Prevalence of intestinal colonization and nosocomial infection with carbapenem-resistant Enterobacteriales in children: a retrospective study

Objective We investigated the epidemiological surveillance of the intestinal colonization and nosocomial infection of carbapenem-resistant Enterobacteriales (CRE) isolates from inpatients, which can provide the basis for developing effective prevention. Methods A total of 96 CRE strains were collected from 1,487 fecal samples of hospitalized children between January 2016 and June 2017, which were defined as the “CRE colonization” group. In total, 70 CRE clinical isolates were also randomly selected for the comparison analysis and defined as the “CRE infection” group. The antimicrobial susceptibility of all strains was determined by the microdilution broth method. Polymerase chain reaction (PCR) was used to analyze carbapenemase genes, plasmid typing, and integrons. Multilocus sequence typing was further used to determine clonal relatedness. Results In the “CRE colonization” group, Klebsiella pneumoniae was mostly detected with a rate of 42.7% (41/96), followed by Escherichia coli (34.4%, 33/96) and Enterobacter cloacae (15.6%, 15/96). The ST11 KPC-2 producer, ST8 NDM-5 producer, and ST45 NDM-1 producer were commonly present in carbapenem-resistant K. pneumoniae (CRKPN), carbapenem-resistant E. coli (CRECO), and carbapenem-resistant E. cloacae (CRECL) isolates, respectively. In the “CRE infection” group, 70% (49/70) of strains were K. pneumoniae, with 21.4% E. cloacae (15/70) and 5.7% E. coli (4/70). The ST15 OXA-232 producer and ST48 NDM-5 producer were frequently observed in CRKPN isolates, while the majority of NDM-1-producing CRECL isolates were assigned as ST45. Phylogenetic analysis showed that partial CRE isolates from intestinal colonization and nosocomial infection were closely related, especially for ST11 KPC-2-producing CRKPN and ST45 NDM-1-producing CRECL. Furthermore, plasmid typing demonstrated that IncF and IncFIB were the most prevalent plasmids in KPC-2 producers, while IncX3/IncX2 and ColE were widely spread in NDM producer and OXA-232 producer, respectively. Then, class 1 integron intergrase intI1 was positive in 74.0% (71/96) of the “CRE colonization” group and 52.9% (37/70) of the “CRE infection” group. Conclusion This study revealed that CRE strains from intestinal colonization and nosocomial infection showed a partial correlation in the prevalence of CRE, especially for ST11 KPC-2-producing CRKPN and ST45 NDM-1-producing CRECL. Therefore, before admission, long-term active screening of rectal colonization of CRE isolates should be emphasized.


Background
Enterobacteriales represent a large family of gram-negative bacteria, which cause a series of nosocomial infections including pneumonia, bloodstream infections, and urinary tract infections.Carbapenems including ertapenem, imipenem, and meropenem are commonly considered the last line of defense to treat multidrugresistant bacterial infections, and the wide and unreasonable use of these antibiotics has consequently resulted in the increasing incidence of antimicrobial resistance.Carbapenemresistant Enterobacteriales (CRE) are regarded as the top-ranked bacteria in the priority list of resistant pathogens of the World Health Organization and have become a severe threat to global public health (1, 2).In China, the overall incidence of CRE infection per 10,000 discharges was 4.0 and differed significantly by region (3).Production of carbapenemases including Klebsiella pneumoniae carbapenemase (KPC) and New Delhi metallo-βlactamase (NDM) was the main mechanism of CRE.Then, the rapid spread of CRE severely limited the options for antibiotics and prolonged the length of hospital stay, ultimately resulting in high mortality.
The family Enterobacteriales naturally colonizes in the gastrointestinal tract, which is the most important reservoir for CRE strains.Previous studies reported that the prevalence of CRE colonization rates was less than or ∼5%, but it would be higher in critical patients in the intensive care unit (4)(5)(6)(7).Furthermore, fecal carriage of CRE isolates is regarded as a risk factor for developing CRE infections.A recent study showed that 50% of the patients with CRE rectal colonization could develop CRE-associated pneumonia and simultaneously were more likely to have a long hospital stay and high hospitalization expense than patients with non-colonized CRE (8).Therefore, early and rapid detection of CRE colonization among high-risk patients will be helpful to implement strategies to prevent CRE dissemination.
Many guidelines recommend that active screening of rectal colonization of CRE is an easy and important tactic to control CRE infection, and rectal swabs are considered as the preferred sample for screening on account of good patient compliance and fewer side effects (9,10).Accordingly, we started the CRE screening for hospitalized children in 2016 and continuously monitored the change in CRE prevalence.However, the distribution of characteristics between the colonized CRE and clinical CRE isolates among children still needs to be summarized.In this study, we will investigate epidemiological surveillance of the colonization/infection of CRE isolates from inpatients, which can provide the basis for developing effective prevention measures for CRE infections.

