multi-locus sequence typing (MLST) was performed to explore the genetic correlation of all the clinical CRKP isolates.²⁶ The positive products were sequenced by Shanghai Jieli Biotechnology, and the results were submitted to K. pneumoniae MLST database (https://bigsdb.pasteur.fr/cgi-bin/bigsdb/bigsdb.pl?db=pubmlst_klebsiella_seqdef) for comparison.

Statistical Analysis

All analyses and graphs were performed using SPSS v.24.0 software (SPSS Inc., Chicago, USA). The chi-square test or Fisher’s exact test was used to analyze categorical variables. Continuous variables were presented as means ± standard deviation (SD), and were evaluated by Student’s t-tests or Mann–Whitney U-test. Multivariable logistic regression analysis was operated to identify independent risk factors for CRKP infection. All biologically plausible variables with a value of P < 0.1 within univariate analysis were included in the following multiple logistic regression model. P < 0.05 was considered as statistically significant, and all probability values were two-tailed distribution.

Results

Distribution of Clinical CRKP Isolates

A total of 66 non-repetitive clinical CRKP isolates were collected from September 2016 to August 2019 in Southwest China. The clinical CRKP isolates were obtained from patients admitted to the 12 different departments, including the majority of neonatology department (27.3%, n = 18) and rehabilitation department (24.3%, n = 16) (Figure 1A). These isolates were mainly originated from sputum (34.8%, n = 23), followed by urine (25.8%, n = 17), blood (21.2%, n=14), secretion (6.1%, n= 4), pleuroperitoneal fluids (6.1%, n = 4), catheter tip (3.0%, n = 2), and pus (3.0%, n = 2) (Figure 1B).

Figure 1 Distribution of 66 clinical CRKP strains. (A) Department distribution of 66 CRKP strains; (B) sample types of 66 CRKP strains.

Antimicrobial Susceptibility Profiles

All of 66 clinical CRKP isolates were multidrug-resistant and resistance rates to 12 antimicrobials reached 100%, which were cefepime, imipenem, piperacillin/tazobactam, ceftriaxone, ampicillin/sulbactam, cefotaxime, cefazolin, cefuroxime, ertapenem, ceftazidime, meropenem, and cefoxitin. In addition, the resistance rates to sulperazon, aztreonam, tobramycin, gentamicin, levofloxacin, ciprofloxacin, cotrimoxazole, and amikacin were 95.5%, 87.9%, 83.3%, 83.3%, 81.9%, 80.3%, 65.2%, and 54.5%, respectively. Remarkably, all the CRKP strains in this study were sensitive to tigecycline and polymyxin B (Table 1).

Carbapenem Resistance Phenotype and Molecular Characteristics

Both EDTA-synergy test and CIM test were shown that 24 (36.4%) clinical CRKP strains produced class B carbapenemases. The sequencing results confirmed that all of which isolates contained bla_NDM, in which 20 isolates carried bla_NDM-1 and 4 isolates carried bla_NDM-5, and 3 isolates possessed bla_NDM and bla_IMP simultaneously. The rest of 42 (63.6%) isolates contained bla_KPC-2. Notably, in addition to the production of carbapenemase, 100% and 95.5% of the clinical CRKP isolates were positive for ESBL and AmpC genes, respectively. Furthermore, virulence genes were detected in all isolates, 54 (81.8%) isolates detected three different virulence genes. Especially for strain Kpn497 and Kpn131, they carried 12 virulence genes and 11 virulence genes, respectively (Table 2).

Table 1 Antimicrobial Susceptibility Profiles of 66 Clinical CKRP Strains

Table 2 Genotypes of 66 Clinical CRKP Strains

Plasmid and Integron-Associated Analysis

KPC-encoding plasmids were positive in 35 transconjugants and all of these plasmids belonged to IncFII. Meanwhile, NDM-encoding plasmids that belong to IncX3 were detected in 24 transconjugants (Table 2).

