Nine regional clinical microbiology laboratories in Nebraska sent stool samples from March 1, 1998, to October 31, 1998, to the Nebraska Public Health Laboratory, University of Nebraska Medical Center. All stool samples that were sent to a participating laboratory with a physician's order to screen for enteric pathogens were included. Thus, all samples were from patients with a differential diagnosis of bacterial gastroenteritis. Patients who had been in the hospital for >2 days before diarrhea developed were excluded. The samples were added to a Para-Pak C&S stool transport container (Meridian Diagnostics, Cincinnati, OH) and sent by courier to the Nebraska Public Health Laboratory.

Samples were plated to cefixime-tellurite sorbitol MacConkey (CT-SMAC) agar plates and screened for typical sorbitol-negative colonies (1). Presumptive colonies were identified as E. coli by API strips (Biomerieux Vitek, Hazlewood, MO) and serotyped with RIM E. coli O157:H7 (Remel, Lenexa, KS). Samples were injected into 10 ml MacConkey broth (Difco, Detroit, MI) and incubated overnight at 37°C. The Premier enterohemorrhagic E. coli (EHEC) assay was performed by using 50 µl of overnight growth. The reaction mixtures were read spectrophotometrically at 450 nm and scored as positive or negative. PCR was performed by first extracting DNA from the overnight culture of MacConkey broth using a QIAamp Tissue kit (Qiagen, Santa Clarita, CA.). The following set of primers, which detects both stx₁ and stx₂, was used: 5'TTTACGATAGACTTCTCGAC3' and 5'CACATATAAATTATTTCGCTC3'(2). E. coli O157:H7 strain G5244 was used as positive control (Centers for Disease Control and Prevention [CDC] strain collection). Samples Shiga-toxin positive by either EHEC enzyme EIA, PCR, or both were plated onto sheep-blood agar (Remel) and streaked for isolation. After overnight growth, multiple E. coli-like colonies were selected for retesting by using the Premier EHEC assay. Positive colonies were identified to species level by using API strips and serotyped by CDC.

Multiplex PCR (3) was performed on isolated Shiga toxin- positive colonies to detect specific genes encoding Shiga toxins 1 and 2 (stx₁ and stx₂), intimin (eaeA), and enterohemolysin A (ehxA).Genomic DNA suitable for pulsed-field gel electrophoresis (PFGE) was prepared (4) and digested with XbaI (Roche, Indianapolis, IN). E. coli O157:H7 G5244 was used as a standard. PFGE patterns were captured by a Bio-Rad Gel-Doc system and were analyzed by Molecular Analyst software (Bio-Rad, Hercules, CA).

Of the nine clinical laboratories that submitted 335 samples during the study period, five submitted samples positive by CT-SMAC culture, EIA, or stx PCR. Fourteen 4.2%) samples were positive by at least one of the methods (Table 1); 13 of these were obtained either through direct culture by using CT-SMAC or through Shiga toxin screening and subsequent colony isolation. Six of the thirteen were serotype O157:H7 or O157:NM; seven were non-O157 serotypes. All seven of the non-O157 isolates were the predominant species found in the culture when the sample was plated on sheep-blood agar. All six E. coli O157:H7 or O157:NM isolates were detected by using CT-SMAC culture and stx PCR; five of six were detected by EIA. All of seven non-O157 isolates were detected by EIA or stx PCR. One sample (isolate B3) that was negative by EIA but positive by PCR and CT-SMAC culture was subsequently found to be positive by EIA when tested individually. One sample from laboratory C was positive for EIA and stx PCR (both tests were weak positives), but no Shiga toxin-positive colony was obtained upon repeat subculture. The low prevalence of this organism in the stool sample may reflect STEC carriage in this patient. By the combined results of both culture and EIA as the reference standard (14 samples positive, 321 samples negative), the sensitivity and specificity of the stx PCR (14 samples positive, 321 samples negative) were each 100%.

PCR was performed on the 13 isolated STEC to detect stx₁, stx₂, eae, and ehxA (Table 2). All isolates, regardless of serotype, encoded eae and ehxA; three of five O157 isolates encoded both stx₁ and stx₂; two of seven non-O157:H7 isolates encoded stx₂ (both O111:NM). PFGE showed that all but two STEC isolates were unrelated; isolates D1 and D2 (both O111:NM), which were isolated from samples sent from the same laboratory, were indistinguishable (data not shown).

