State Key Laboratory for Diagnosis and Treatment of Infectious Diseases, the First Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou, Zhejiang, China
Corresponding author: L. Li, State Key Laboratory for Diagnosis and Treatment of Infectious Diseases, the First Affiliated Hospital, College of Medicine, Zhejiang University, No. 79, Qingchun Road, Hangzhou, 310003, China
Diarrhoeagenic Escherichia coli (DEC) infection is a major health problem in developing countries. The prevalence and characteristics of DEC have not been thoroughly investigated in China. Consecutive faecal specimens from outpatients with acute diarrhoea in nine sentinel hospitals in southeastern China were collected from July 2009 to June 2011. Bacterial and viral pathogens were detected by culture and RT-PCR, respectively. DEC isolates were further classified into five pathotypes using multiplex PCR. The O/H serotypes, sequence types (STs) and antimicrobial susceptibility profiles of the DEC isolates were determined. A total of 2466 faecal specimens were collected, from which 347 (14.1%) DEC isolates were isolated. DEC was the dominant bacterial pathogen detected. The DEC isolates included 217 EAEC, 62 ETEC, 52 EPEC, 14 STEC, one EIEC and one EAEC/ETEC. O45 (6.6%) was the predominant serotype. Genotypic analysis revealed that the major genotype was ST complex 10 (87, 25.6%). Isolates belonging to the serogroups or genotypes of O6, O25, O159, ST48, ST218, ST94 and ST1491 were highly susceptible to the majority of antimicrobials. In contrast, isolates belonging to O45, O15, O1, O169, ST38, ST226, ST69, ST31, ST93, ST394 and ST648 were highly resistant to the majority of antimicrobials. DEC accounted for the majority of bacterial pathogens causing acute diarrhoea in southeastern China, and it is therefore necessary to test for all DEC, not only the EHEC O157:H7. Some serogroups or genotypes of DEC were highly resistant to the majority of antimicrobials. DEC surveillance should be emphasized.
Diarrhoea caused by Escherichia coli (E. coli) is a major health problem in developing countries. Diarrhoeagenic E. coli (DEC) strains are divided into six pathotypes: enteropathogenic E. coli (EPEC), Shiga toxin-producing (or enterohemorrhagic) E. coli (STEC), enterotoxigenic E. coli (ETEC), enteroaggregative E. coli (EAEC), enteroinvasive E. coli (EIEC) and diffusely adherent E. coli (DAEC) [1, 2]. Global epidemics and outbreaks of all DEC types except DAEC are frequently reported . From May to July of 2011, a large-scale outbreak of DEC caused 3816 documented infections (including 54 deaths) in Germany, 845 of which (22%) involved haemolytic uraemic syndrome (HUS). The outbreak strain was characterized as an extended-spectrum β-lactamase (ESBL)-expressing E. coli strain of serotype O104:H4 with virulence factors of both EHEC and EAEC . China has always been a hotspot of such epidemics. For example, diarrhoeal epidemics caused by E. coli O157:H7 were reported in Jiangsu and Anhui provinces in 1999, leading to 20 000 cases of infection and 195 cases of HUS, with 177 deaths . DEC, the major pathogen causing diarrhoea , is primarily spread by water and food and grows in warm, moist places. Although southeastern China has always been a hotspot for diarrhoea, with potential risks of DEC epidemics, there is a lack of DEC surveillance data. In this study, we chose the southeastern region as a representative region for diarrhoeal epidemics in China to investigate the epidemiological and aetiological characteristics of DEC.
Materials and Methods
Nine hospitals, including seven general hospitals, one children's hospital and one community hospital, located in different areas of southeastern China were selected as surveillance sites (Fig. S1). The subjects were outpatients with acute diarrhoeal disease (defined as three or more watery or loose stools during a 24-h period with a duration ≤14 days). The sentinel hospitals administered patient questionnaires, collected and packaged faecal specimens, isolated and identified the bacteria and stored the isolates. The specimens were frozen at −20°C, and the isolates were stored at −80°C in trypticase soy broth (TSB) containing 20% glycerol. The preserved specimens and isolates were delivered to our laboratory on dry ice for further identification every 2 weeks. The questionnaires covered demographic characteristics, illness symptoms, the results of routine stool tests and medications taken before the visit. The faecal specimens were collected after informed consent was obtained from the patients.
