Risk Factors and Prevalence of mcr-1-Positive Escherichia coli in Fecal Carriages Among Community Children in Southern Taiwan

Colistin is the last resort antimicrobial for treating multidrug-resistant gram-negative bacterial infections. The plasmid-mediated colistin resistance gene, mcr-1, crucially influences colistin’s resistance transmission. Human fecal carriages of mcr-1-positive Escherichia coli (E. coli) were detected in many regions worldwide; however, only a few studies have focused on children. Therefore, we identified the prevalence and risk factors of mcr-1-positive E. coli in fecal carriages among community children in Southern Taiwan. In this study, 510 stool samples were collected from April 2016 to August 2019 from the pediatric department at a medical center in Southern Taiwan. These samples were collected within 3 days after admission and were all screened for the presence of the mcr-1 gene. Diet habits, travel history, pet contact, and medical history were also obtained from participants to analyze the risk factors of their fecal carriages to mcr-1-positive E. coli. Antimicrobial susceptibility testing was determined using the VITEK 2 system and the broth microdilution test. Twelve mcr-1-positive E. coli. were isolated from 2.4% of the fecal samples. Through multivariate analysis, frequent chicken consumption (at least 3 times per week) had a significantly positive association with the presence of mcr-1-positive E. coli in fecal carriages (adjust odds ratio 6.60, 95% confidence interval1.58– 27.62, p = 0.033). Additionally, multidrug resistance was more common in mcr-1-positive E. coli. (75.0% vs. 39.5%, p = 0.031) than in non-mcr-1-positive Escherichia coli. Furthermore, the percentage of extraintestinal pathogenic E. coli in mcr-1-positive isolates was 83.3%. Some multi-locus sequence types in our mcr-1-positive E. coli were also similar to those isolated from food animals in the literature. The prevalence of fecal carriages of mcr-1-positive E. coli was low among community children in Southern Taiwan. Our data shows that chicken consumption with a higher frequency increases the risk of mcr-1-positive E. coli. in fecal carriages.


INTRODUCTION
Antibiotic resistance is a global public health challenge of our time. Colistin (polymyxin E), first discovered in 1947, is a polycationic peptide antibiotic (Storm et al., 1977). It mainly targets the bacterial cell membrane and increases its permeability, which results in the leakage of cell contents, thereby ultimately causing cell death . Furthermore, it has broad-spectrum antimicrobial activities, and it is used to treat multidrug-resistant gram-negative infections (Poirel et al., 2017). Before 2015, studies showed that colistin resistance was regulated by mutation in chromosomal genes (Olaitan et al., 2014). However, in 2016, Liu et al. (2016) discovered the plasmidmediated colistin resistance gene-mcr-1. Since then, the mcr-1 gene has been identified as part of several bacterial species in humans, animals, and the environment, thus, posing a great threat to treatment (Elbediwi et al., 2019).
The human gut is a reservoir of antimicrobial resistance genes. In the gut, antimicrobial resistance genes can spread through horizontal gene transfer (Rolain, 2013;McInnes et al., 2020), which plays an important role in transmitting drug-resistant bacteria. Previous studies have also reported that the prevalence of mcr-1-positive Enterobacteriaceae in fecal carriages was 0.4-15.0% in Asia (Chan et al., 2018;Shen et al., 2018;La et al., 2019;Wu et al., 2019), 0-0.35% in Europe (Terveer et al., 2017;Zurfluh et al., 2017b), and 38.3% in South America (Giani et al., 2018), showing great variability in different geographic areas and populations. In the aspects of risk factors, several literatures from China reported antibiotic use before admission, the consumption of meat and aquaculture products were linked to mcr-1-positive Enterobacteriaceae in fecal carriages (Wang et al., 2017b;Shen et al., 2018). Furthermore, traveling to Southeast Asia has also been associated with mcr-1-positive Escherichia coli (E. coli) in fecal carriages (Nakayama et al., 2018). Despite these results, most studies on the risk factors for mcr-1 Enterobacteriaceae in fecal carriages were from the adult population, and studies among children were scarce. Furthermore, E. coli is a common inhabitant of the human intestinal tract and has the highest mcr prevalence among all bacterial species (Elbediwi et al., 2019). Therefore, we conducted a prospective study to identify the prevalence and risk factors of mcr-1-positive E. coli in fecal carriages of community children in Southern Taiwan.