Study design and bacterial isolates
To compare the colonization/infection of CRE isolates from inpatients, this study was designed in Shanghai Children's Hospital between January 2016 and June 2017.The inclusion criteria were as follows: Fecal sample was collected from hospitalized children aged 0 to 18 years who received the routine fecal culture testing, and only the first fecal sample of each inpatient was enrolled.The exclusion criteria were as follows: (1) patients older than 18 years; (2) patients from an outpatient clinic; and (3) patients with duplicate fecal samples.A total of 1,487 fecal samples from 1,487 inpatients were transported to clinical microbiology for fecal culture testing and were further screened by performing the meropenem disk screening method.Moreover, 96 CRE strains were detected with positive results, which were defined as the "CRE colonization" group.Meanwhile, we randomly selected 70 CRE clinical isolates from the 210 clinical CRE isolates using the random number generation function in Microsoft Office Excel 2010 (Microsoft Corporation, Redmond, WA, USA), and these CRE strains were isolated from different samples including sputum, blood, urine, and pus, which caused nosocomial infections.These clinical isolates were incorporated for the comparison analysis and defined as the "CRE infection" group.The clinical information of all enrolled inpatients including patient demographics, hospital ward, use of antibiotics, and treatment outcomes was also collected from electronic medical records.All CRE isolates were identified by the matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) system using MALDI Biotyper (Bruker Daltonik GmbH, Bremen, Germany).These isolates were stored at −80 • C in a 40% glycerol broth medium for further analysis.

Antimicrobial susceptibility testing
The minimum inhibitory concentrations (MICs) of antimicrobial agents including cefazolin, cefuroxime, ceftazidime,

Detection of carbapenemase genes
All the CRE isolates were screened for the presence of carbapenemase genes including bla NDM , bla KPC , bla OXA−48 , bla IMP , bla AIM , bla VIM , bla GIM , and bla SIM by polymerase chain reaction (PCR) as previously described (11) (Supplementary Table 1S).The amplicons were sequenced, and nucleotide sequences were further analyzed and compared with the sequences available at the National Center for Biotechnology Information (NCBI) website (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Multilocus sequence typing
Genetic evolutionary relationships were identified for K. pneumoniae, E. coli, and Enterobacter cloacae using multilocus sequence typing (MLST) schemes.The housekeeping genes of K. pneumoniae (ropB, gapA, mdh, pgi, phoE, infB, and tonB) and E. coli (dinB, icdA, pabB, polB, putB, trpA, trpB, and uidA) were performed according to the protocol available on the website of Institute Pasteur MLST (https://bigsdb.pasteur.fr/).Meanwhile, amplification of housekeeping genes for E. cloacae was also conducted as described on the PubMLST website (https://pubmlst.org/).The STs were assigned by sequencing and submitting the sequences to the MLST database for alignment.MEGA 7.0 software was used for Phylogenetic analysis.

Plasmid replicon analysis
For all CRE isolates, the molecular typing of the plasmids was performed using the PCR-based replicon typing, and 17 pairs of primers (HI1, HI2, I1, L/M, N, FIA, FIB, W, Y, P, FIC, A/C, T, FIIAs, F, K, and B/O) were amplified as described previously.
In addition, we used new primers typing IncX groups to IncX1-IncX4 and new primers for ColE-type plasmid (12)(13)(14).The amplified products were electrophoresed in 1.5% agarose gels, stained with Green Nucleic Acid Stain (Sangon, Shanghai, China), and visualized under ultraviolet light using the Gel Doc 2000 system (Bio-Rad).

