Characteristics of diarrheagenic Escherichia coli among children under 5 years of age with acute diarrhea: a hospital based study

The definition of diarrhea was as at least 3 abnormal appearance stools (loose, watery, mucus or bloody) in 24 h, with at least one of the following symptoms: nausea, vomiting, abdominal pain, or fever above 37.2 °C. The diarrhea which lasted 14 days or less at the time of presentation was defined as acute diarrhea, otherwise, was defined as persistent diarrhea. Persistent diarrheal Children were excluded from the present study [16].

From August 2015 to September 2016, stool samples from acute diarrheal children under 5 years of age were collected at Tongji hospital (the largest teaching hospital in central China, which has more than 4000 beds and treats patients from the six surrounding provinces) [19]. All samples were collected under the parents’ or legal guardians’ permission. Demographic information for each patient, such as age, sex and clinical symptoms were collected.

All stool samples collected were processed by routine microbiological tests to identify. Briefly, MacConkey (Mac) agar were used to isolate the pathogens and incubated for 24 h at 37 °C. Three suspicious colonies with E. coli morphology (including lac + or lac-) were selected from Mac agar plates and all of them were identified by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) using the MALDI Biotyper (Bruker Daltonik GmbH, Leipzig, Germany). Biochemical tests were carried as supplements for E. coli identification. O157:H7 was screened by sorbitol-Mac.

DEC was characterized by PCR as previously [19]: tEPEC (eae and bfp), aEPEC (eae or bfp), STEC (eae and stx1 and/or stx2), ETEC (elt and/or estIa or estIb), EIEC (virF and ipaH) and EAEC (aggR and/or pic or astA). The PCR assay was carried out as follow: Boiling method was used for template DNA preparation. PCR was performed in 20 μl final volume. The reactions were run with the following cycling conditions: 94 °C for 2 min, 35 cycles of denaturation at 95 °C for 20 s, annealing at 60 °C for 30 s and primer extension at 72 °C for 45 s followed by a final extension at 72 °C for 5 min. EPEC CMCC44155, ETEC CMCC44815, EIEC CMCC44825 (from the National Institute for the Control of Pharmaceutical and Biological Products in China); EAEC serotype O42, STEC EDL933(from Chinese Center for Disease Control and Prevention) were served as the positive controls. E. coli DH5α, which lacks all the diarrhoeagenic genes, was used as a negative control.

Carbapenemase genes (blaKPC-2, blaGES, blaIMP-4, blaVIM-1, blaNDM-1and blaOXA-48) were screened in carbapenem-resistant DEC strains (MIC ≥ 4 ug/ml to imipenem or meropenem). Positive strains were further studied by sequencing. Strains showing significantly decreased susceptibility to ceftazidime or cefotaxime (MIC ≥ 32 ug/ml) were further studied by PCR amplification and sequencing of the extended-spectrum β-lactamase genes (ESBL genes, including blaSHV, blaTEM and blaCTX-M). Isolates showing high resistance to quinolones (MIC ≥ 32 ug/ml to ciprofloxacin or levofloxacin) were further tested for amino acid changes in the plasmid-mediated quinolone-resistant genes gyrA and gyrB, according to the methods reported by Yenkao et al. [21]. Quality control came from the strains identified by sequencing before. GenBank database (http://​www.​ncbi.​nlm.​nih.​gov/​) was employed to confirm the subtypes of antibiotic resistance genes.

Between August 2015 and September 2016, 684 stool samples were collected from diarrheal children under 5 years of age. In the population studied, most children (52.0%) were less than 24 months, 14.7% were 24–35 months, 16.9% were 36–47 months, and 16.4% were 48–59 months old, respectively. Boys accounted for 61.6% of the study population. Most patients were admitted in the summer (36.8%), followed by spring (23.5%), winter (21.6%), and autumn (18.0%). About half of the admitted children with acute diarrhea showed abdominal pain, followed by fever and vomiting (Details shown in Table 1).

