April 1998

cytokines. This study aimed to define factors which promoted survival by inhibition of apoptosis. Methods. Human colonic SEMFs were cultured as described (Y.R. Mahida et al Am J Path - in press). Apoptosis was quantified by observer blinded counting of propidium iodide stained cells. Expression of cell surface proteins was determined by FACS analysis. Extra-cellular matrix (ECM) protein synthesis was demonstrated by immuno-fluorescent microscopy. Apoptosis in non-adherent cells was determined by FACS analysis with annexin-V-FITC staining for phosphatidyl serine (PS) exposure. Results. Absolute serum deprivation of established adherent cultures of SEMFs for 24 hours caused only 3% -+ 1% (n--4) apoptosis and addition of cycloheximide up to 250pM caused no significant increase suggesting that FCS derived or soluble autocrine factors are not sole determinants of SEMF survival. Cells harvested with trypsin / EDTA which failed to re-adhere to plastic became apoptotic (PS+ve) within 8 hours suggesting adherence is critical for survival. 76%-+ 2.4% (n=16) of cells seeded onto plastic in the absence of FCS survived, the remaining 24% undergoing apoptosis within 24 hours. Immunofluorescence staining demonstrated that these cells synthesised and laid down the ECM protein fibronectin. Supplementing the media at the time of seeding with FCS reduced apoptosis to 68% -+ 3% of the levels seen with matched controls. Fibronectin reduced the levels of apoptosis to 66% -+ 5% of the levels seen with matched controls i.e. when cells were seeded with no additives (p < 0.01). When cells were seeded without supplements, exposure to the peptide RGDS (but not RGES) increased apoptosis 2.1 -+0.2 fold (n=8, p=0.0006) and inclusion of the selective fibronectin inhibitor peptide GRGDNP (but not GRADSP) increased apoptosis 1.8-+0.25 fold (n=8, p=0.01). Cycloheximide (50pM) included at the time of seeding inhibited production of fibronectin thus increasing the absolute level of apoptosis when cells were seeded in the absence of supplements to 96% -+ 4% (n=4). This was reduced to 26% -+ 4% (p=0.001) by the addition of FCS at the time of seeding. This was not a non-specific effect of FCS since fibronectin alone reduced cycloheximide induced apoptosis to 44%-+ 6% (p=0.005) showing that this ECM protein potently signals survival in SEMFs. We then demonstrated by FACS analysis that these cells express [31 and ~V133 integrins - two receptors for extracellular fibronectin. Blocking antibody to ¢tVI33 had no effect on cells seeded without serum. Anti-131 integrin antibody increased apoptosis 3.6 -+ 0.2 fold over control cells seeded without serum (n=4, p=0.002). Conclusion. Fibronectin, either when constitutively expressed or when added as a supplement, can promote SEMF survival by inhibition of apoptosis. Thus the ECM is likely to play a critical role in determining SEMF number in vivo. Fibronectin can bind to both ctVI33 and 131 integrins but only the latter appears to be important in mediating survival signals for human colonic SEMFs. This work was supported by the Wellcome Trust

• G1562

SMALL BOWEL LYMPHOMA IN SILENT CELIAC DISEASE: A CAUSE FOR CONCERN? SD Johnston, RGP Watson Dept. Medicine, Royal Victoria Hospital, Belfast, Northern Ireland

Small bowel lymphoma is a well recognized complication of typical celiac disease from which a gluten-free diet is known to give some protection. It is not clear whether this risk also applies to silent or screening-detected celiacs. Aim: To determine if small bowel lymphoma and/or adenocarcinoma are associated with the presence of unrecognized celiac disease at the time of presentation. Methods: A retrospective search was carried out by the five main Pathology laboratories serving Northern Ireland. Snomed searches were used to identify cases of adenocarcinoma (SBA) and lymphoma (SBL) affecting the small intestine between 1987-1996. The pathology reports were obtained and analysis of these was carded out with respect to clinical features, site of pathology, type of tumour, grading and the presence or absence of distant villous atrophy. Cases in which the primary tumour was thought to have spread from an adjacent site, such as, colon or ovary, and neuroendocrine tumours were excluded. Results: One hundred and thirty-eight cases were identified of whom 69 (43 male; mean age 60.2 yrs.) had SBL and 69 (51 male; mean age 68.0 yrs.) had SBA. Comparing the clinical presentation of SBL to SBA, perforation (10 v 1, p=0.009) was more common whereas small intestinal obstruction (20 v 20, p=l.0) and small intestinal mass (13 v 15, p=0.67) did not differ significantly. Fifty-four patients in the SBL group had a small intestinal resection compared to 36 in the SBA group (p=0.001). Eight cases were diagnosed in the SBL group by OGD compared to 21 in the SBA group (p=0.002). Considering the SBL group, B cell lymphomas occurred in 20 cases T cell lymphomas in 24 cases and 25 were unclassified. Considering the SBA group these were well differentiated in 7, moderately differentiated in 27, poorly differentiated in 11 and unclassified in 24. There was one known celiac in the SBL group and none in the SBA group (p=l.0). Villous atrophy at a distant site was recorded in 13 cases in the SBL group compared to none in the SBA group, all of these being T cell lymphomas (enteropathy- associated T cell lymphoma). Conclusions: Similar numbers of SBL and SBA occurred in the study period. Their clinical presentation was similar although perforation was more common in the SBL group. One-third of all SBL and over half of the T cell lymphomas had distant villous atrophy indicating that small bowel lymphoma is an important complication of silent celiac disease.