Detection and sequencing of integrons
The class 1, 2, and 3 integrons were screened by PCR amplification for intI1, intI2, and intI3 genes.Furthermore, amplification of the intI1 variable region was performed using the primers 5 ′ -CS and 3 ′ -CS under the conditions and annealing temperatures as described previously (15,16) (Supplementary Table 1S).To identify different types of gene cassettes, same-sized amplicons were analyzed by restriction fragment length polymorphism (RFLP) with digested RsaI and Hinf I genes (TaKaRa Bio Inc., Tokyo, Japan), which were dependent on the species.Amplicons showing the same pattern were considered to be identical gene cassettes, and then, one representative product of each distinct RFLP was purified and sequenced by Sangon Biotech Co. Ltd (Shanghai, China).The sequencing results were analyzed using the BLAST program on the NCBI website (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Statistical analysis
The statistical analysis was performed using SPSS 25.0 for Windows (version 25.0; SPSS Inc., Chicago, IL, USA).For continuous variables, data were described as median, which was compared using the Mann-Whitney U-test.Multivariate analysis was used to identify the independent predictors, and we calculated the odds ratios (ORs) and the 95% confidence intervals (CIs) for each variable.A value of P ≤ 0.05 was considered statistically significant.

Clinical characteristics of enrolled patients
A total of 96 non-duplicate samples were detected as CRE-positive with a carriage of 6.5% (96 out of 1,487).The clinical characteristics of enrolled patients between the "CRE colonization" group and the "CRE infection" group are presented in Table 1.The CRE strains between the two groups were predominantly isolated from the department of neonatology, with severe underlying diseases including pneumonia, sepsis, and gastrointestinal infections.More than 50% of patients in the "CRE colonization" group and the "CRE infection" group had a history of invasive procedures and antibiotic exposures such as carbapenems (P ≤ 0.05).It is worth noting that 14.6% of CRE carriers in the "CRE colonization" group had CRE infections after the detection of CRE.The death rates between the "CRE colonization" group and the "CRE infection" group were 2.1 and 2.9%, respectively.Multivariate analysis showed that the underlying diseases of gastrointestinal infections (OR = 0.044, 95% CI: 0.006-0.342,P = 0.003) and invasive procedures (OR = 180.081,95% CI: 16.681-1,944.066,P < 0.001; Table 2) are independent risk factors for the colonization/infection of CRE isolates.

FIGURE
The carbapenemases of carbapenem-resistant Enterobacteriales (CRE) strains from intestinal colonization and nosocomial infection.

