A l 1 5 2 AASLD ABSTRACTS GASTROENTEROLOGY, VOI. IO8, No. 4

• EXTRACELLULAR BRAIN AMINO ACID CHANGES IN RELATION TO NEUROLOGICAL STATUS IN PORTAL-SYSTEMIC ENCEPHALOPATHY: RESULTS OF AN IN VIVO CEREBRAL MICRODIALYSIS STUDY. V.L. Raghavendra Rao, R. Audet and R.F. Butterworth, Neuroscience Research Unit, HOpital Saint-Luc (University of Montreal), Montreal, Qua., Canada H2X 3J4

Portal-systemic encephalopathy (PSE) is characterized by neuropsychiatric symptoms progressing through stupor and coma. Previous studies of human autopsy material suggest that amino acids may play a role in the pathophysiology of PSE (LavoJe et aL, J. Neurochem., 1987). As part of a series of studies of the role of brain amino acids in PSE, extracellular amino acids were measured sequentially in relation to neurologic status using in vivo cerebral microdialysis in frontal cortex o f portacaval shunted rats administered ammonium acetate (NH4OAc) (3.8 mmol/kg, i.p.) to precipitate severe PSE. Sham-operated animals served as controls. Amino acids were measured as their o-phthalaldehyde derivatives by HPLC with f luorescence detection. Portacaval shunting alone resulted in selectively increased extracellular concentrations of glutamine (increased 3-fold,~p<0.01), alanine (increased 38%, p<0.01), aspartate (increased 4 4 % , p<0.05), phenylalanine (increased 170%, p<0.01), tyrosine (increased 140%, p<0.01) and tryptophan (increased 63%, p<0.01). Deterioration of neurological status following NH4OAc treatment resulted in further increases of extracellular phenylalanine, tyrosine and alanine but n o further increase of aspartate, glutamate or tryptophan. Recovery of neurological function was accompanied by normalization of extracellular concentrations of alanine and phenylalanine. Thus, whereas, alterations of neurotransmitter precursor amino acids (phenylalanine, tyrosine, tryptophan) as well as the excitatory amino acid aspartate characterize early PSE, impending hepatic coma is characterized by superimposed alterations of brain energy metabolism, characterized by increased alanine. Both neurotrans- mitter-related events and a deficit of cerebral energy metabolism appear to be modified in PSE. Therapeutic measures aimed at the correction of the cerebral energy deficit (such as L-carnitine) could be useful in human PSE. (Funded by The Medical Research Council o f Canada)

DOES COLON CANCER DEVELOP AFTER LIVER TRANSPLANTATION FOR PRIMARY SCLEROSING CHOLANGITIS ASSOCIATED WITH INFLAMMATORY BOWEL DISEASE? M. Rahmi.~n D. Chu, L. Kim Schluger, F. Klion, S. Guy, H. Bodenheimer. Division of Liver Diseases and Liver Transplantation, The Mount Sinai Medical Center, New Yok, NY. Objective: " To assess the development of colon cancer after liver transplantation for primary sclerosing cholangitis associated with inflammatory bowel disease. _Methods:, Thirty eight patients underwent orthotopie liver transplantation for primary sclerosing cholangitis associated with ulcerative or crohn's colitis between September 3, 1988 and May 19, 1994. The medical records of these patients were retrospectively reviewed. Results: 2/38 (5%) died within 3 months because of graft non-function or surgical complication. Two patients died at 14 months and 36 months due to cholangiocarcinoma and prostate cancer respectively. Two patients had colon surgery prior to liver transplantation. One patient had a segmental resection for colon cancer 6 months prior to liver transplantation and remains well at 63 months. One patient had a proctocolectomy for fulminant colitis 20 years prior to liver transplantation., 32/38 patients were evaluated for the development of colon cancer after liver transplantation. The age ranged from 17- 73 years with a mean age of 47: The mean duration of colitis prior to liver transplantation was 11.3 years with a range of 5-30 years. Colonoscopy was performed within 14 months of liver transplant in each patient. No colorectat malignancies were detected at the time of liver transplantation. 2/32 (6%) had low grade dysp!asia prior t o liver transplantation. One of these patients had a proctocolectomY six months after successful liver transplantation for high grade dysplasia. The other patient remains cancer free 24 months after liver transplant. One Patient with a 30 year history of ulcerative colitis developed high grade dysplasia 7 months after liver transplantation. No patient developed colon cancer. The range of followup after liver transplantation is 12-79 months with a mean followup of 37 months. Conclusions: To date, no increased risk of development of colon carcinoma in patients who have undergone liver transplantation for PSC associated with ulcerative or crohn's colitis is identified. We recommend continued screening for the development of dysplasia and colon carcinoma as indicated by clinical symptoms and duration of colitis.

• REGULATION OF TUMOR NECROSIS FACTOR a (TNF) BIOACTIVITY BY KUPFFER CELLS (KC) DURING THE REGENERATIVE RESPONSE TO LIVER INJURY. R.Rai, S.Q. Yang and A.M. Diehl: GI Div., Johns Hopkins University, Baltimore, MD.