Stool culture and phenotypic identification
Two aliquots of each faecal specimen were made immediately upon arrival at the laboratories of the sentinel hospitals. The first aliquot was plated on selective media for enteric pathogens, passaged in enrichment broth and subcultured after overnight incubation at 35°C. The identities of all isolates were confirmed using the Vitek-2 Compact bacterial identification system (bioMerieux, Marcy l'Etoile, France). The second aliquot was tested for the presence of viruses. Rotavirus was detected using a commercial ELISA kit (IDEIA™ Rotavirus; DAKO Ltd, Ely, Cambridgeshire, UK). Calicivirus, astrovirus and adenovirus were detected using reverse-transcriptase PCR and real-time PCR. Five lactose-fermenting colonies suggestive of E. coli were analysed from each MacConkey agar plate for each subject. DEC isolates were further classified into five pathotypes by multiplex PCR. The target genes, primer sequences, PCR conditions, amplification parameters and interpretation of the results were in accordance with previous methods described by Müller et al. . EPEC strain 44155 (escV+, bfpB+), STEC strain EDL933 (stx1+, stx2+ and escV+), ETEC strain H10407 (elt+, estIa+ and estIb+), EIEC strain 44825 (invE+) and EAEC strain O42 (astA+, aggR+ and pic+), which were kindly provided by the Chinese Center for Disease Control and Prevention, were used as positive control strains, and E. coli ATCC25922 was used as a negative control strain for virulence factor genes.
O and H antigen determination was performed using the tube agglutination test with specific antisera (50 O antisera and 22 H antisera (Denka Seiken Co. Ltd, Tokyo, Japan) and 14 O antisera (State Serum Institute, Copenhagen, Denmark)) according to the instructions for the reagents.
Antimicrobial susceptibility testing and detection of β-lactamase genes
The isolates were tested for susceptibility to a panel of 18 antimicrobials using the Kirby-Bauer disc-diffusion method. ESBL-producing isolates were screened using a phenotypic confirmatory test using cefotaxime, ceftazidime, cefotaxime/clavulanate and ceftazidime/clavulanate. The results were interpreted according to the guidelines of the Clinical Laboratory and Standards Institute (CLSI, 2010). Intermediate susceptibility was analysed as resistance. Multidrug resistance (MDR) was defined as acquired non-susceptibility to at least one agent in three or more antimicrobial categories based on the methods of Magiorakos et al. . The ESBL-positive isolates were tested by PCR for the presence of blaTEM, blaSHV, blaCTX-M-1 and blaCTX-M-9 group genes, and amplicons were sequenced . The DNA sequences were compared with sequences in GenBank or the β-lactamase classification system (http://www.lahey.org/studies/) to confirm the subtypes of the β-lactamase genes. The quality control isolates included E. coli ATCC25922, Staphylococcus aureus ATCC25923 and Klebsiella pneumoniae ATCC700603.
Data management and analysis
Demographic and laboratory data were entered into the information system of the National Infectious Diseases Surveillance Platform Project. Statistical analyses were performed using SPSS 17.0. The statistical significance of the differences between the groups was tested by the χ2 test or Fisher's exact test. All p-values reported were two sided, and p < 0.05 was considered to be statistically significant.
A total of 2466 faecal specimens were collected from nine sentinel hospitals from July 2009 to June 2011, including 1353 specimens from men and 1113 specimens from women (1.2:1). Among the patients, 909 (36.9%) were ≤5 years old, 123 (5.0%) were 6–18 years old, 1200 (48.7%) were 19–60 years old and 234 (9.5%) were >60 years old.
Prevalence of DEC
The results of the pathogen isolation from the 2466 samples are shown in Table 1. A total of 347 (14.1%) DEC isolates were collected, and EAEC (217, 62.5%) was the most prevalent pathotype identified, followed by ETEC (62, 17.9%), EPEC (52, 15.0%), STEC (14, 4.0%) and EIEC (1, 0.3%). One (0.3%) strain was astA+, aggR+ and elt+. Among the STEC stains, stx2+ strains accounted for 78.6%, and none was stx1+ and stx2+ (Table 2).
Table 1. Frequency of isolated enteric pathogens from surveillance conducted in southeastern China from July 2009 to June 2011
No. of samples tested
No. (%) of positive samples
Rotavirus was detected using a commercial ELISA kit. Calicivirus, astrovirus and adenovirus were detected by reverse-transcriptase PCR and real-time PCR. Virus detection was performed on part of the specimens only.
Among the 347 isolates, 142 (40.9%) were typable for their O antigens, 6 (1.7%) were defined as ‘rough’, and the O antigen types of 199 (57.3%) isolates could not be determined. A total of 35.5% (77/217), 76.7% (43/62), 38.5% (20/52) and 14.3% (2/14) of EAEC, ETEC, EPEC and STEC isolates were O antigen typable, respectively. The dominant O serogroups were O45 (23 isolates, 6.6%), O6 (22, 6.3%) and O25 (14, 4.0%), and the others belonged to 24 serogroups. The EAEC, ETEC and EPEC isolates belonged to 23, 11 and 9 serogroups, respectively. O45, O148 and O45 were the predominant serogroups of EAEC, ETEC and EPEC isolates, respectively. One STEC isolate was identified as O157:H7 (Table 2).
