Ho',vever, the anemia persisted despite treatment of these lesions. In both of these patients the anemia resolved after the antral polyps were removed endoscopically. No follow-up information was available in three of the patients Conclusions: Hyperplasric polyps of the gastric antrum are a rare (< 1% of patients undergoing upper endoscopy for iron deficiency anemia), but significant cause of gastrointestinal blood loss in older patients with anemia. Removal of the polyps using endoscopic or surgmal methods may be required for resolution of the blood loss along with oral iron supplementation. Conjectures: Gastroenterologists should be aware that h)`~erplastic polyps of the gastric antrum might result in gastrointestinal blood loss and iron deficiency anemia.

T1680

Referral Rates are Low for Genetic Evaluation in Patients Meeting Bethesda Guidelines for HNPCC Shflpa Grover, Elizabeth Tschoegl, Laoti Russo, Sapna Syngal

INTRODUCTION: The Bethesda criteria aim to identify individuals who may benefit from mierosatellite instability (MS1) testing and further genetic evals~ation for hereditary nonpolyp- osis colorectal cancer (HNPCC). The prevalence of individuals with colorectal cancer (CRC) who meet the Bethesda criteria for HNPCC and the frequency of their referral for genetic evaluation is unknowu. In addition, the relative usefulness of each of the seven Bethesda criteria has not been reported. MEIHODS: 433 individuals with CRC presenting for their first visit to a gastroenterology cancer clinic at a specialized cancer center completed a family history question2aaire regarding cancer diagnoses in first and second-degree relatives. Family history forras were reviewed in addition to physicians' notes to identify individuals who satisfied the Bethesda criteria, and to determine the percentage referred for genetic evaluation (MSI testing and genetic assessment) Tumor histology was reviewed to determine the prevalence of right-sided and undifferentiated cancers. RESULTS: Twenty percent (87/433) of CRC patients met the Bethesda criteria for HNPCC. Of these individuals, 1% (1/87) fiflfilted the Amsterdam cmena, 11% (ll /87) had synchronous CRCs and 28% (24/87) of patients with CRC had a first-degree refative with an HNPCC associated tumor, w/th one cancer diagnosed at age <45 yrs. 71% (62/87) of subjects had a diagnosis of CRC prior to 45 yrs. Of these young onset CRCs, 13% (8/62) were fight-sided undifferentiated tumors and 8% (5/62) demonstrated signet cells on histopathology. Four percent (17/433) of individuals in the entire cohort and 16% (14/87) of subjects who fulfilled the Bethesda criteria were reterred for genetic evaluation. Individuals who satisfied the Bethesda criteria were more likely to be referred for genetic testing (p<0.0001). In the Bethesda subgroup, 7].% (10/14) of sobjects that received recommendations for genetic testing had young onset CRC6. However, these individuals were not more likely to be referred for evaluation than individuals who met the first three Bethesda criteria (p=0.5). CONCLUSIONS: Although individnals who fulfill the Bethesda cntena are more likely to be advised a genetic assessment, the raaNrity of patients remain unrecognized. Increased awareness of the Bethesda guidelines as well as referral for genetic evaluation are necessary for the appropriate management of high-risk CRC patients.

T1681

Genetic Predisposition for Alcohol-Associated Upper Aerodigestive Tract Cancer and Hepatocellular Carcinoma in Heavy Drinkers with the Alcohol Dehydrogenase 3"1 Allele Felix Stickel, Ntis Horaann, Monika Beneso~a, Andrea Jacobs, Detlef Schuppan, Eckhart G. Hahn, Christian Conradt, Dieter Ludwig, Ina Zuber, Hubert E. Bhira, Helmut K Seitz