Among 684 collected stool samples, 54 were positive for DEC (overall prevalence 7.9%). The most frequent pathotype was EPEC (50.0%), followed by EAEC (20.4%), ETEC (14.8%), EIEC (3.7%), and STEC (3.7%). Meanwhile, the remaining 7.4% cases were co-infected with more than one DEC pathotype (1 tEPEC and EAEC; 3 aEPEC and ETEC). Notably, among EPEC-infected cases, aEPEC accounted for 77.8%, while tEPEC accounted for just 22.2% (Fig. 1). Clinical data were available for 54 children with single or mixed DEC infection (shown in Additional files 1, 2 and 3). Fever was observed as the most frequent symptom (51.9%) among DEC-infected children, followed by vomiting (25.9%), abdominal pain (11.1%), bloody diarrhea (7.4%), and nausea (3.7%). Two children infected with non-O157:H7 STEC, both had bloody stools and showed severe symptoms of diarrhea. Worse symptoms were also discovered in children infected with more than one DEC pathotype.

Demography analysis showed that the younger the child was, the more prone to infection by DEC (chi-squared test, P < 0.001). When age stratification was done, the frequencies of DEC diarrheal episodes occurring in different age groups were 17.5% (0–11 months), 6.3% (12–23 months), 5.0% (24–35 months), 5.2% (36–47 months), and 1.8% (48–59 months) (Fig. 2). The isolate rate of DEC and the isolate number in subgroups by seasonality are shown in Fig. 3. The isolation rate of DEC showed a distinct seasonal variation, with a higher rate in the summer (7.1%) and autumn (17.1%) months (chi-squared test, P = 0.002). In addition, EPEC, the most dominant pathotype, tended to occur more in children less than 24 months of age (66.7%) and in the summer/autumn period (81.5%).

Among the 27 EPEC strains, only 11 (40.7%, 11/27) strains belonged to the classic EPEC serogroups, accounting for 33.3% (7/21) of aEPEC strains and 66.7% (4/6) of tEPEC strains. Furthermore, 22.2% strains belonged to serotype O86:k61; 11.1%, to O55:K59; and 7.4%, to O125:K70.

The observed prevalence of resistance to 16 antibiotics amongst DEC are shown in Fig. 4. The highest resistance rate was detected for ampicillin (77.8%), followed by 64.8% for co-trimoxazole, 59.3% for cefazolin and cefuroxime; 57.4% for cefotaxime; 50.0% for ciprofloxacin; and ≥ 30% for gentamicin, aztreonam, levofloxacin, cefepime, and ceftazidime. For other antibiotics tested, cefoxitin, amikacin, piperacillin/tazobactam, imipenem, and meropenem showed efficacy against most DEC strains, with the resistance rate being < 30%. Although more than 90% DEC showed sensitivity to carbapenems, there was a 9.3% resistance rate. In addition, 36 (66.7%) isolates of DEC were multidrug-resistant, and 1 ETEC isolate showed resistance to all the antibiotics tested; this highlights the increasing trend of extensively drug-resistant bacteria. aEPEC, sharing high prevalence among DEC-infected patients, also had a high resistance rate, e.g., 81.0% for ampicillin, 66.7% for co-trimoxazole, and 14.3% for carbapenems (Fig. 4).

