Human Infections with Non-O157 Shiga Toxin–producing Escherichia coli, Switzerland, 2000–2009

We characterized 97 non-O157 Shiga toxin (stx)–producing Escherichia coli strains isolated from human patients during 2000–2009 from the national reference laboratory in Switzerland. These strains belonged to 40 O:H serotypes; 4 serotypes (O26:H11/H–, O103:H2, O121:H19, and O145:H28/H–) accounted for 46.4% of the strains. Nonbloody diarrhea was reported by 23.2% of the patients, bloody diarrhea by 56.8%. Hemolytic uremic syndrome developed in 40.0% of patients; serotype O26:H11/H– was most often associated with this syndrome. Forty-five (46.4%) strains carried stx2 genes only, 36 strains (37.1%) carried stx1, and 16 (16.5%) strains carried stx1 and stx2. Genes encoding enterohemolysin and intimin were detected in 75.3% and 70.1% of the strains, respectively. Resistance to >1 antimicrobial agent was present in 25 isolates. High genetic diversity within strains indicates that non-O157 stx–producing E. coli infections in Switzerland most often occurred as single cases.

The common feature and main virulence factor of STEC is production of stx1 or stx2 proteins. Human virulent STEC strains often may also contain other virulence factors such as intimin (eae), a protein essential for the intimate attachment and the formation of attaching and effacing lesions on gastrointestinal epithelial cells, and E. coli hemolysin (ehxA) (2).
Little data are available for clinical non-O157 STEC infections in humans, including those in Switzerland, a country with a small but disproportionately high population of travelers. Therefore, we characterized all non-O157 STEC strains collected by the Swiss National Centre for Enteropathogenic Bacteria (Zurich, Switzerland) during 2000-2009, characterized these strains according to clinical and anamnestic data, and compared these results with data from other countries in Europe and the United States.

Strains
A total of 97 non-O157 STEC strains obtained from the Swiss National Centre for Enteropathogenic Bacteria were characterized. Strains were isolated during 2000-2009 from fecal samples of human patients with a reasonable clinical suspicion of infection with STEC. Samples sent to the Centre from hospitals or private practitioners are representative for Switzerland and the period screened. lin, Germany). Nonmotile strains were investigated with respect to their fl agellar genotypes by using PCR and HhaI digestion of PCR products as described (3).

Genotyping
Pulsed-fi eld gel electrophoresis (PFGE) was performed according to the PulseNet protocol (Centers for Disease Control and Prevention, Atlanta, GA, USA) (8) and by using restriction enzyme XbaI and the CHEF-DR III system (Bio-Rad, Hercules, CA, USA). Pulse times were ramped from 5 to 50 sec for 19 h at an angle of 120°. Salmonella enterica serovar Braenderup strain H9812 (BAA 664; American Type Culture Collection, Manassas, VA, USA) was used as a reference. GelCompar II software (Applied Maths NV, Sint-Martens-Latem, Belgium) was used for pattern comparison. PFGE patterns were considered clonally related if they had a similarity coeffi cient >80% (Dice similarity index and unweighted pair-group with arithmetic mean method).

Antimicrobial Drug Susceptibility Testing
Strains were tested for antimicrobial drug resistance by the disk-diffusion method according to protocols of the Clinical and Laboratory Standards Institute (9). The panel of antimicrobial drug disks (Becton Dickinson, Sparks Glencoe, MD, USA) used contained ampicillin, amoxicillin/clavulanic acid, ceftazidime, cefalothin, ciprofl oxacin, cefpodoxime, cefotaxime, cefuroxime, cefepime, cefoxitin, gentamicin, and tetracycline. E. coli strain 25922 (American Type Culture Collection) was used as a quality control.

Strains
The 97 strains were isolated from 95 patients. Two STEC strains (968-03I and 968-03II; 2244-08I and 2244-08II) were isolated from 2 patients (online Appendix Table, www.cdc.gov/EID/content/17/2/178-appT. htm). During the period of screening, an increasing number of strains per year were registered. This increase could have been caused by the offi cial reporting system for STEC, which was initiated in Switzerland in 1999; or by a country-wide reporting program for HUS, which was initiated in Switzerland in 2004 and may have increased physician awareness for HUS.
No data were available for use of antimicrobial drugs. However, it is generally assumed in Switzerland that no antimicrobial drugs should be given to patients with acute diarrhea if laboratory test results are not available.