Study Population and Data Collection
This prospective study was conducted from April 2016 to August 2019 at the pediatric department of Kaohsiung Veterans General Hospital in Southern Taiwan. Children aged below 18 years who were admitted at the pediatric department were enrolled. We contacted them as soon as they were available after admission. If they agreed to participate in the study after an interview, we described how their stool samples would be collected, after which they were given a standardized questionnaire that included demographic data, hospitalization in the last 3 months, antibiotic use in the last 3 months, history of traveling abroad, dietary habits, intake of unboiled drinking water, and history of pet contact. All participants and their legal guardians were then provided with an informed written consent. Exclusion criteria were non-willing participants, those unwilling to complete the informed written consent, those without fecal samples within 3 days after admission, and those with incomplete data in their questionnaire. The Ethics Committee of the Kaohsiung Veterans General Hospital (VGHKS 16-CT2-04, VGHKS 18-CT3-11) approved this study.
Isolation of E. coli and Detection of the mcr-1 Gene Fecal samples were collected using cotton swabs, and then the swabs were transported in a COPAN Transystem R (Copan Diagnostics, Inc., Brescia, Italy). Subsequently, the swabs were plated on a CHROMagar TM ECC plate (CHROMagar, Paris, France) and incubated without CO 2 at 35-37 • C for 24 h. E. coli colonies presented blue coloration, and up to two E. coli colonies were randomly selected for further analysis. Furthermore, all selected E. coli strains were screened for mcr-1 genes. We used polymerase chain reaction with the primers; CLR5-F (5 -CGGTCAGTCCGTTTGTTC-3 ) and CLR5-R (5 -CTTGGTCGGTCTGTA GGG-3 ) for mcr-1 gene screening (Liu et al., 2016). Additionally, all selected E. coli strains were spread on a CHROMagar TM ESBL plate (CHROMagar, Paris, France) to recognize the extended-spectrum β-lactamase (ESBL) E. coli isolates. We also checked for ESBL genes and carbapenemase genes, including; bla CTX -M , bla SHV , bla TEM , bla OXA-1 , bla IMP , and bla VIM using polymerase chain reaction (PCR) with previously described primers and methods (Chia et al., 2005;Sidjabat et al., 2009;Wu et al., 2019). Moreover, we used polymerase chain reaction to identify whether mcr-1positive E. coli isolates belonged to extraintestinal pathogenic E. coli (ExPEC). Therefore, ExPEC was defined as E. coli isolates harboring at least two of the following five genes: S and F1C fimbriae (focG + sfaS), kpsM II, papA, afa, and iutA (Johnson et al., 2003). The E. coli multilocus sequence typing (MLST) scheme was also determined among mcr-1-positive E. coli isolates using seven housekeeping gene sequences (adk, fumC, gyrB, icd, mdh, purA, and recA) (Chia et al., 2005). Then, we used the goeBURST algorithm 1 to evaluate the genetic relatedness of mcr-1-positive E. coli (Francisco et al., 2009).

Statistical Analysis
We used SPSS version 20.0 for Windows (SPSS Inc., Armonk, NY, United States) to perform our statistical analyses. For univariate analysis, Chi-square or Fisher's exact tests were used to analyze categorical variables. An independent t-test was also used to analyze continuous variables. Furthermore, variables with a p-value of <0.1 in the univariate analysis were assessed using binary logistic regression analysis in the multivariate analysis. A p-value of<0.05 was considered statistically significant.