Among 5 (9.3%) carbapenem-resistant DEC strains, 3 (60%) isolates were positive for carbapenemase genes, with 2 harboring blaNDM-1 and 1 harboring blaKPC-2 (Table 2). One EIEC strain, harboring blaKPC-2, blaCTX-M-65, and blaTEM-1, showed resistance to most antibiotics tested, and was sensitive only to amikacin. Both blaNDM-1-positive strains showing high resistance to imipenem and meropenem (MIC ≥128 ug/ml) were aEPEC. There were 30 (55.5%) DEC isolates displaying high resistance to ceftazidime or cefotaxime (MIC ≥32 ug/ml). Among them, 93.3% were positive for ESBL genes, with blaTEM-1 (43.3%) and blaCTX-M-55 (33.3%) being the most common types. Positive rates of other ESBL genes detected were 20.0% for blaCTX-M-15, 10.0% each for blaCTX-M-14 and blaTEM-214, and 3.3% each for blaCTX-M-65 and blaCTX-M-137. blaAmpC genes and other carbapenemase genes were not observed. In addition, 16 (29.6%) isolates of DEC were found to be highly resistant to ciprofloxacin or levofloxacin (MIC ≥32 ug/ml), but no gyrA or gyrB genes were observed. Notably, aEPEC strains were found to harbor drug-resistance genes at a high rate. Among 3 carbapenemase-positive DEC strains, 2 strains positive for blaNDM-1 were aEPEC, accounting for 66.7% of carbapenem-resistant aEPEC isolates. Meanwhile, 52.4% of cephalosporin-resistant aEPEC strains were positive for ESBL genes.

The prevalence of DEC in our study was 7.9% (54/684), lower than reports from other developing countries [8, 22] but similar to the results of studies in China [4, 19, 23]. This suggests the influence of different regions in the distribution of DEC. The primary pathotypes were EPEC (54.0%) and EAEC (22.0%), corresponding with the results reported by Wang et al. [15]. EPEC, first named in 1995 by Neter et al. [24], has been described as the most frequent DEC pathotype in many developing countries [4, 25, 26]. Our study reflected that EPEC constituted 54.0% of DEC isolates, a little higher than that reported by Yu et al. [4], which supported the need for follow-up epidemiological studies in childhood diarrhea. EAEC strains have been associated with traveler’s diarrhea in both developing and industrialized countries [27, 28]. Studies carried out in Brazil and Mexico revealed EAEC to be the primary pathotype, with respective rates of 50% [12] and 52.1% [6]. In the present study, EAEC (20.5%) ranked as the second most common DEC associated with infected children, reflecting the difference in distribution across geographical areas. STEC, a subgroup of DEC strongly related to severe human illnesses [29], continued to be uncommon. However, the infection rate was a little higher (4.0%) in the present study compared to 0.4% in Beijing [23] and 0% in Shanghai [7]. Although no O157:H7 was observed, both STEC-infected children in this study presented with bloody stools. These findings should caution clinicians to carefully monitor the prevalence of STEC in children with severe diarrhea. Consistent with the findings of Patzi-Vargas et al. [25], EIEC was observed at a very low frequency (4.5%).

When age stratification was analyzed, the infection rate of DEC was found to decrease with age (P < 0.05), consistent with a report by Gomes et al. [13]. The lower prevalence in older children might be attributed to age-related immunity, which has also been observed in other studies [22, 25]. Generally speaking, EPEC is among the most important pathogens infecting children under 2 years of age in the developing country [6, 30]. Supporting this view, in the present study, 66.7% of EPEC infection cases were of children less than 24 months old. Seasonal variation was also found in DEC infection, especially in the EPEC group, which occurred most frequently in the late summer/early autumn period (P < 0.05). Similar seasonal patterns have also been observed in earlier studies [15, 25], indicating that DEC infection is strongly related to environmental factors such as temperature and humidity.