Using neutralizing antibodies to TNF, we showed that an injury related cytokine (TNF) plays a key role in triggering proliferation of surviving hepatocytes after partial hepatectomy (PH). TNF enhances hepatocyte proliferation and acts within minutes of PH to activate a growth related kinase (jun nuclear kinase (jnk)) and certain growth-regulatory genes (c-jun,, jun B. C/EBP ,8, C/EBP 6). TNF is not detected in plasma post-PH, suggesting local release. In order to clarify if Kupffer cells (KC) produce TNF post-PH, gadolinium chloride (GdCI) (10 mg/kg i.v.) was used to deplete KC: PHs were done 24 h later and induction of TNF- regulated transcription factors and liver regeneration were compared in GdCI-treated and NaCI-treated controls 0.5-48 h post-PH. GdCI depleted KC since phagocytosis and lysis of I a25- and CrSl-labelled E.Coli were decreased ~79% 24 and 48 h post-GdCl. In control, little c-Jun, .C/EBP B or C/EBP 6 protein was detected on immunoblots of nuclear extracts pre- PH but within 3-6 h post-PH nuclear concentrations of c-Jun and C/EBP p protein each increased 10 fol~ and that of C/EBP 6 increased 16 fold. Gel mobility shift assays with specific antisera demonstrated that the DNA binding activities of c-Jun, C/EBP B and C/EBP 6 also increased post-PH in controls. GdCI had no effect on the nuclear levels of any of these proteins before PH but amplified the induction of all 3 proteins post-PH by 200-400%. It also increased hepat0cyte proliferative activity at both 24 h and 48 h ;}ost-PH. At 24 and 48 h, nuclear expression of PCNA, an S phase specific antigen, was significantly greater in GdCI-treated rats (62% and 72%) than controls (51% and 19%). The number of mitoses/HPF were also doubled at 24 and 48 h (3-2 HPF in controls vs. 6-8/HPF in GdCI, p < 0.05). However. despite improving regenerative induction of several hepatic transcription factors and increasing hepatocyte proliferation. GdCI caused significant mortality (27% vs. 0% in controls) within 8 h post-PH. CONCLUSIONS: GdCI-mediated KC depletion increases hepatocyte proliferation after PH but is associated with increased mortality. Since I GdCI enhances the expression of genes which are inhibited by TNF neutra zaton, KC depleton may de-repress PH nducton of TNF bioactivity in other cell types.

NEURAL NETWORKS (NN) AS MEASURES OF ILLNESS SEVERITY IN ALCOHOLIC HEPATITIS (All) S. Rajan t, M. Bonacini 2, and P. Lapuerta 1, Division of General Internal Medicine I and Division of GI and Liver Diseases 2, University of Southern California, Los Angeles, CA.

The Maddrey discriminant function (MDF) is the most widely used prognostic indicator in AH. AIM: We developed and evaluated NNs as a new measure of illness severity in AH utilizing commonly available clinical and laboratory parameters. PATIENTS: Charts with complete data on admission WBC. albumin, bilirubin, creatinine, protime, and the presence or absence of ascites, encephalopathy, GI bleeding, and SBP were selected for review in patients satisfying diagnostic criteria for AH. Records on 59 deaths and 85 survivors were selected (144 total) in order to obtain adequate samples of both outcomes. METHODS: NNs and logistic regressionmodels (LR) were developed to predict in-hospital mortality with the above-mentioned parameters on NeuralWorks ® Professional II/Plus software. Model development and validation was organized according to a modified jack-knife technique (Wu et at.. Radiology 1993: 187:81-87). The MDF was based on bilirubin and protime, which was calculated by converting our lab's protime (%) into seconds. A regression analysis of MDF values with outcomes in our dataset was performed to enhance the predictive value Of the MDF scores. Our evaluation included a comparison of outcome predictions and construction of receiver operating characteristic (ROC) curves of true positive vs. false positive rates for NN. LR, and MDF, RESULTS: NNs provided scores which correlated with mortality (mean NN score 0.59 in deaths. 0.28 in survivors, p <0.001). NN scores were able to successfully stratify patients into low. intermediate, and high risk groups with mortality rates of 9,7%. 23.8%, and 64.7% respectively. Laboratory and clinical parameters both had significant impacts on NN predictions. NNs predicted 109 of 144 individual outcomes correctly, but the MDF predicted only 93 ( p < 0.05), and LR predicted 104. Areas under the ROC curves were 79.8%, 79.8% and 71.6%, respectively, for NN. LR. and MDF. with larger areas indicating a higher probability of correctly ranking patients. CONCLUSIONS: NNs are promlsing tools for measuring illness severity in AH. They can stratify patients according to mortality risk. and they have a greater predictive value than MDF. Analysis of clinical and laboratory values together provides a better outcome prediction than simply two laboratory values (bilirubin and protime) alone. With the current database size and variables collected the NN and LR model performances are similar.