A total of 340 isolates (seven isolates died) were analysed and found to have 166 different sequence types (ST), of which 78 were novel types. The most prevalent genotypes were ST complex (STC) 10 (87, 25.6%), ST38 (11, 3.2%) and ST226 (10, 2.9%). STC10 includes ST10, ST48, ST218 and 20 other types, and ST10 is the predominant type among STC10 isolates (25, 28.7%). The EAEC, EPEC and ETEC isolates belonged to 114, 40 and 24 STs, and STC10 accounted for 19.3% (41/212), 21.2% (11/52) and 55.7% (34/61) of isolates, respectively. The data are presented in Fig. 1(a). A total of 59.1% (13/22) of O6 isolates, 42.9% (6/14) of O25 isolates, 80.0% (8/10) of O148 isolates and 75.0% (6/8) of O159 isolates belonged to ST48, ST1491, ST94 and ST218, respectively. Two isolates were EAEC-O104:H4-ST678.
Antimicrobial resistance and ESBL genes
Of the 342 isolates (five isolates died), 91.8% were resistant to ampicillin; 52.3–57.6% were resistant to cefazolin, tetracycline, ampicillin-sulbactam and trimethoprim-sulphamethoxazole; and <10.0% were resistant to piperacillin-tazobactam, cefoxitin and amikacin. All isolates were sensitive to imipenem and meropenem. EAEC, ETEC, EPEC and STEC exhibited different antimicrobial resistance patterns (Table 3). The ESBL genotypes were determined for 104 of the 118 ESBL-positive isolates, and TEM-type ESBLs were not detected. Nine isolates harboured an SHV-type ESBL: SHV-12 (4), SHV-5 (2), SHV-2 (2) and SHV-57 (1). The CTX-M ESBLs were detected in 98 isolates. Of these, 33 belonged to the CTX-M-1 group (17 CTX-M-55, 12 CTX-M-15, 2 CTX-M-3 and 2 CTX-M-22) and 65 belonged to the CTX-M-9 group (48 CTX-M-14, 11 CTX-M-65, 3 CTX-M-27, 2 CTX-M-24 and 1 CTX-M-9). Three isolates carried two ESBL genotypes simultaneously, SHV-12+ CTX-M-65, CTX-M-14+ CTX-M-15 and CTX-M-14+ CTX-M-55.
Table 3. Antimicrobial resistance of diarrhoeagenic Escherichia coli (DEC) isolates from outpatients with acute diarrhoea
Comparison of the percentages of resistant isolates among the different pathotypes of diarrhoeagenic E. coli; NA, not applicable. Boldface indicates statistical significance (p < 0.05).
In the O45, O6 and O25 serogroups, MDR isolates accounted for 91.3% (21/23), 31.8% (7/22) and 21.4% (3/14) of isolates, respectively, while ESBL-producing isolates accounted for 34.8% (8/23), 13.6% (3/22) and 21.4% (3/14), respectively. In the STC10 group, MDR and ESBL-producing isolates accounted for 50.6% (44/87) and 19.5% (17/87) of isolates, respectively. However, 88.0% (22/25) of the ST10 isolates were MDR, whereas only 28.0% (7/25) were ESBL positive. ST38 (11), ST69 (9), ST31 (8), ST93 (7) and ST648 (7) isolates were all MDR. A total of 90.1% (10/11) of ST38 isolates and 85.7% (6/7) of ST648 isolates were ESBL positive, all of which only carried the CTX-M-14 gene. Isolates belonging to the serogroups or genotypes of O6, O25, O159, ST48, ST218, ST94 and ST1491 were highly susceptible to the majority of antimicrobials. In contrast, isolates belonging to O45, O15, O1, O169, ST38, ST226, ST69, ST31, ST93, ST394 and ST648 were highly resistant to the majority of antimicrobials. The relationships between the serogroups or genotypes and antimicrobial resistance patterns are shown in Figs 1 and 2.