The resistance rates of DEC to first-line therapeutic drugs were high, e.g., 77.8% to ampicillin and 64.8% to co-trimoxazole, higher than the rates we reported before [19, 31] but lower than those reported by Chen et al., where 91.8% DEC were resistant to ampicillin [32]. In this study, 36 (66.7%) DEC isolates were multidrug-resistant, comparable with 70.2% in the study by Chen et al. [32]. By molecular analysis, the ESBL genes blaTEM-1 and blaCTX-M-55 were the genetic determinants responsible for resistance to cephalosporins. ESBL-producing strains have been reported to be recently changing from the blaTEM or blaSHV type to blaCTX-M [18], which was consistent with our findings. blaCTX-M-15 and blaCTX-M-14 have been the most common cephalosporin-resistant genotypes isolated from humans [33], while blaCTX-M-55 was the dominant type in the present study, followed by blaCTX-M-15 and blaCTX-M-14. The molecular characterization of the isolates suggested that blaCTX-M-55 is most closely related to blaCTX-M-15, with only a single amino acid substitution, indicating that blaCTX-M-55 might be just a derivative of blaCTX-M-15 [34]. In 2014, a nationwide investigation of ESBL- and AmpC-producing E. coli first reported that the incidence of blaCTX-M-55 exceeded that of blaCTX-M-15 in China [35], which warns of the prevalence of new variants of blaCTX-M. At present, carbapenems are the first option to treat ESBL-resistant strains, but the resistance rate of carbapenems (9.3%) presented a rising trend, with all the DEC showing sensitivity to imipenem in our earlier study [19, 31]. Carbapenem resistance in the Enterobacteriaceae is mainly attributed to the production of carbapenemases, the most common one being blaKPC [36] and the predominant one in China being blaKPC-2 [37, 38]. In our study, both blaKPC-2 and blaNDM-1 were found to be associated with carbapenem resistance among DEC. blaNDM-1, first reported in 2008, demonstrated a current and pressing example of the rapidity with which it could disseminate globally [39]. Recently, NDM-1-producing Klebsiella pneumoniae was detected from the neonatal ward in our hospital [40], which reflected the spread of this resistance gene in China. In contrast with a published finding [19], mutations in the plasmid-mediated quinolone resistance genes gyrA and gyr B were not the major reason for resistance to quinolone antibiotics in the present study, suggesting that some other resistance mechanism might exist, e.g., overexpression of efflux pumps and decreased expression of outer membrane porins.

Interestingly, aEPEC was found to be the crucial subtype of EPEC, accounting for 77.8% of EPEC strains and 42.0% of DEC strains. Although the association of aEPEC with diarrhea is still controversial [6], recent epidemiological studies have suggested an increasing identification of aEPEC in both developed and developing countries [28, 41], with some strains leading to diarrheal outbreaks [42]. A study from 13 developing countries showed that aEPEC isolates were responsible for 78% (131/169) of EPEC cases in children [43], which was similar with our study. Nonetheless, many existed studies lack the discrimination between tEPEC and aEPEC [4, 15, 16], which have made parallel contrast among different areas unavailable. Our study indicated aEPEC to be dominant and closely related with diarrhea among children, which has rarely been reported in China so far.

Hernandes et al. indicated that approximately 81% of the reported aEPEC strains did not belong to the classical EPEC serogroups, and 26.6% of them were O non-typeable [44]. In the present study, only 33.3% (7/21) of aEPEC strains belonged to the classical EPEC serogroups, indicating that serotyping might fail to detect many aEPEC strains. However, serotyping is still frequently used in many clinical laboratories, including in China. This raises concerns regarding correct diagnosis and might need to be addressed urgently.

aEPEC strains identified in Mexico were shown to possess high antibiotic resistance [6]. Similarly, aEPEC strains isolated from a food-poisoning outbreak in China were found to have high multidrug resistance, including high resistance to both quinolones and extended-spectrum cephalosporins [45]. In our study, aEPEC strains have developed resistance to many commonly used clinical drugs. Resistance rates of aEPEC to quinolones and extended-spectrum cephalosporins were all >30%. Furthermore, 2 blaNDM-1 positive strains were both aEPEC strains. This implied that the drug resistance rate was high among aEPEC strains, warranting more attention to be focused on this critical aspect.

This study also had several limitations. Some DEC pathotypes were too few to demonstrate any association with age and seasonal patterns, which had been shown in other studies [15, 25]. Hence, larger sample size, extensive coverage area, and longer monitoring time are needed to yield an overall picture of DEC prevalence in childhood diarrhea. What is more, DAEC, another less well-defined pathotype, was not detected in our study for the difficulties in its identification and classification.