PFGE Genotyping
PFGE conducted for strains of the most prevalent serogroups (O26, O103, O121, and O145) showed that the patterns of O26 strains were heterogeneous (similarity coeffi cient range 49%-94%) except for 5 strains ( Figure 2). Patterns of O145 strains were also heterogeneous (similarity coeffi cient range 56%-95%) except for 2 strains ( Figure  3). Patterns of O103 and O121 strains were heterogeneous (similarity coeffi cients ranges 86%-97% and 62%-91%, respectively) (Figures 4, 5). For O26 and O145 strains with the same PFGE patterns, no obvious epidemiologic link was observed between patients and different regions of Switzerland. Moreover, no information was available for about risk factors for these patients.

Discussion
Over the 10-year study period (2000-2009), HUS caused by infection with non-O157 STEC was detected in 38 (40%) of the 95 patients investigated. This frequency of HUS cases in our study was higher than that in other studies (10,11). In contrast to other studies, in which all diarrhea samples were screened for STEC, the set of non-O157 strains in our study was isolated from patients with a reasonable clinical suspicion of infection with STEC. This fact could be the reason for a higher proportion of HUS and bloody diarrhea cases in our study group. Fourteen (36.8%) HUS cases were caused by STEC O26:H11; the remaining cases were caused by other serotypes. Of the patients with HUS, 20 (52.6%) were female and 18 (47.4%) were male. Furthermore, 30 (78.9%) were <5 years of age and only 4 (10.5%) were >60 years of age.
Of the remaining major O groups, O145 (6 isolates of O145:[H28] and 4 isolates of O145:H25/H -) was the second most common non-O157 serogroup isolated in this study (10.3% of all isolates). Eight O145 isolates had positive results for stx2, eae, and ehxA; 1 isolate that did not ferment sorbitol had positive results for stx1, eae, and ehxA; and 1 isolate that had positive results for stx1 and stx2 also had positive results for eae and ehxA. These strain characteristics are similar to those of isolates from Finland  and Germany (20,21). Four of the O145 strains were associated with bloody diarrhea, and 5 were associated with HUS, which is similar to symptom distribution associated with STEC O145 reported by Karch et al. (18).
The third most common non-O157 STEC serotypes isolated were O103:H2/H -(6 strains) and O121:H19 (6 strains). In contrast to fi ndings from Germany (19), none of the O103 strains from Switzerland were associated with HUS.
Among the STEC O121:H19 group, all but 1 of the strains were motile, fermented sorbitol, and were stx2, eae, and ehxA positive; the 1 exception was ehxA negative. Four patients had bloody diarrhea and HUS developed in 2 patients. These fi ndings confi rm the results of Johnson et al. (11), which showed that strains of serogroup O26, O103, O121, and O145 are more likely to be associated with cases of HUS.
Although STEC O111 is one of the most common serogroups in countries such as Germany, Austria, and Australia and is often associated with HUS (11,22), we found only 2 O111:Hstrains that had the H8 genotype. One of these strains had stx1, eae, ehxA, and the other strain had stx1, stx2, eae and ehxA. The patients from whom these 2 strains were isolated sought treatment for bloody diarrhea and HUS, respectively.
According to FoodNet 2009 (23), the most common non-O157 STEC serogroups in the United States are O26 (28.9%), O103 (20.0%), and O111 (14.9%). The frequency of O26 (28.9% of all isolates) in the United States is the same as that in Switzerland. However, the frequencies of STEC O103 and O111 in the United States are higher than those in Switzerland (STEC O103, 6.2%; STEC O111, 2.1%).