Intestinal Disorders A383

G1563

FECAL PRODUCTION AND DETOXIFICATION OF SULPHIDE. J~lr~,ensen J, Mortensen PB. Dept. of Medicine CA, Section of Gastroenterology, Rigshnspitalet, DK-2100 Copenhagen, Denmark.

Background." The oxidation of butyrate in colonocytes may be diminished in ulcerative colitis (UC). Hydrogen sulphide, found in supranormai concentrations in feces from patients with UC, is known to inhibit butyrate oxidation in colonocytes and could play a pathogenetic role in this disease. Hydrogen sulphide is, in part, bound to metal ions in the colonic lumen (metal sulphides). Only the free fraction of this compound is thought to be toxic. Purpose: To evaluate the fecal ability to produce and detoxify sulphide. Methods: A) Stools from two healthy volunteers were collected and mixed to a 25% wffwt homogenate with anaerobic isotonic NaC1 by paddle impaction in a stomacher. Homogenates with and without substrate addition were incubated for 0, 6 and 24h at 37°C, added to glass tubes topped with nitrogen, and ultracentrifuged. The supernatant was sterile filtered and free sulphide was measured in the filtrates by the methylene blue method of Cline, modified by Florin. Filtrates were processed from homogenates incubated and supplemented as shown in the Table. B) Stools from 10 healthy volunteers, 10 patients with inactive and 10 patients with active UC were collected and sulphide was measured in the filtrates as described in A). Results: A)

Concentration of sulphide (mmol/l)

Substrate addition 000h 0 6h 24h None 0.00 0.00 + NariS (3 retool/l) 0.13 0.07 0.07 + cysteine (10 mg/g) 0.26 0.26 0.36 + methionine (10 mg/g) 0.00 0.00 0.00 + blood (1 ml/10ml) 0.06 0.06 0.04

B) Mean _+ SEM. Free fecal sulphide concentrations (calculated from the concentrations in filtrates) for healthy volunteers and patients with inactive and active UC were 0.05-+0.03, 0.05-+0.01 and 0.13-+0.04 mmol/1, respectively. Conclusions: The fact that addition of sodium sulphide to make a 3 mmol/l homogenate resulted in only 0.13 mmol/l of free sulphide at Oh indicates a fecal binding capacity of sulphide of around 95%. Of the dietary sulphur-containing amino acids cysteine and methionine, the former was a larger and rapid sulphur contributor to the production of sulphide. The results in B) support earlier findings of higher sulphide concentrations in patients with active UC. The concentrations of free sulphide found in this study in healthy volunteers and in patients with UC presumably conceal a substantial production of sulphide, of which the main part, however, is detoxified by the binding to metal ions.

G1564

INCREASED PLASMA FIBRINOLYTIC ACTIVITY IN HUMAN GASTROINTESTINAL ANGIODYSPLASlA. F. Junouera. E. Saperas, A. AnglEs*, J. Monasterio* and J-R Malagelada. Digestive System Research Unit and *Hemostasis and Thrombosis Unit. Hospital General Universitari Vail d'Hebr6n, Barcelona. Spain.

Angiodysplasia (AGD) is a major cause of recurrent GI bleeding. An association of GI angiodysplasia with clotting disorders has been reported. However, no systematic investigation of bleeding diathesis as a relevant pathogenetic factor in angiodysplasia-associated gut hemorrhage has been previously undertaken. Aim: To investigate whether bleeding GI angiodysplasia is associated with a specific clotting disorder. Methods: Blood samples were obtained from 21 patients with gastrointestinal AGD (10M/llF; mean age: 71 -+ 2; age range: 58-90 yrs), three months after the last episode of bleeding. Plasma levels of von Willebrand factor (vWf) were measured by ELISA. D-dimer (D-D), plasminogen activator inhibitor 1 (PAl- 1) and tissue-plasminogen activator (t-PA) activity, as markers of fibrinolysis were measured by ELISA and a functional assay, respectively. Tissue factor pathway inhibitor (TFPI) and factor VII activated (FVIIa-rTF) as markers of clotting activation via extrinsic pathway were analyzed by ELISA and a clotting assay, respectively. As controls, we similarly studied 14 patients with previous duodenal ulcer bleeding (10M/4F; mean age: 68.1 -+ 2; age range: 59-79). Results: All AGD patients had normal plasma levels of vWf. Mean plasma vWf levels were in fact higher in AGD (208 _+ 12%) than in controls (143-+ 11%) (p<0.05). D-D levels (661-+80 ng/ml) and t-PA levels (2.04 -+ 0.14 UI/ml) were also higher than in controls D-D (395 -+ 99 ng/ml), t-PA (1.6 -+ 0.1 UI/ml) (p < 0.05), whereas levels of PAI-I were similar in both groups. Plasma levels of F VIIa-rTF (62-+7 mU/ml) and TFPI (69.5-+ 2.9 ng/ml) in AGD patients were similar to controls F VIIa-rTF (50 -+ 9 mU/ml) and TFPI (69.5-+ 2.6 ng/ml). There was no correlation between plasma levels of any of these factors with age, sex, associated diseases or treatments used. However, PAI-1 levels (31.5 -+ 11 ng/ml) were lower in AGD patients with high bleeding rate (> 2 epis0des/patient/year) than in those with low bleeding rate (< 2 episodes/patient/year) PAI-1 (47 +- 14 ng/ml) (p < 0.05). In a regression analysis, plasma levels of PAI-1 was a significant predictor of further hemorrhage (p < 0.05). Conclusion: Increased plasma fibrinolytic activity may contribute to bleeding from human gastrointestinal AGD. A low plasma PAI-1 levels is a risk factor for AGD bleeding tendency.