In a 1997 study of 30,000 diarrheal stool samples, E. coli O157:H7 was the fourth most prevalent bacterial enteric pathogen (5). However, the incidence of non-O157 STEC in the United States is not well established. Studies from Europe have shown that the prevalence of STEC in diarrheal samples is 0.3% to 9.3%; serogroup O157 prevalence is 0% to 2.7%. In Australia, serotype O111:NM is the most frequent cause of serious human disease and has been associated with outbreaks. In a recent study of 3,289 diarrheal samples from clinical laboratories in the United States, non-O157 STEC were more prevalent than O157 serotypes (6). STEC was found to be as prevalent (1.2%) as Shigella sp. (1.4%), and almost as prevalent as Salmonella sp. (2.4%) and Campylobacter sp. (2.0%). Testing for E. coli O157:H7 alone would have missed up to 50% of STEC.

Our study is the first to address the prevalence of non-O157 STEC in diarrheal samples from the Great Plains region of the northern United States, where cattle and other animal reservoirs of STEC are abundant. In our study, 4.2% of the samples were positive for STEC by CT-SMAC culture, PCR, or Meridian EHEC EIA. Though this prevalence is higher than previously reported in the United States, other studies have shown that northern states have a higher prevalence of E. coli O157:H7 (7). In addition, Nebraska has a large rural population, whose members may have contact with animal reservoirs that carry STEC. Five different non-O157 STEC serotypes were isolated: O111:NM, O26:H11, O145:NM, O103:H2, and Orough:H2. Four of these are associated with HUS (O111:NM [8-10], O26:H11[11], O145:NM [12], and O103:H2 [13]). In addition, serotypes O111:NM and O26:H11 have been associated with diarrhea in weaned calves (14). Although most STEC cases are linked to eating undercooked hamburger (15), contact with food animals has also been implicated as a source of infection (16).

Shiga toxins 1 and 2 are the main virulence factors associated with hemorrhagic colitis and HUS, presumably because they interact with endothelial cells at the site of infection and in the glomeruli and arterioles of the kidney (17). stx₁ and stx₂ are highly related yet immunologically distinct. STEC produce other accessory virulence factors, including intimin (eae) and enterohemolysin A (ehxA). The former is responsible for the characteristic histopathologic feature known as the attaching and effacing (A/E) lesion (18); ehxA is a hemolysin encoded by the 90-kb virulence-associated plasmid found in most STEC infecting humans (19). A study of 237 isolates from 118 serotypes found a significant association between stx₂ and eae in isolates that caused hemorrhagic colitis and HUS in humans (20) and between ehxA and severe disease. STEC isolates from asymptomatic human carriers usually do not encode eae and therefore may not have a mechanism to adhere to intestinal epithelial cells (21). However, STEC isolates lacking the eae gene have been associated with hemorrhagic colitis and HUS (22). All STEC isolates obtained in this study encoded both eae and ehxA. The six O157 isolates also encoded stx₂, as did both O111:NM isolates, which indicates that at least these eight isolates could produce serious disease. Five of seven non-O157 isolates encoded stx₁ only; however, STEC isolates associated with hemorrhagic colitis and HUS that encoded only stx₁ have been reported (23).

All O157:H7, O157:NM, and O26:H11 isolates in our study were distinct by PFGE RFLP patterns, which suggests that these cases were sporadic. However, two O111:NM isolates (D1 and D2) from the same community had indistinguishable PFGE RFLP patterns. The isolates were from different patients and were cultured within 1 day of each other. This suggests that an outbreak of O111:NM may have occurred that was not detected by standard laboratory techniques. However, without epidemiologic information, these results are difficult to interpret.

This study demonstrated that non-O157 STEC serotypes are at least as prevalent as serogroup O157 in diarrheal samples from Nebraska. Non-O157 STEC isolates presumably were the cause of diarrhea in 7 of 14 positive samples. These non-O157 isolates carried known STEC virulence genes and were the predominant organism found in culture. Other bacterial pathogens such as Salmonella, Shigella, and Campylobacter were not isolated from these seven samples. Our results suggest that stx PCR is as sensitive and specific as CT-SMAC culture and EIA combined, and therefore may be used as an alternate method to diagnose diarrheal infections caused by STEC. Clinical laboratories may need to implement serotype-independent methods to avoid underdiagnosis of STEC-mediated bacterial gastroenteritis.