RESULTS
We collected 510 non-duplicated fecal specimens from patients in our pediatric ward. Among these specimens, 268 yielded 514 E. coli isolates. The characteristics of these 510 participants are demonstrated in Table 1. The participants' ages ranged from 1 day to 17 years (mean age was 1.5 years), and males accounted for 58.6% (299/510). Hospitalization and antibiotic therapy in the last 3 months were recorded in 9.6% (49/510) and 7.8% (40/510) of participants, respectively. In addition, 7.5% (38/510) of the participants had traveled abroad in the last 12 months. Among them, 10 had been to China, 23 to Japan, 2 to Korea, 5 to Southeast Asian nations, and one to the United States. The prevalence of fecal carriages of mcr-1-positive E. coli was 2.4% (12/510). Looking further, we divided the study period into three segments. The prevalence was 1.6% (2/124) from April 2016 to March 2017, 2.5% (8/324) from April 2017 to March 2018, and 3.2% (2/62) from April 2018 to August 2019. Subsequently, we compared the variables between individuals with and without fecal carriages of mcr-1-positive E. coli through univariate analysis. Individuals with fecal carriages of mcr-1positive E. coli were predominantly male [Odds Ratio (OR) 8.02, 95% confidence interval (CI) 1.03-62.61] and people frequently ate chicken (OR 3.13, 95% CI 0.77-12.72), with a consumption of between 1 and 3 times per week. Another set of included individuals (OR 7.10, 95% CI 1.72-29.35) had a chicken consumption frequency of at least 3 times per week. Furthermore, antibiotic use in the past 3 months was borderline associated with mcr-1-positive E. coli carriage (OR 4.15, 95% CI 1.08-16.00) ( Table 1). Through binary logistic regression analysis, chicken consumption with a higher frequency (least 3 times per week) was the only factor that had a significant association with fecal carriages of mcr-1-positive E. coli (adjust OR 6.60, 95% CI 1.58-27.62, p = 0.033) ( Table 1).
The distribution of reasons for admission between participants with and without mcr-1-positive E. coli in fecal carriages are presented in Table 2, showing that individuals that were hospitalized due to infectious diseases had a higher proportion of mcr-1-positive E. coli in their fecal carriages than those hospitalized due to non-infectious diseases (3.6 and 0.9%, respectively, p = 0.041), and individuals hospitalized due to urinary tract infection (UTI) had the highest proportion (7.4%) of mcr-1-positive E. coli in their fecal carriages.

DISCUSSION
The prevalence of mcr-1-positive E. coli in fecal carriages of patients included in the study was 2.4% (12/510), which was lower than what had been reported in children (4-38.3%) from either community or hospital settings in other countries, such as China, Bolivia and India (Bi et al., 2017;Hu et al., 2017;Giani et al., 2018;Purohit et al., 2019). However, the results were similar to other recent studies in Taiwan, including data from clinical isolates and commensal E. coli Wu et al., 2019). These findings demonstrated that the prevalence of the mcr-1 gene was still low in Taiwan. However, an upward trend of the prevalence over time was found. Furthermore, in this study at the individual level, a significant predictor of fecal carriages in mcr-1-positive E. coli was chicken consumption. A study from China of approximately 90% adults also demonstrated that meat consumption, especially pork and mutton, was associated with mcr-1 prevalence in a provincial region's level (Shen et al., 2018). Another epidemiological study from southern China using vegetarian and non-vegetarian classifications did not find diet as a risk factor for mcr-1 in fecal carriages (Wang et al., 2017b). Additionally, previous studies revealed a correlation of drug-resistance bacteria between contaminated food and the human gut. Sørensen et al. (2001) observed the same resistant strains in stool after ingestion of chickens containing those resistant bacteria. Donabedian et al. (2003) also observed related pulsed-field gel electrophoresis patterns of gentamicin-resistant isolates from human stools and meat. A recent review article indicated that foodborne transmission was a pathway of mcr-1-positive E. coli transmission (Elbediwi et al., 2019). Literature also revealed that chicken had a higher percentage of the mcr-1 gene than other meats (Kuo et al., 2016). Therefore, chicken consumption is a possible pathway to transmit mcr-1-positive E. coli to humans. Nevertheless, more studies should investigate the detailed mechanisms. Traveling abroad was proposed as a risk factor for drugresistant gene transmission (Hu et al., 2020). Limited studies on the relation between traveling abroad and fecal carriages of mcr-1-positive E. coli have been reported. von Wintersdorff et al. (2016) and Nakayama et al. (2018) demonstrated that traveling to Southeast Asia or southern Africa increases the risk to mcr-1positive E. coli in fecal carriages. However, traveling abroad was not associated with mcr-1-positive E. coli in fecal carriages in this study. The period between the time of travel and fecal collection may affect the results. In those two studies, fecal samples were collected within no more than 3 weeks after travel events. In the study presented here, travel history was traced back to 12 months before the fecal sample collection. Previous studies revealed that the proportion of drug-resistant bacterial colonization declined as time after international travels increased. For example, only about 10% of fecal colonization of ESBL E. coli cases remained positive after 12 months follow-up (von Wintersdorff et al., 2016;Arcilla et al., 2017;OstholmBalkhed et al., 2018). Additionally, people in Taiwan have more opportunities of contacting people from China and Southeast Asia due to their geographic location, which may weaken the influence of traveling to these countries. Therefore, more studies should investigate the relationship between international travels and mcr-1-positive E. coli in fecal carriages.
