Shiga Toxin–producing Escherichia coli Serogroups in Food and Patients, Germany

We compared 61 Shiga toxin–producing Escherichia coli (STEC) serogroups from 448 food isolates with 71 STEC serogroups from 1,447 isolates from patients in Germany. Two thirds (41/61), representing 72% of food isolates, were also found in patients. Serogroups typically isolated from patients with hemolytic uremic syndrome were rarely found in food.

We compared 61 Shiga toxin-producing Escherichia coli (STEC) serogroups from 448 food isolates with 71 STEC serogroups from 1,447 isolates from patients in Germany. Two thirds (41/61), representing 72% of food isolates, were also found in patients. Serogroups typically isolated from patients with hemolytic uremic syndrome were rarely found in food.
S higa toxin-producing Escherichia coli (STEC) of serogroups other than O157 (non-O157 STEC) account for 80% of STEC gastroenteritis reports in Germany's national surveillance database (1). Some of the non-O157 serogroups unequivocally cause disease comparable in severity to that caused by STEC O157, such as the hemolytic uremic syndrome (HUS) (2). Numerous, but not all, STEC serogroups have been linked with human disease.
Food is an important transmission vehicle for human STEC infection, especially in outbreaks (3), and many different STEC serogroups are isolated from food (4). Yet the public health relevance of many of these STEC serogroups, which includes their ability to cause human disease and the frequency with which this may occur, has not been investigated.
In Germany, identifi cation of STEC in patients' stool and in food is based on detection of Shiga toxin or of a Shiga toxin gene and subsequent isolation of STEC strains (4,5). This allows, in principle, ascertainment of all STEC strains, independent of their serogroup. To assess the public health relevance of STEC isolated from food, we compared those strains with those isolated from patients.

The Study
Information on STEC isolates from food came from 2 sources. The fi rst source was the Federal Institute for Risk  Assessment, which received isolates from German governmental food inspection laboratories for strain characterization from 2005 through 2007 (food source 1). These STEC isolates originated from routine food samples taken by food safety authorities across Germany, according to a nationwide sampling scheme that focused during the sampling period mainly on red meat, ground raw meat, and stabilized meat products. The second source was the Max-Rubner Institute in Germany, which had conducted a series of investigations in conveniently selected meat-processing companies in Germany from 1996 through 2004 (food source 2). Information on STEC isolates from patients came from a laboratory-based sentinel in existence from 1999 through 2004, coordinated by Germany's National Reference Center. The sentinel has been described elsewhere (6). In brief, private laboratories across Germany agreed to screen stool specimens of gastroenteritis patients for the presence of Shiga toxin 1 and Shiga toxin 2 with an enzyme immunoassay if predefi ned criteria were met (e.g., patients with diarrhea were <5 years of age, bloody diarrhea was mentioned on the laboratory request form). Positive samples were sent to the National Reference Center, where STEC strains were isolated and subtyped by various methods (including serotyping). We calculated frequencies and proportions of STEC serogroups separately for food and patient isolates. Serogroups were compared for matches in both groups. Because the clinical outcome associated with human STEC infection was not systematically recorded, we additionally compared serogroups of food isolates with a compilation (available on the Internet) of literature reports of STEC serotypes and their association with human illness (7). We acknowledged an association with human illness if a symptom at least as severe as diarrhea was specifi ed for a serogroup. The proportion of serogroups in patient and food isolates was compared by using the Wilcoxon signed rank test. Within selected serogroups, we examined serovars (classifi ed by O and H antigen, e.g., O157:H7) to assess comparability between food and patient isolates because the serovar is a better proxy for genomic background of the strains than is the serogroup.
Of the 61 food serogroups, 41 (67%) were also identifi ed in patients ( Figure). These serogroups comprised 72% (242/339) of food isolates with a known serogroup. Similarly, 78% (19/25) of serogroups isolated from game, accounting for 70% (44/63) of isolates, occurred also in patients. The Internet search showed a published association with human illness for at least 41 (67%) of all food serogroups; the phrase "at least" is used because 5 serogroups (O174, O176-O179) were offi cially acknowledged as genuine O-groups after May 2003 (10), which according to the website is the date of its last update (7). Moreover, some serogroups exclusively found in food in this study (and not listed on the website) have been described as sporadic patient isolates from Germany (11) and elsewhere (12).
Overall, a signifi cant inverse correlation was found between the ranking of the serogroup proportion in patients and in food (p<0.01). This fi nding is illustrated by the following: of the 41 serogroups found in food and in patients, 33 accounted each for <1% of the patient isolates. In total, they represented only 9% of patient isolates but 45% of food isolates. Conversely, the 3 most frequently identifi ed serogroups in patients, O157, O103, and O26, represented 46% of the patient isolates but only 3% of food isolates. These 3 serogroups account for 85% of STEC isolated in pediatric HUS patients in Germany (2). Notably, the virulent serogroup O157 was found in 5 (1%) food isolates. This result is compatible with results of studies conducted in other countries that identifi ed only few, if any, O157 strains among STEC strains isolated from ruminant meat, particularly beef (13,14).

Conclusions
Two thirds (41/61) of serogroups from food were also isolated from patients, comprising 72% of food isolates with a known serogroup. These serogroups included, albeit uncommonly, those typically identifi ed in pediatric and nonpediatric HUS patients. An association with human illness has been published for more than two thirds of food serogroups. These fi ndings suggest that many STEC strains isolated from food in Germany are pathogenic for humans. Notwithstanding, the most frequent STEC serogroups in patients, except O91, were only rarely found in food.
The incongruent serogroup distributions of STEC isolates from food and from patients likely refl ect the nonprobabilistic sampling schemes and differing sampling periods that underlie these populations. In addition, differences in pathogenicity among serogroups, a different se-rovar distribution at the serogroup level, and the fact that foodborne transmission is only 1 transmission route (5) should also contribute to the observed differences. Game might be a relevant, and as yet underappreciated, source for human STEC infection in Germany. Epidemiologic studies are needed to assess the risk associated with consumption of or contact with game.
Part of the work of the Max-Rubner Institute was fi nancially supported by the AiF-Project 12606N, funded by the Research Association of the German Food Industry. The laboratory-based sentinel surveillance of human STEC infection was funded by the German Ministry of Education and Research, Project "Emerging Foodborne Pathogens in Germany" (grants 01KI9901 and 01K10202).