Nearly 85% (n = 56) strains were intl1-positive, of which 46 strains contained variable region. A total of 6 cassette arrays were found among them. The cassette arrays were revealed in this study included: addA2 (82.6%), ddfrA12 (6.5%), dfrA12-addA2 (4.3%), dfrA27-arr3 (2.2%), gcuF-dfrA12 (2.2%), and bla_OXA-10-addA1-aacA4 (2.2%) (Table 2) In addition, ISCRI was detected in 25 strains.

Genetic Environments of bla_KPC-2 and bla_NDM-1

All of 42 KPC-2-producing strains could be divided into three different types (A, B, and C) based on the genetic structures that compared with pKP048 (GenBank Accession No.FJ628167). Type C was the most prevalent (n = 31), followed by type A (n = 8) and type B (n = 3). Type A exhibited the same structure as pKP048. The downstream of ISKpn6-like was deletion in type B, and three genetic elements downstream of bla_KPC-2 were deletion in type C, including ISKpn6-like, tnpR, and tnpA (Figure 2A).

Figure 2 Comparison of the genetic elements surrounding the blaKPC-2 and blaNDM-1 genes identified in this study. (A) Comparison of the genetic environments of blaKPC-2, reference sequences: K. pneumoniae (pKP048, GenBank Accession No.FJ628167); (B) comparison of the genetic environments of blaNDM-1, reference sequences: A. lwoffii (pNDM-BJ01, GenBank accession NO. JQ001791).

Furthermore, the genetic structure of pNDM-BJ01 (GenBank accession No. JQ001791) was used to compare with each NDM-1-producing strain, revealing A, B, and C types among these strains. As for type A (n = 12), the ISAba125 upstream of the bla_NDM−1 was truncated, all downstream genes of tat were complete deletion. ISAba125 upstream of the bla_NDM−1 was deletion in type B compared with type A. Type C presented as the IS30 upstream of bla_NDM-1 and followed by ble_MBL, trpF, tat, cutA, groES, groEL (Figure 2B).

MLST of Clinical CRKP Isolates

A total of 13 MLSTs were identified among all the clinical CRKP isolates, and ST11 (56.2%, n = 37) was the most frequent ST type, followed by ST4495 (19.7%, n = 13), ST2407 (4.6%, n = 3), ST147 (3.0%, n = 2), ST307 (3.0%, n = 2), ST37 (3.0%, n = 2), ST15 (1.5%, n = 1), ST17 (1.5%, n = 1), ST152 (1.5%, n = 1), ST23 (1.5%, n = 1), ST405 (1.5%, n = 1), ST318 (1.5%, n = 1), and ST467 (1.5%, n = 1).

Risk Factors and Multivariate Analysis of CRKP Infection

Statistically significant differences were observed for respiratory disease (P = 0.034), renal disease (P = 0.039), hematological disease (P = 0.02), tracheal cannula (P < 0.001), and use of β-lactams/β-lactamase inhibitor combination (P = 0.009) between CRKP and CSKP groups (Table 3). Multivariate analysis revealed that hematological disease (odds ratio [OR], 2.568; 95% confidence interval [95% CI], 1.106 to 5.964; P = 0.028), tracheal cannula (OR, 4.883; 95% CI, 1.797 to 13.265; P = 0.002), and use of β-lactams/β-lactamase inhibitor combination (OR, 4.271; 95% CI, 1.760 to 10.365; P = 0.001) were independent risk factors for CRKP infection.