Chronic alcohol consumption is associated with an increased risk tbr upper aerodigestive tract cancer (UADTC) and hepatocellular cancer (HCC). Acetaldehyde (AA), a known carcinogen in rodents, is believed m be primarily involved in ak'ohoi-associated carcinogenesis. Indeed, Asians with high AA levels following alcohol ingestion due to a mutation o f AA dehydrogenase 2 (ALDH2) have an increased risk of UADTC In Caucasians, polymorphism of ADH3 occurs and the ADH3*I allele encodes for an enzyme with a higher capacity" to generate AA than ADH3*2. Recent epidemiological studies resulted in contradictory resuhs regarding ADH3 po]ymorphism and alcohol-associated cancer. We, therefore, determined the ADH3 geno- types in heavy alcohol consumers with and without cancer. Altogether, 187 patients with UADTC (laryngeal cancer 41, oropharyngeal cancer 46, esophageal cancer 100), 75 with aIcohotic cirrhosis and HCC, 424 age-matched alcoholic patients without cancer (alcoholic cirrhosis 164, alcoholic pancreatitis 146, alcoholics without organic disease 114), and 48 health),' controls were included. ADH3 genotypes were assessed by PCR followed by restriction fragraent length pnlymorphism on white blood cell DNA. The allele frequency of the ADH3* 1 allele was slgnifieanfly increased in heavy drinkers with upper UADTC and in patients with HCC in cirrhotic livers compared to age-raatched alcoholic controls without cancer (UADTC: 61%, HCC: 61%, alcoholic controls without cancer: 46%, heafihy controls: 50%; p<0.01, beth instances). In addition, homoaygosit)`' of ADH3* 1 was also significantly higher in cancer patients (249 alcoholics with cancer: 35% vx 424 alcoholics without cancer: 19%; p<O.00 i). These data suggest that heavy drinkers horaozygons for the ADH3* 1 allele carry a genetic risk to develop UADTC and HCC, respectivel

T1682

Non-lnsulin Dependent Diabetes Mellitus (N1DDM) and Colorectal Cancer Risk Yu-Xiao "fang, Sean Heraressy, James D. Lewis

Background & Aims: Hyperinsulinemia in tire context of NIDDM may play an important rote in the pathogenesis ot colorectal cancer. However, the clinical evidence for the association is inconclusive. In this study within a large general practice population, we sought to determine whether the prevalence of NIDDM is higher among patients with colorectal cancer than those without colorectal cancer. Methods: We carried out a case-coturol study in the United Kingdom based General Practice Research Database (GPRD). All patients with incident colorectal cancer d~agnoses in the GPRD were included as cases (n = 10,473). Up to 10 controls v~thont a diagnosis of colorectal cancer were matched with each case on three

conditions: sex, year of birth, and duration of GPRD follow-up prior to the case index date. The pnmary exposure was an),' diagnosis of NIDDM recorded in the GPRD prior to the index date. Odds ratios (ORs) were calculated using conditioilal logistic regressmn. Potential confounders examined included cholecystectomy, smoking status, body mass index (BMI) & NSAID/aspirin nse. Results: A prior diagnosis of NIDDM was associated with a significantly increased risk of colorectal cancer, with a crude OR of 1.55 (95% CI: 1.37-1.76) and a multi,,~ariable OR of 1.35 (95% CI: 1.19-1 53). Much of the difference between the crude and rnuhivariable results was due to adjustment for BMI. The association between NIDDM and colorectal cancer was observed in both men (muhivariable OR 129; 95% CI: 1.09~ 1.51) and women (nudtivariable OR 1.43; 95% CI: 1.18-1.75). Conclusions: N1DDM is associated with an increased risk of colorectal caner risk in both males and females. BMI appears to account for a portion of the risk increase

T1683

Sensitivity, specfity and clinical value of MMR immunohis tochemistry for the predict ion of microsatellite instability Karsten Schulmann, Frank E. Brasch, Erdraute M. Kanstmann, Stephan A. Hahn, Wolff Schmiegel

