Characterisation of STEC and other diarrheic E. coli isolated on CHROMagar™STEC at a tertiary referral hospital, Cape Town

Sample collection and processing

This study is part of a larger study on STEC in human and non-human sources in Cape Town. At the National Health Laboratory Services (NHLS) laboratories at Groote Schuur Hospital (GSH), stool specimens are not routinely screened for diarrhoeic E. coli. Physicians are advised to contact the laboratory within seven days if there is a clinical suspicion of diarrhoeic E. coli infections such as STEC. Diarrhoea, in this study, was defined as more than three loose stools in less than 24 h. Freshly collected diarrhoeic stool specimens (in a screw-capped stool collection container and delivered in a temperature regulated box) submitted to the NHLS at GSH, Cape Town, between September 2014 and May 2015 for microbiological testing (as part of standard patient care) were screened for diarrhoeic E. coli. All the stool samples from diarrhoea samples were tested irrespective of the age of the patient as recommended by the Centre for Disease Control (CDC) [13]. In order to isolate tellurite resistant STEC and other diarrheic E. coli, samples were either inoculated directly onto CHROMagar™STEC (CHROMagar Microbiology, Paris, France) or subjected first to overnight enrichment in Tryptic Soy Broth (TSB; Oxoid, Basingstoke, U.K) before inoculation onto CHROMagar™STEC (Fig. 1). Mauve colonies, indicative of growth and therefore tellurite resistance, were visually identified and selected (a maximum of 5 per sample) for characterisation. Pathotypes and serotypes were determined as previously described [11].

Briefly, to dertermine the pathotypes, end point PCR reactions targeting the fimbrial adhesion gene (daaC) for diffusely adherent E. coli (DAEC), the anti-aggregation protein transporter gene (aat) for enteroaggregative E. coli (EAggEC), heat-stable (ST) and heat-labile (LT) enterotoxin genes of enterotoxigenic E. coli (ETEC), the invasive plasmid antigen gene (ipa) for enteroinvasive E. coli (EIEC), and the intimin coding gene (eae) for enteropathogenic E. coli (EPEC) were done as previously described [14]. An in-house developed duplex real-time PCR assay was used to screen for stx₁ and stx₂. The stx + E. coli isolates were tested for Shiga toxin production using the immunocard STAT® EHEC (Meridian Biosciences Inc., Cincinnati, Ohio, United States). Subsequently, clinical data were reviewed for all diarrhea cases that carried STEC to assess for presentation of typical haemolytic uremic syndrome (tHUS).

To determine the serotypes, serology was done to detect the somatic O- antigens using reagents supplied by the Statens Serum Institut, Copenhagen, Denmark and the tube agglutination method as previously described [15]. Serology to detect the H-antigens was not done.

Culture characteristics on sorbitol MacConkey

Briefly, mauve isolates identified as E. coli using the VITEK 2 system (bioMérieux, USA) were cultured on Sorbitol MacConkey and incubated for 18 h at 35 ̊C. Colourless colonies were considered to be non-sorbitol fermenting. Antimicrobial susceptibility testing.

Isolates were tested for susceptibility to 19 antibiotics following the Clinical Laboratory Standards Institute guidelines (CLSI, 2015). Since VITEK offers a limited MIC range, we further re-tested these isolates using the broth microdilution based GNX2F sensititre MIC plate (Thermo-scientific, USA) to confirm the MICs observed using VITEK 2 systems. MIC values from broth microdilution were used as the final reference for classification as resistant (R), intermediate (I) or susceptible (S). E. coli ATCC 25922 was used as the control strain. The antibiotics included in the broth microdilution panel on the GNX2F sensititre plate were: amikacin (AMK), aztreonam (AZT), cefepime (FEP), cefotaxime (CTX), ceftazidime (CAZ), ciprofloxacin (CIP), colistin (COL), doripenem (DOR), doxycycline (DOX), ertapenem (ETP), gentamicin (CIN), imipenem (IMP), levofloxacin (LEV), meropenem (MEM), minocycline (MIN), piperacillin-tazobactam (TZP), polymixin B (POL), ticarcillin-clavulanic acid (TIM), tigecycline (TGC), tobramycin (TOB), and trimethoprim – sulfamethoxazole (SXT).

MIC data were entered into WHONET 5.6 software. Resistance to third-generation cephalosporins and or carbapenems was confirmed using the ESB1F sensititre MIC plates (Thermo-scientific, USA).

Data analysis

Data on cultural characteristics, serotypes, and pathotypes was entered into Microsoft Office Excel 2016 (Microsoft Corporation, USA) software, coded and exported to STATA version 12 for analysis. Further analysis was done using the R-statistical package [16]. Results were entered in WHONET 5.6 (World Health Organization, Geneva) and interpreted in accordance with current Clinical Laboratory Standards Guidelines [17].

