AGA Abstracts T2038 A Case-Control Study of the Risk Factors for Advanced Serrated Polyps Christopher S. Huang, Francis A. Farraye, Courtney Shepherd, Shi Yang, Michael J. O'Brien Background: It is postulated that some hyperplastic polyps (HPs) can progress to advanced serrated polyps (ASPs), which include sessile serrated adenomas and dysplastic serrated polyps, and ultimately to cancer. Endpoint cancers of this serrated pathway are frequently characterized by microsatellite instability (MSI). Little is known about the factors that determine risk of this progression. In this study we tested the hypothesis that smoking, previously shown to be associated with MSI cancer, is an independent risk factor for ASPs. Specific Aim: To identify the risk factors for ASPs in patients with one or more serrated polyp (SP).Methods: Case-control study. Cases were defined as patients with at least one ASP, and controls were those with only typical HPs. All polyps were classified using the criteria of Torlakovic et al.[Am J Surg Pathol 2003;27:65-81]. Subjects with conventional adenomas or history of colon cancer were excluded. Results: A total of 267 subjects with 512 SPs were included (101 cases, 166 controls). Univariate tests of associations (Table 1) showed that polyp size and location were strongly associated with risk of ASPs. Current smoking appeared to be protective. Of note, smokers were on average 3.4 years younger than non-smokers (53.0 vs. 56.4, respectively, p=0.005). On multivariate logistic regression analysis (Table 2), larger polyp size and presence of proximal SPs proved to be independent risk factors for ASPs. Current smoking and presence of distal SPs were associated with reduced risk, although these trends were not statistically significant.Conclusions: In this study, larger polyp size and proximal location were associated with advanced pathology in serrated polyps, but smoking was not. Table 1 Univariate tests Table 2 Multivariate analysis T2039 Colorectal Cancer Incidence Rates for American Indian and Alaska Native Populations Vary Five-Fold Between Regions, with a Propensity Toward the Distal Colon and Later Stage At Diagnosis David G. Perdue, Carin I. Perkins, Jeannette Jackson-Thompson, Steven S. Coughlin, Faruque Ahmed, Donald Haverkamp, Melissa A. Jim BACKGROUND. Colorectal cancer (CRC) is a leading cause of cancer morbidity and mortality in American Indian and Alaska Native (AI/AN) populations. Problematically, AI/ANs are often misclassified as other races in medical records, resulting in underestimates of disease burden. Though the overall incidence of CRC in AI/ANs has been reported to be lower than that for non-Hispanic Whites (NHWs), summary statistics may obscure important regional rate differences for this vastly diverse and heterogeneous group. METHODS. To reduce racial misclassification, we linked high-quality National Program of Cancer Registries (NPCR) and Surveillance, Epidemiology, and End Results (SEER) data with Indian Health Service (IHS) user population records to compare CRC incidence, stage at diagnosis, and colonic location between AI/AN and NHWs. We limited our analysis to populations living in counties served by the IHS and stratified the results by IHS service regions. RESULTS. From 1999- 2004, the overall risk of CRC was 9% lower among AI/ANs than NHWs (RR=0.91, 95% CI 0.87-0.95). However, there was a nearly five-fold regional variation in incidence among AI/ANs, with rates ranging from 21.0 per 100,000 in the Southwest to 102.6 per 100,000 in Alaska. By comparison, rates for NHWs ranged from 46.8 per 100,000 in the Southwest to 55.2 per 100,000 in the East. Incidence rates were significantly higher among AI/ANs than NHWs in the Alaska (RR=2.03, 1.78-2.31), Northern Plains (RR=1.39, 1.26-1.52), and Southern Plains (RR=1.16, 1.08-1.25) regions, while rates were lower in the Pacific Coast (RR=0.80, 0.72-0.88), East (RR=0.65, 0.53-0.79) and Southwest (RR=0.45, 0.40-0.50). AI/ ANs were more likely to have advanced CRC at initial diagnosis with a rate ratio to localized disease of 1.92 (95% CI 1.74-2.12) compared to 1.48 (95% CI 1.46-1.50) for NHWs. Females were more likely than males to be diagnosed with CRC proximal to the splenic flexure among both AI/ANs and NHWs (females 40.1% and 50.1%, males 33.5% and 40.3%, respectively). Overall, AI/ANs