Concentrations of interleukin 6 and tumour necrosis factor in serum and stools of children with Shigella dysenteriae 1 infection

Serum interleukin 6 (IL-6) and tumour necrosis factor (TNF) were measured in children with dysentery during an epidemic caused by Shigella dysenteriae 1. IL-6 and TNF were also measured in fresh stool filtrates from children with acute gastroenteritis. The median serum IL-6 concentration was raised significantly in the children with complications (haemolytic uraemic syndrome, leukemoid reaction, thrombocytopenia, thrombocytosis, and severe colitis lasting more than one week) during the first week (n= 18, 9-7728 pg/ml; median 107) and in the second week (n= 13, 5312 pg/ml; median 77), compared with convalescent sera (n=10, <3-85 pg/ml; median 39; p<002 and <0*05 respectively). The median IL-6 concentration during the first week was significantly higher in the group with complicated disease than in those with no complications (n=8, <3-37 pg/ml; median 5; p<0001). Although serum TNF concentrations were significantly raised in the complicated group during the first and second weeks of the illness and in the uncomplicated group compared with convalescence, there was no significant difference in the TNF concentrations between the complicated and uncomplicated groups. IL-6 was detectable in stool filtrates from eight of 13 children with S dysenteriae 1 infection and four of eight children with S flexneri infection. It was not detectable in Cryptosporidia, rotavirus, or adenovirus infections, those with pathogen-negative acute diarrhoea or controls. Seven of 13 children with S dysenteriae 1 and three ofnine children with Sflexneri infections had TNF detectable in stools. Two of four children with adenovirus infection also had TNF detected in stools. None of the children with Salmonella, Cryptosporidia, rotavirus or children with pathogen-negative diarrhoea and controls had detectable TNF in stool filtrates. It is postulated that the local and generalised vasculitis observed in shigellosis may be related to a direct effect of Shiga toxin on endothelial cells or caused by cytokine production stimulated by endotoxin, or both. (Gut 1993; 34: 194-198) Most children with dehydration caused by watery diarrhoea can be effectively treated with oral rehydration therapy. By contrast mortality from bacterial dysentery still occurs primarily in patients with Shigella dysenteriae 1 infections. The problem of dehydration is eclipsed by the systemic effects of the infection that include haemolytic uraemic syndrome.' In Galle, Sri Lanka, 5-10% of children admitted with bacterial dysentery during epidemics develop haemolytic uraemic syndrome (unpublished observations). Rahaman et al first reported the association between S dysenteriae 1 infection and haemolytic uraemic syndrome. A potential common factor between haemolytic uraemic syndrome associated with infections caused by S dysenteriae 1 and certain strains ofEscherichia coli (most notably E coli serotype 0157:H7)' is the production of a cytotoxic enterotoxin (Shiga toxin and Verotoxin respectively). Shiga toxin and Verotoxin (VT) share biological effects that may be related to their ability to inhibit protein synthesis.45 The release of endotoxin during shigellosis could also damage the capillary endothelium, either as part of a generalised Schwartzman reaction6 or as a result of release of cytokines such as tumour necrosis factor (TNF)7 or interleukin 6 (IL-6).8 We therefore measured IL-6 and TNF concentrations in serum and stool from children with bacterial dysentery to determine whether there was a correlation between these measurements and the severity of the disease and appearance of complications, especially haemolytic uraemic syndrome.