Introduction: The identification of MSI-H cancers requires the analysis of at least five defined raicmsatellite markers. The analysis consists of the dissection of tumor and normal tissue by micro- or macrodissection, DNA isolation, PCR amplification and geleIectrophoresis which is time consummg and expensive. In contrast, Immunohistochemistry for MMR proteins can be easily implemented in routine histology, is cheap and requires no normal tissue. Methods: We prospectively determined the MSI stares according to the Bethesda guidelines of 153 lesions from 131 patients. Immunohistochemiatry (MLH1, MSH2, MSH6) was performed prior to the MSI study and was blinded until the MSI status was determined. We analysed 112 CRC, 20 colorectal adenomas, and 21 other extracolonic cancers of the HNPCC tumor spectrum. Extracoloinc cancers were mainly obtained from patients fnlfillmg the Amsterdam (AMS) criteria. 43 patients met the AMS-Criteria, 54 met at least one of the Bethesda-Criteria, 2 patients had a MTS, 12 patients did not fulfill the AMS- or Bethesda cmeria. In 20 patients no familiy history could be obtained Results: MSI-I-I ',,,,as detected m 40/i 12 (36%) of CRC, 6/20 (25%) of colorectal adenomas, and in 12/21 (57%) extracolonic cancers (AMS 68%, Bethesda 26%). 44/58 (76%) of MSI-H cancers showed loss ot one or two MMR proteins. Two additional MSI-H cases were not examined by IHC. In contrast none of 87 MSS/MSI-L cancers showed an abnormal MMR protein expression Discussion: The sensitivity of iramnnhistochemistry for the prediction of MSI-H is 76%, whereas specifity is 100% The negative predicitive value is 86% and the positive predicirive value is 100%. To our opinion MSI analysis could be restricted to patients with normal MMR protein expression By applying this two step diagnostic algorithm time and costs of MSI-H analysis can be reduced significantly without reducing the sensitivity ",,,'hen compared to MSI analysis alone Furthermore imraunohistoehemlstry gnides sequencing of MMR genes in HNPCC.

T1684

Familial Colorectal Cancer (CRC) Risk Communicat ion among Colorectal Adenoma (CRA) Patients Paul C. Schroy IlI, Subodh Lal, Sheifa Wilson, Timothy C Heeren, Alan Geller, Francis A. Farraye

Background: The first-degree relatives (FDRs) of CRA patients diagnosed at age <60 may be at comparable risk to the FDRs of CRC patients and may warrant screening beginning by age 40 (Gastruenterolog)`, 1997;112:594-642). The main objective of this study was to assess the extent to which channels of communication exist between physmians, CK~ patients and their FDRs regarding familial risk and screening recommendations Methods: We con- ducted a telephone survey of 147 consecutive newly diagnosed English-speaking CRA patients age <60 treatedby 12 endoscopists at Boston Medical Center (n = 115) or Harvard Vanguard Medical Associates (n = 32). Information regarding patient demographics~ awareness of familial risk, and extent of risk communication to FDRS was elicited. Results: A total of 81 (55%) of eligible patients responded to the survey. Responders ranged in age from 38 to 59 years (median, 54 years) and were mostly male (59%) and white (67%). Few responders (n=27; 33%) were aware that their FDRs were at increased risk of CRC Only 16 of the 27 (59%) knowledgeable patients identified a physician (primary care > endoscopist) as the source. Awareness was strongly associated (p<0.O01) with patient-initiated information- seeking from their primary care providers rather than unsolicited physician communication. Awareness was not associated with age, gender, race/ethnicity, site, endoscopist, indication, or level of education. Most knowledgeable CRA patients (n=23, 85%) reported that they had informed one or more FDRs about their diagnosis and told them to either get screened (41%) or tell their primary care physician at their next visit (22%). The majority (68%) of responders felt that it was the patient's responsibility to notify their FDP, s. Conclusions: Communication about familial CRC risk between CRA patients and their FDRs is poor. Failure to notify FDRs is mostly attributable to lack of CRA patient awareness about the implications of their diagnosis, which in turn, may reflect a lack of provider knowledge (Am J Gastm 2002; 97:1031-6), lack of provider commumcation, and/or ineffective pro- vider communication.

1"1685

Pancreatic Endocrine Tumors (PET) in Von-Hippel Lindau (VHL) Disease : Genotype / Phenotype Correlations Olivier Corcos, Anne Couvelard, S. Giraud, Valerie Vilgrain, S. Richard, Dermot O'Toole, Philippe Raszniewski, Pascal Hammel

PET account for 12-15% of pancreatic lesions in patients (patients) with \~tL disease. However, their characteristics are ill-defined. Airas-1) to assess clinical, morphological and histological patterns of PET in "~IL disease : 2) to look for genotype-phenotype correlations after characterization of germline mutations. Patients and methods - PET in patients with

A-547 AGA Abstracts