Discussion
Recently, the CRE isolates have been increasingly identified in children, and their horizontal spread among different species has contributed to a dilemma of no drugs available for clinical antiinfective therapy.Children are considered a special population, and once they are infected with CRE strains, it will greatly limit the choice of antimicrobial agents.Colonization of CRE strains in the gastrointestinal tract implies the potential for nosocomial transmission, and rapid active screening of CRE is essential for reducing and controlling CRE infections.Notably, the proportion of CRE colonization in hospitalized patients ranged from 15.5 to 52% and varied among different countries, different crowds, and different departments (17-20).This study revealed a carriage rate of 6.5%, which was lower than the previous literature.Furthermore, evaluating the special risk factors for the colonization/infection of CRE isolates is crucial for identifying high-risk patients.Consistent with other studies (21,22), invasive procedures such as mechanical ventilation were identified as risk factors for colonization or infection with CRE.Additionally, long hospitalization stays and exposure to antibiotic use also increase the likelihood of crosstransmission of CRE isolates.The diversity of CRE bacterial species was observed in this study, and the most common CRE isolates both in intestinal colonization and nosocomial infection was CRKPN, which is in agreement with the findings of previous studies (23,24).Notably, the rapid increase of CRKPN in recent years is a critical threat to global health and causes high mortality (25).In addition, the proportion of CRECO in the "CRE colonization" group was higher than those in the "CRE infection" group, while the proportion of CRECL in the "CRE colonization" group was lower than those in the "CRE infection" group.It could be speculated that CRECO isolates were easier to detect in fecal samples, but it should be further verified by expanding the number of specimens.CRECL represents the third most common CRE followed by CREKPN and CRECL isolates, and the increased incidence of these isolates has constantly been reported worldwide, which requires more clinical attention (26,27).
The prevalence of carbapenemase types among CRE strains has varied by regions: metallo-β-lactamases (MBLs) including NDM and VIM were dominant in isolates collected in Africa, the Middle East, and Asia/Pacific; KPC was mainly detected in Latin America and North America; and OXA-48-like carbapenemases were predominant in Europe (23).However, a study conducted in China displayed that the most prevalent carbapenemase gene was bla KPC−2 among K. pneumoniae isolates from adult patients and bla NDM among E. coli isolates from children (28).In this study, among CRKPN isolates, KPC-2 producers were the most detected in the "CRE colonization" group, while OXA-232 producers were predominantly found in the "CRE infection" group.Notably, KPC-2-bearing ST11 CRKPN isolates were widely disseminated throughout China (28,29) and had been reported to be colonized in the intestinal tract (30).In recent years, OXA-48-like carbapenemase, including OXA-162 and OXA-181, was increasingly prevalent in CRE strains (31,32), and OXA-232 carbapenemase, a variant of OXA-181 with one amino acid substitution, was found in ST15 K. pneumoniae strains in Shanghai, China (33).Even though in this study, the OXA-48-like carbapenemase was not distributed in colonized CRKPN isolates, the clinicians should still be aware of the emergence of OXA-48-like carbapenemase in the gastrointestinal tract, which was predominantly distributed in fecal colonization (34).Interestingly, the frequency of NDM-carrying isolates has increased worldwide, and NDM-type has spread globally and is detected in various species.Among CRECO strains, NDM-type carbapenemase, including NDM-1 and NDM-5, was frequently detected in the "CRE colonization" group and the "CRE infection" group.Meanwhile, NDM-1-type carbapenemase was the most common among ST45 CRECL isolates both in the "CRE colonization" group and the "CRE infection" group, showing the high similarity of CRECL between colonization and infection.The trend of  carbapenemase types in the colonization/infection of CRE isolates should be further concerned.Mobile genetic elements containing integrons and plasmids highlight their role in the horizontal spread of carbapenemase genes among CRE isolates (35).Other resistant genes in mobile genetic elements also contribute to resistance to antibiotics and result in therapeutic failure of antibiotics.This study revealed that various plasmids were detected in CRE isolates, and no difference in the distribution of plasmids was found between the "CRE colonization" group and the "CRE infection" group.A high incidence of IncF-type plasmid in bla KPC−2 -carrying CRKPN isolates was observed, while IncX type, especially IncX3 and IncX2, was observed in most bla NDM -carrying CRE isolates.The presence of IncF and IncX3 plasmid-mediated the clonal dissemination of carbapenemase genes reported throughout the world (36,37).Furthermore, integrons with classes 1-3 were known as multidrug-resistant integrons that may ultimately contribute to the wide spread of "superbugs" including CRE (38).A high frequency of class 1 integrons was identified in the "CRE colonization" group (74.0%) and the "CRE infection" group (52.9%), which is in agreement with an earlier report (16).Urgent surveillance of mobile genetic elements among CRE isolates should be conducted to control the spread of these pathogens.
Our study still has several limitations.First, this study was performed in one hospital, and the results were not suitable for other institutions.Therefore, multicenter design and comprehensive capture of the colonization/infection of CRE isolates should be further conducted.Second, even though we recommended CRE screening for all hospitalization patients, the active screening was still not frequently performed for all inpatients.The fecal sample was collected from hospitalized children who received the routine fecal culture testing.It might have led to the low CRE-positive rate.Finally, the sample size was relatively small and could not fully reflect the prevalence of intestinal colonization and nosocomial infection of CRE strains.

Conclusion
In conclusion, this study described the prevalence of intestinal colonization and nosocomial infection with CRE isolates in children, and KPC-2, NDM-1, NDM-5, and OXA-232 carbapenemases were the main types found in the "CRE colonization" group and the "CRE infection" group.Then, a few intestinal colonized CRE strains were probably linked to clinically infected strains, especially for ST11 KPC-2-producing CRKPN and ST45 NDM-1producing CRECL.Therefore, in the case of the outbreak of these high-risk CRE clones from intestinal colonization, we should reinforce the surveillance of CRE from intestinal colonization and long-term active screening of rectal colonization of CRE isolates before admission should also be emphasized.

FIGUREFrontiers
FIGURE Phylogenetic tree of MLST data among CRE isolates analyzed by MEGA .(A) Phylogenetic tree of carbapenem-resistant Klebsiella pneumoniae; (B) Phylogenetic tree of carbapenem-resistant Escherichia coli; (C) Phylogenetic tree of carbapenem-resistant Enterobacter cloacae.
TABLE Clinical characteristics of patients with CRE positive in this study.
* A comparison between two groups.A P-value of ue.A on between inhibitorstically significant.CRE, Carbapenem-resistant Enterobacteriales, PICU, pediatric intensive care unit; UTI, urinary tract infection.
TABLE Multivariate analysis for factors associated with CRE.