Table 3 Clinical Characteristics of CRKP and CSKP Strains

Discussion

The clinical CRKP strains were first reported in 1997,^27,28 and this type of resistance did not become common within that decade. However, with the increasing use of carbapenems in recent years, CRKP strains have been distributed around the world at a boosting rate, and it was listed as one of the critical-priority bacteria by World Health Organization (WHO).²⁹ In this work, the antimicrobial susceptibility profiles, molecular characteristics, plasmid and integron-associated analysis, genetic environments of bla_KPC-2 and bla_NDM-1, and MLST were evaluated among 66 clinical CRKP isolates. Meanwhile, the risk factors for CRKP infection were also investigated. All the CRKP strains were MDR or even extensively drug-resistant (XDR), in this case, the tigecycline and polymyxin B were the last resort options for CRKP infection. However, the treatment was hardly successful because of the low blood concentrations when tigecycline was used in monotherapy.³⁰ Adverse events, such as hypofibrinogenemia, also resulted in treatment termination associated with tigecycline therapy.³¹ In addition, nephrotoxicity and unclear dosing also limited the widespread clinical usage of polymyxin B.^32,33 Importantly, tigecycline- and polymyxin B-resistant CRKP strains have been reported in many regions,^34,35 therefore, the optimal therapeutic strategy for CRKP infection still needs to be explored.

Focusing on the phenotype and molecular analysis of CRKP isolates, the results revealed that the primary resistance mechanism of CRKP was carbapenemase production, which was consistent with other researches.^36,37 The bla_KPC-2 and bla_NDM-1 were two major carbapenemase genes in our study, accounting for 63.6% and 30.3%, respectively. KPC-2 is the most commonly identified enzyme among over 20 reported KPC variants so far,³⁸ and the majority of NDM-1-producers also harbored various other resistance genes but mostly were still susceptible to polymyxin B and tigecycline,³⁹ which were highly correlated with our results. Moreover, IMP is another important metallo-β-lactamase, and bla_IMP-1 emerged in Japan in the late 1980s.⁴⁰ The bla_IMP-4 was one of the most prevalent bla_IMP variants since it was first identified in Acinetobacter spp. in the mid-1990s,⁴¹ particularly in China and Australia.^42,43 It is worth noting that bla_IMP often co-exists with other resistance genes,⁴⁴ which is consistent with our results. In our work, 3 strains were identified co-carrying bla_NDM and bla_IMP genes, including two isolates with bla_NDM-1 and bla_IMP-4 and one with bla_NDM-5 and bla_IMP-4. Coexistence of bla_NDM-1 and bla_IMP-4 was still rarely distributed in China since first reported in 2012, but the strain co-harboring bla_NDM-5 and bla_IMP-4 has never been published before.⁴⁵

In this work, bla_KPC-2 was only found on IncFII plasmids, and bla_NDM was only detected on IncX3 plasmids. The IncFII plasmid family is commonly low copy number, not only carries multiple antimicrobial resistance and virulence genes, but also can replicate and disseminate in many different species of Enterobacteriaceae.⁴⁶ Furthermore, several studies have revealed that IncX3 plasmids were the predominant plasmid type that harboring a variety of bla_NDM genes.^47–49 Of note, IncX3 plasmids can frequently disseminate in humans, animals, and environment, and more commonly spread in Asia, including China,^49,50 Myanmar,⁵¹ South Korea,⁵² and India.⁵³ In addition, 46 CRKP strains carried variable regions of intl1, and 7 resistance gene cassettes were detected among them. The addA2 (40/46) and addA1 (1/46) were the genes responsible for aminoglycoside resistance. The dfrA12 (6/46), dfrA27 (1/46) contributed to trimethoprim resistance. Furthermore, a β-lactams-resistant gene bla_OXA-10, a gentamicin-resistant gene arr3, and a rifampicin-resistant gene aacA4 were also detected.