Detection of diarrhoeic E. coli in stool

The inhibitory action of potassium tellurite on coliforms was first reported by Fleming in 1940 [18]. Tellurite-containing media have been routinely used to screen for STEC [19], but not the other E. coli pathotypes. However, Hirvonen et al., 2012 reported the formation of mauve colonies by other tellurite resistant diarrhoeic E. coli on this medium [8]. Of the 733 stool samples processed in this study, 33 (5%) contained tellurite-resistant diarrhoeic E. coli. However, tellurite-containing media have limited ability to select for STEC. In a study conducted on STEC from human, animals, and food in Austria, the prevalence of tellurite resistance amongst STEC was 26% - therefore 74% of the STEC would not be detected by the tellurite-containing medium. [20]

Tellurite resistance and antimicrobial resistance are often observed in highly pathogenic diarrhoeic E. coli strains [20]. Commercial selective media such as CHROMagar™STEC and Sorbitol MacConkey with cefixime and potassium - tellurite use this property to select STEC O157: H7 and other pathogenic STEC. There was a relatively low prevalence (5%, 33/733) of tellurite-resistant diarrhoeic E. coli in stool samples submitted to NHLS, GSH. We were unable to determine the prevalence of tellurite susceptible STEC in this study; this is because these would not grow on CHROMagar™STEC.

STEC

Five STEC were detected in this study. All the five were detected after enrichment in TSB, and none was obtained by directly streaking on CHROMagar™STEC. Since the other diarrheic E. coli pathotypes were detected with and without enrichment, it can, therefore, be speculated that TSB enrichment is more beneficial for STEC isolation relative to the other diarrheic E. coli pathotypes. Overall, the number of diarrheic E. coli obtained after enrichment in TSB was significantly higher than the number obtained without (p = 0.004). This may suggest that enrichment in TSB may be a beneficial step as regards recovering STEC (and other pathotypes) from a stool sample. The STEC diagnostic strategy routinely employed by the NHLS clinical laboratory at the GSH targets only O157 STEC and is based on the non-sorbitol fermenting attribute of STEC O157: H7. Therefore, this strategy would miss the non-O157 STEC and sorbitol-fermenting E. coli O157. Of the five STEC reported in this study, only one was non-sorbitol fermenting and belonged to serotype O101. The five non- O157 STEC that we report in this study would have been missed (four because they were sorbitol fermenting and one because it was not O157) given that the routinely used method is based on absence of sorbitol fermentation and targets only E. coli O157:H7. Only two of the five STEC tested positive for Shiga toxin production by immunoassay because the other three did not express the stx genes.

The relatively low rate of STEC isolation in this study could be due to several reasons: (1) The stx primers may not have been suitable for detection of all the stx gene variants [21]. (2) The tellurite susceptible STEC could not be isolated on CHROMagar™STEC. (3) This study had a short sampling frame and not all children presenting with diarrhoea may have had a stool specimen taken. (4) There could have been a low prevalence of STEC in stool samples processed, especially given that the stool specimens were not from patients presenting with bloody diarrhoea, or HUS. There are higher chances of recovering STEC from the stool of patients with bloody diarrhoea or HUS as opposed to those without [22].

In South Africa, as in other African countries, there may be a lower prevalence of STEC as compared to America and Europe. This may be related to the type of diet given to ruminants in America and Europe that favour the proliferation of STEC in cattle-the primary ruminant reservoirs for STEC [23].

It is important to detect STEC in stool because the use of antibiotics such as the quinolones leads to bacterial lysis and toxin release which increases the chances of HUS in infected patients [23]. Four of the five STEC detected in this study carried the stx ₁ genes. Possession of stx ₂ genes has been associated with the more severe form of illness [24]. Also, we detected eae in three of the five STEC, while two did not possess the eae gene. The eae gene which is located in the locus of enterocyte effacement (LEE) codes for the intimin protein which is necessary for the formation of attaching-effacing lesions in the intestinal tract [25]. STEC that carry eae (LEE-positive STEC) have been shown to cause more severe disease.

Per the manufacturer of CHROMagar™STEC, the commonly encountered STEC serotypes should form mauve coloured colonies on this medium. However, the serotypes were categorised as “common” (O157, O26, O45, O145, O111, O121, and O103) based on outbreaks that occurred in developed countries. In this study, STEC belonged to serotypes O101 and O186. These serotypes were not detected in an earlier study conducted at the NICD (2006 to 2009) which screened 2378 diarrhoeic E. coli isolates. In that study, STEC in stool had not been purposively targeted (and thus CHROMagar™STEC was not used) but was an incidental finding among EPEC (the strains carried the eae gene) that had been sent to the central public health laboratory for serotyping. The 14 STEC identified in that study belonged to serotypes O4, O5, O21, O26, O84, O111, and O157 [26]. Other related studies reported STEC serotypes O8 and O9 in pigs in South Africa [27]. We did not detect E. coli O157: H7 despite high sensitivity of CHROMagar™STEC for this serotype [3].

Clinical relevance of STEC carriage

On review of the clinical records of the five patients that carried STEC, none presented with typical Hemolytic Uremic Syndrome (HUS). However, the assessment of clinical relevance of STEC carriage was not possible due to the protracted time lag between the onset of sickness in the community and reporting to primary health care and eventual referral to tertiary health care. Earlier research has shown that only a small percentage of acute diarrhea cases in South Africa report to the health centres [28].

Limitations of this study

CHROMagar™STEC only permits the growth of tellurite resistant STEC and not the tellurite susceptible strains. This study did not have a long sampling timeframe, and not all children that presented with diarrhoea may have had stool specimens taken. Only diarrhoeic E. coli that possessed virulence genes were characterised; therefore, we might have missed strains that lost the virulence genes.