In Germany, STEC O91 (H14/H21) is currently the fourth most common STEC serogroup isolated (24). However, we have detected only 1 O91:H10, stx2-positive, eaeand ehxA-negative, sorbitol-fermenting strain in Switzerland during the past decade. This strain was isolated from a 60-year-old woman in whom HUS developed. STEC O91 isolates that express fl agellar antigen H10 have been detected in different countries, albeit at low frequencies (25)(26)(27). O91 is usually eae negative and the most common serogroup isolated from adult patients (25). Another eaenegative serogroup is O117, which is often associated with travelers, mainly to Asia, Africa, and Cuba (28). We identifi ed 3 (3.1%) O117:H7 strains that were stx1 positive and stx2, eae, and ehxA negative. This virulence pattern was identical to that for 20 STEC strains reported by Olesen et al. (28). One of our strains was associated with a patient who traveled to India, a country that has been reported as the origin of O117 infections (29). However, no information was available regarding travel for the other 2 patients in our study from whom the O117:H7 stx1-positive, stx2-, eae-, and ehxA-negative strains were isolated. Two of these patients had bloody diarrhea, but no clinical data were available for the third patient.
To detect genetic similarities and epidemiologic relationships among STEC strains, we performed PFGE on representatives of serogroups that occurred at high frequencies. O26 PFGE patterns were heterogeneous except for 2 strain sets, which had 2 and 3 strains without any obvious epidemiologic relationship to each other. However, no information for risk factors, such as contact with animals, traveling, or eating animal derivedfood, was available for these 5 patients.
PFGE patterns for O145 strains were also heterogeneous, except for 2 similar strains without any epidemiologic relationships to each other that were isolated in 2008 (K 145:25-1550-08 and K 145:25-2208-08; both stx2 and eae positive, 1 from a 2-year-old girl who was treated for HUS, and 1 from a 1-year-old boy who had bloody diarrhea). However, no information for risk factors, such as contact with animals, traveling, or eating animal-derived food, was available for these 2 patients. PFGE patterns of O103 and O121 strains were the most heterogeneous; none of the strains had the same patterns.  The stx gene distribution among all 97 non-O157 STEC isolates showed that 45 (46.4%) strains had only stx2, 36 (37.1%) had only stx1, and 16 (16.5%) had stx1 and stx2. Studies from Spain (17) and Finland (20) showed a similar distribution of stx genes among non-O157 STEC isolates. These fi ndings are in contrast to those from the United States (Minnesota), in which non-O157 STEC strains generally showed a higher frequency of isolates that had stx1 (10).
The presence of the eae gene has been reported to have some predictive value for STEC seropathotypes that are associated with epidemic disease and consequently associated with severe disease such as bloody diarrhea and HUS (12,30,31). Therefore, we performed statistical analysis with multinomial regression and binary logistic regression for our data.
Strains associated with HUS, compared to those associated with bloody diarrhea, were more likely to harbor stx2 and eae, but the presence of only 1 of these virulence factors was not signifi cantly associated. In our study, of 29 patients who provided eae-negative isolates, 17 had bloody diarrhea, 4 had bloody diarrhea and HUS, 4 had HUS, 1 had nonbloody diarrhea, and 3 had no clinical data available. Strains from the 8 patients with HUS comprised a variety of serogroups (O20, O82, O91, O148, O153, O181, and ONT); only 5 had the stx2 gene. In our study, 21.1% of isolates from HUS patients were eae negative. The 3 eae-negative, stx2-negative strains had only stx1 or stx1 and ehxA. Strains with such a pattern of virulence factors are notable because they are less likely to cause HUS than strains harboring stx2. However, these fi ndings are consistent with epidemiologic data from other countries (19,27), which indicate that certain eae-negative STEC strains cause hemorrhagic diseases in humans. A report from the United States (Minnesota) indicated that non-O157 isolates that had only stx1 can cause severe illness (bloody diarrhea or HUS) (32).
In conclusion, high genetic diversity within strains indicates that non-O157 Shiga toxin-producing E. coli infections in Switzerland most often occurred as single cases. Because little data are available for clinical non-O157 STEC infections in humans, our results may provide useful information for analysis of these strains.