Presentations focused on the molecular characterization of VTEC virulence genes and mobile genetic elements and the evolution of pathogenic clones. The diagnostic and typing assays performed by National Reference Laboratories were also discussed. The discussion underlined the existence of a previous paradigm for the definition of pathogenic strains, based on the association of a few VTEC serotypes (O157, O26, O111, O103, and O145) that cause serious human disease. A new paradigm was proposed that depends on the detection and subtyping of virulence genes regardless of serotype. Consensus was reached that 1) the term enterohemorrhagic E. coli based on serogroups can be misleading, 2) an alternative nomenclature based on patterns of virulence genes associated with severe disease is needed, and 3) the combination of virulence genes that make a strain pathogenic needs to be determined.

The incidence of VTEC in human infections and animal reservoirs is poorly defined in many European countries. Therefore, participants recommended that surveillance in humans and livestock should be improved and that standardized, rapid, and cost-effective molecular detection and typing methods for surveillance should be developed.

Future challenges in host-microbe interactions will be to decipher the mechanisms by which the type III secretion system required for colonization of bovine and human intestines becomes fully functional and how the bacterial effector proteins injected into the host cell and the regulation of their expression manipulate host signal transduction pathways. The humoral antibody response to VTEC is also important in determining the course of the disease. Research in this field may be exploited for developing standardized procedures for the serodiagnosis of human infections.

Neither the mechanism of VT translocation from the intestinal lumen to the blood stream and from blood components to renal endothelial cells nor intracellular trafficking of the toxin is fully understood. Only a small number of genes, which are mainly involved in inflammatory responses, are upregulated in human endothelial cells exposed to VT. VT2 has stronger effects than VT1, and this observation may provide an explanation for the epidemiologic association between VT2-producing strains and HUS. A better understanding of the kinetics of VT from the intestinal lumen to the renal endothelium is needed.

As for E.coli O157, non-O157 VTEC are common in ruminants, yet many types are not commonly implicated in human disease. Differences in prevalence and serotype distribution are reported and could be due to different diagnostic approaches. The harmonization of sampling and detection strategies and the development of molecular assays based on the detection of virulence genes may improve the comparison of data. Such approaches could also be applied to the detection of VTEC in foodstuffs.

Intervention strategies are, as yet, poorly developed and need to be based on fundamental knowledge of the colonization processes in ruminants. In cattle herds, individual animals may be heavily colonized by E. coli O157, and these animals can be considered to be "supershedders." Therefore, the level of public health risk and the related intervention strategies should be related to the level, rather than presence or absence, of animal colonization and excretion.

Factors involved in the survival in and transmission from animal reservoirs should be better determined. In this respect, the development of appropriate experimental models of infection would be crucial.

Animal colonization may be influenced by management factors such as diet, general husbandry practices, and season. The immune status of the host, including innate immunity, has a significant effect on the levels of colonization, and vaccination may yet offer a means to reduce VTEC carriage by food-producing animals. Increasing our knowledge of host responses and innate immunity to this infection would enable the development of targeted vaccines to reduce animal colonization.