Background and aims: Smoking tobacco has opposite effects on the different forms of inflammatory bowel disease (IBD). It predisposes to the development of Crohn's disease (CD) yet is associated with a reduced incidence of ulcerative colitis (UC). We have studied sib pairs discordant for both smoking and IBD phenotype (UC or CD) to investigate whether smoking determines the type of IBD that develops in individuals with very similar genetic susceptibility. Patients: Smoking habits and disease characteristics were analysed in 242 IBD pedigrees (658 patients). Within this group there were 339 affected sibling pairs of whom 89 were discordant for smoking when diagnosed. Results: Smoking at diagnosis was associated with development of CD (odds ratio (OR) 3.55; 95% confidence limits 2.50-5.02; p<0.001) in all of the familial patients, with increases when analysed for ileocaecal disease, fibrostenosis, and intestinal resection. Smokers were also protected from UC (OR 0.28; 0.2-0.4; p<0.001). Of 89 sibling pairs discordant for smoking at diagnosis, 23 were also discordant for disease type-in 21 of these, CD occurred in the smoker and UC in the non-smoker (OR 10.5; 2.6-92; p<0.0001). Conclusions: Smoking is a strong environmental risk factor for Crohn's disease and increases the likelihood of needing surgery. However, sib pairs who are discordant for both smoking and IBD type almost always show CD in the smoker and UC in the non-smoker, and so in some cases tobacco consumption acts on IBD genetic predisposition to shift the phenotype from UC towards CD. The explanation of part of the apparent "protective" effect of smoking on sporadic UC may be that the form of IBD that develops in a proportion of smokers is not UC but CD.

Background-Investigation of anti-colon antibodies may be simplified if a sensitive method and homogeneous source of antigen were available. Aims-To examine the anti-colon antibody response using human colonic carcinoma cell lines as antigen. Subjects-Patients with inflammatory bowel disease and other gastrointestinal disorders and healthy controls were studied. Methods-Comparative enzyme linked immunosorbent assays (ELISAs) were performed to assess the value of whole Caco-2, HT-29, and LS-180 cells as antigen. The antigenic determinants of the immune response were characterised by western blot analysis. Results-Sera demonstrated immunoreactivity against each of the cell lines, but diVerent epitopes were recognised. Applying whole Caco-2 cells as antigen in an ELISA, the prevalence of anti-colon antibodies was significantly greater in patients with ulcerative colitis (36%) than Crohn's disease (13%), other gastrointestinal disorders (13%) and healthy controls (0) (p<0.05). The immune response was not associated with one predominant antigen. Conclusions-Fixed whole cell ELISA is a simple and feasible method for studying the anti-colon antibody response. This response is non-specific, being directed against multiple antigens, and is likely to be an epiphenomenon of inflammatory bowel disease, more so for ulcerative colitis than Crohn's disease. (Gut 1999;44:196-202)