A higher proportion of participants who were hospitalized due to UTI were positive for mcr-1-positive E. coli in the study. The association between UTI pathogens and the gut microbiome had FIGURE 1 | goeBURST diagram showing the genetic relatedness of mcr-1-positive E. coli isolates from the study and food animals reported in the literatures of Taiwan (Kuo et al., 2016;Liu J. Y. et al., 2020). The number within the node is the ST. Node size is based on the number of isolates with specific STs. ST of isolates in the study (from fecal samples of children) is marked with a dark blue frame. However, isolates with the same ST from the study and food animals of Taiwan reported in the literatures are marked with a red frame. Others without frame markings are ST of isolates from food animals in the literature of Taiwan. (Node color: light green: group founder; light blue: common node).
been reported before (Moreno et al., 2008;Paalanne et al., 2018). Paalanne et al. (2018) revealed that the gut microbiome was associated with the risk of febrile UTI in children. However, we did not collect urine data from the participants in the study. The relevance of UTI pathogens and fecal carriages of mcr-1positive E. coli thus needs further study. The four participants with UTI with mcr-1-positive E. coli isolates were under 5 years. Furthermore, two of them had a history of UTI and antibiotic exposure in the last 12 months. Except for the undefined one, the other three isolates belonged to ExPEC. Interestingly, the positive rate of mcr-1-positive E. coli presence in the fecal carriage of participants with acute gastroenteritis was low (1.7%, 1/58).
Some limitations exist in this study. First, the diet information relied on self-reports provided by the participants' parents or caregivers. Recall bias is inevitable. Second, the number of participants with mcr-1-positive E. coli in fecal carriages was relatively small, which would not detect certain risk factors. Third, mcr-1 gene is detected mainly from Enterobacteriaceae. Although E. coli has the highest mcr prevalence among all bacterial species (Elbediwi et al., 2019), there are still some common bacterial species with mcr gene, such as Klebsiella pneumoniae, that have not been analyzed in the study. Furthermore, the study was limited to mcr-1 gene, which is the most widely disseminated mcr gene worldwide (Ling et al., 2020). However, other mcr variants may exist. Last, the diversity of STs of mcr-1-positive E. coli not only means clonal expansion but also the possibility of certain plasmid transmission in the community. In our preliminary result, some of the plasmid in the mcr-1 positive E. coli had the same sequence as the plasmid identified in food animals, as reported by studies published in Taiwan (unpublished data).
In conclusion, the prevalence of mcr-1-positive E. coli in fecal carriages is low among community children in Southern Taiwan who were included in this study. Nevertheless, we can found a gradual upward trend of the prevalence from 2016 to 2019. This phenomenon shows the need for continuous follow-up. A higher frequency of chicken consumption was also associated with fecal carriages of mcr-1-positive E. coli. However, the genetic relationship between our colonized isolates and isolates from chicken in Taiwan warrants further evaluation.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by VGHKS 16-CT2-04, VGHKS 18-CT3-11. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.