The strain causing the 2011 German outbreak belonged to an EAEC lineage that had enhanced pathogenicity and antimicrobial resistance due to the acquisition of genes for Shiga toxin 2 (stx2) and antimicrobial resistance . This outbreak had serious consequences, emphasizing the importance of detecting all DEC strains, not just EHEC O157:H7 . Our current study demonstrated that the DEC isolation rate was the highest among bacterial pathogens. In Europe, Clostridium difficile is considered a frequent cause of community-acquired diarrhoea . However, in China, stuides on C. difficile are limited and focus primarily on hospital-acquired diarrhoea . The prevalence of C. difficile in community-acquired diarrhoea should be studied further. The DEC isolation rate was 14.1%, which was higher than the rate of 9.4% reported in the USA  but lower than the rate of 28.6% reported in South Africa . A report from Brazil showed that economic status and health conditions were important factors that influence the prevalence of DEC . The relatively high EAEC prevalence (62.5%) in this study was higher than that reported in other regions [12, 13, 15-18]. The outbreak strain in Germany harboured an unusual combination of virulence factors from EAEC and STEC . Interestingly, in our study, there was one DEC isolate that carried the EAEC virulence genes astA and aggR as well as the elt gene of ETEC. This result indicates the possibility that EAEC strains can integrate other virulence genes to generate outbreak isolates. STEC can produce Shiga toxin (Stx1 and Stx2), namely verotoxin, which causes different symptoms and may lead to fatal HUS. The current study found a lower STEC detection rate, which is consistent with the results from neighbouring areas such as Vietnam and Thailand [15, 17]. Tarr et al. reported that Stx2 toxin was more likely to cause HUS. In our study, the number of STEC isolates carrying stx2 was notably higher than that carrying stx1.
The true population structure of DEC isolates from the stool specimens of diarrhoea patients has rarely been studied using MLST. Previous studies focused primarily on extra-intestinal pathogenic E. coli (ExPEC), especially ExPEC-ST131 . Few studies using MLST have focused on specific pathotypes of E. coli isolated from stool specimens, such as EAEC or ETEC [21-23]. Our study was the first to use MLST to determine the pathogenic characteristics of five pathotypes of DEC. The results showed that the most common genetic lineage was STC10 (25.6%). STC10 accounted for 55.7% of ETEC isolates. Some studies have reported that EAEC in Nigeria belongs to multiple lineages, among which STC10 accounted for 21.3% . In our study, the EAEC isolates had 114 STs, and the most common type was STC10 (19.3%). This result indicates that the distribution of DEC genotypes differs among regions. Our current study discovered two EAEC-O104:H4-ST678 isolates that were genetically homologous to the German outbreak isolates, but the stx2 gene was not found. However, future surveillance efforts should pay attention to EAEC-O104:H4-ST678 isolates.
Ampicillin, trimethoprim-sulphamethoxazole and fluoroquinolones are widely used as the first choice to treat intestinal tract infections because of their ready availability and low cost. In the current study, the ampicillin resistance rate of DEC was 91.8%, which is higher than the rates of 67.7% in Nicaragua and 62.0% in Iran [24, 25]. The ciprofloxacin resistance rate was much higher than that in Nicaragua and Vietnam [24, 26]. This phenomenon could be explained by the widespread use or abuse of these antimicrobials in animal husbandry and aquaculture in this region of China . In our study, 34.5% of the DEC isolates were ESBL positive, and the common genotypes were CTX-M-14, CTX-M-55, CTX-M-15 and CTX-M-65. CTX-M-14 and CTX-M-15 are widely distributed in extra-intestinal and intestinal isolates throughout the world. However, CTX-M-55 and CTX-M-65 were primarily found in animal intestinal isolates in China . The modes of transmission of these two CTX-M genes between animals and humans should be investigated in future work. The antimicrobial resistance patterns of different serotype/genotype isolates were distinct. For ExPEC, Suzuki et al. reported that O86 isolates were more resistant to sulphamethoxazole/trimethoprim, whereas O25 isolates were more frequently resistant to ceftazidime, cefoxitin, ciprofloxacin and chloramphenicol. E. coli ST131, a worldwide pandemic clone, causes predominantly MDR infections and is associated with O25:H4 and CTX-M-15 . However, the situation was different in the DEC isolates included in our current study. Four ST131 isolates, one O25 isolate and three MDR isolates were negative for the CTX-M genes. In addition, isolates with the serotypes O6, O25, O159, ST48, ST218, ST94 and ST1491 were highly susceptible to the majority of antimicrobials. In contrast, isolates with the serotypes O45, O15, O1, O169, ST38, ST226, ST69, ST31, ST93, ST394 and ST648 were highly resistant to the majority of antimicrobials. Given these results, the true relationships between serotypes or genotypes and antimicrobial resistance patterns were not clear in our study. A larger number of clinical isolates may be needed to verify the complex relationships.
In conclusion, DEC accounted for the majority of bacterial pathogens responsible for acute diarrhoea in southeastern China, and it is necessary to test for all DEC, not just EHEC O157:H7. Some serogroups or genotypes of DEC were highly resistant to the majority of antimicrobials. DEC surveillance should be emphasized to monitor epidemic trends, reduce outbreaks and improve the effectiveness of antimicrobial therapy.
We thank all the participants from the nine hospitals for the data collection and testing of specimens. We also gratefully acknowledge the Chinese Center for Disease Control and Prevention for supporting this project.
This work was supported by the National Key Programs for Infectious Diseases of China (2012ZX10004210). The authors declare no conflicts of interest.