A previous study has proved that the genetic environment of bla_KPC-2 in pKP048 from China was considered as a combination of the Tn3-based transposon and the partial Tn4401 structure.⁵⁴ In our study, KPC-2 producers presented a similar structure (tnpA-tnpR-ISkpn8-bla_KPC-2), but the environment surrounding bla_KPC-2 was partly different from other reports in China due to the specific insertions and deletions.^55,56 Additionally, all the NDM-1 producers carried the highly conserved region (bla_NDM−1-ble_MBL-trpF-tat) surrounding the bla_NDM−1 gene, which was identified not only in K. pneumoniae, but also in Acinetobacter lwoffii and Enterobacter Cloacae.^25,57,58

Furthermore, the results showed that more than 80% clinical CRKP strains simultaneously carried virulence genes entB, fimH, and mrkD. The enterobactin that encoded by entB is one of the ubiquitous siderophores in K. pneumoniae, which is an important factor involving in iron acquisition.⁵⁹ The fimH and mrkD mediate bacterial adhesion through encoding type I and type III fimbriae of K. pneumoniae implied that these two genes may contribute to biofilm formation and development. Zhang et al have also observed that 95% tested K. pneumoniae isolates co-carrying fimH and mrkD genes in their study.⁶⁰ Additionally, the genes peg-344, iroB, iucA, rmpA, and rmpA2 locate on the HVKP virulence plasmid and are regarded as regulators of hypervirulent phenotype.⁶¹ Notably, two carbapenem-resistant HVKP (CR-HVKP) strains were found in this work, which contained bla_KPC-2 gene. There are two hypotheses to explain the evolution of this novel phenotype. First, CRKP strains acquire HVKP-specific virulence plasmid; second, HVKP strains obtain the carbapenem-resistant genes by acquisition of resistance plasmid or by the insertion of resistance determinants into HVKP-specific virulence plasmid.^59,62 Compared with the second hypothesis, the first one was more likely to occur.^63–65

According to the results of MLST, the 42 KPC-2-producing CRKP strains were grouped into 7 ST types, of which 35 strains belonged to ST11. It has been reported that most of KPC-2-producing strains belong to clone group (CG) 258; meanwhile, ST11 and ST258 are the dominant ST types.⁶⁶ ST258 has spread around the world since it emerged in the early 21st century, particularly in North America, Latin America, and several European countries.^66,67 However, ST11 is more prevalent in Asia, and usually accounts for more than half of all KPC-2-producing CRKP strains in China.^47,68,69 In addition, the 24 NDM-producing CRKP strains were classified into 8 ST types, and ST4495 was the most common ST type. Remarkably, ST4495 (4-1-99-1-9-5-5) has never been detected before, indicating a novel clone carrying bla_NDM-1 has spread in Southwest China.

Hematological disease, tracheal cannula, and exposure to β-lactams and β-lactamase inhibitor combination were independent risk factors for CRKP infection. Many studies have demonstrated that the history of tracheal cannula was independent risk factor for CRKP infection.^70–73 Meanwhile, exposure to antibiotic was also associated with CRKP infection, such as exposure to quinolones,⁷⁴ β-lactams and β-lactamase inhibitor combination, and carbapenems,⁷⁵ which was consistent with our study. Notably, hematological disease was rarely found to be a risk factor for CRKP infection in other studies, but it is reasonable due to patients with hematological disease were accompanied with severe immune function deficiency and were more prone to be infected. There are some limitations in our study, first, the number of patients included in this study is relatively small, although this is a common problem in studies assessing risk factors for multidrug-resistant microbial infections.⁷⁴ Secondly, the study was performed in a single-center setting, so that some important risk factors may be missed.

Conclusion

The current study revealed the high prevalence of CRKP infection caused by NDM-1 and KPC-2 producers, and ST11 was the most common lineage. To the best of our knowledge, the novel ST4495 and co-exist of bla_NDM-5 and bla_IMP-4 were the first reported in this work. Moreover, plasmid, integron, ISCR, and other mobile genetic elements endowed bacteria a rapid adaptation ability in changing environments; meanwhile, the genetic environments of bla_KPC-2 and bla_NDM-1 showed polymorphism. Logistic regression model indicated that hematological disease, tracheal cannula, and use of β-lactams and β-lactamase inhibitor combination were independent risk factors for CRKP infection, suggesting that appropriate clinical management and antimicrobials treatment were necessary for retarding the selection of CRKP strains. Effective measures to prevent and control the further dissemination of CRKP strains are highly needed.