had proportionately more distal cancers than NHWs with A-606 AGA Abstracts proximal to distal rate ratios 0.63 and 0.81 respectively. CONCLUSIONS. The heterogeneity of AI/AN populations and their risk factors for CRC are reflected in the five-fold difference in CRC rates between IHS regions. Efforts to make CRC screening a priority, resources to overcome barriers to endoscopic screening, and work to engage AI/AN communities in culturally-appropriate ways to participate are needed. T2040 Poor Compliance with MSI-Analysis in Patients with Colorectal Cancer At High Risk for Lynch Syndrome Margot G. van Lier, J. de Wilt, J. Wagemakers, W. Dinjens, R. Damhuis, E. J. Kuipers, M. E. Leerdam van Background: Lynch syndrome (LS) is responsible for approximately 3-5% of all colorectal carcinomas (CRC). Microsatellite instability (MSI) is the molecular hallmark of LS. In 2004 the revised Bethesda criteria were published to improve the efficiency of recognizing LS by identifying LS related malignancies that should be analyzed for MSI. The aim of this study was to evaluate whether MSI-analysis was performed in CRC patients at high risk for LS according to the Bethesda guidelines in the Southwestern part of the Netherlands. Methods: All patients diagnosed with invasive CRC in 2005 and 2006 in the study region were selected from the database of the regional Comprehensive Cancer Center Rotterdam (CCCR). Data from the department of pathology performing all MSI-analyses in the complete CCCR region between January 2005 and August 2007 were available. Patients were included if they met any of the following criteria, derived from the Bethesda guidelines: (1) patients diagnosed with CRC < 50 years, (2) patients < 60 years with CRC displaying mucinous or signet-ring differentiation or medullary growth pattern, and (3) patients diagnosed with a second LS- associated tumor prior to the diagnosis of CRC in 2005/2006. Patients were excluded when they had previously been identified as LS mutation carrier. MSI performance rates were tabulated and analyzed using chi square statistics and multivariate logistic regression. Family history and presence of tumor-infiltrating lymphocytes or Crohn's-like lymphocytic reaction in the tumor could not be part of the analyses since these data were missing. Results: Of the 1882 patients diagnosed with CRC, 173 (9%) met any of the inclusion criteria. One known LS mutation carrier was excluded. MSI-analysis had been performed in 23 (13%) of the remaining 172 patients. MSI was performed in 18 (23%) of the 79 included patients < 50 years of age, and in one (5%) of the 21 patients between 50-60 years with high-risk pathology markers. Of the 72 patients with a second LS-related tumor, 4 (5,5%) were referred for MSI-analysis. In patients with CRC < 50 years MSI-analysis was significantly more often performed than in patients fulfilling one of the other 2 inclusion criteria (p < 0,01). Gender and tumor localization showed no significant correlation with performance of MSI-analysis. Conclusions: Despite multidisciplinary approach of oncology and easy access to a tertiary referral center and clinical genetics facilities, there is marked underutilization of MSI-analysis in patients at high risk for Lynch syndrome. As a result Lynch syndrome might be underdiagnosed in patients with CRC and their relatives. T2041 Evaluation of Mutation Prediction Models in Lynch Syndrome D. Ramsoekh, Monique E. van Leerdam, Anja Wagner, E. J. Kuipers, E. W. Steyerberg Background: Lynch syndrome (LS) is a common hereditary colorectal cancer syndrome that is characterized by a high risk for developing colorectal cancer (CRC). Therefore, it is of utmost importance to detect subjects that are affected by LS. However, the diagnosis of LS is hampered by the absence of specific diagnostic features, such as the presence of hundreds of polyps in the colon. In the recent years, several models have been developed to predict the likelihood of carrying a germline mutation. The aim of the present study was to validate five prediction models for the presence of mutations in the mismatch repair genes. Methods: We used data of 321 families who were referred to the department of Clinical Genetics of the Erasmus Medical Center between 1995 and 2004 because of a family history of CRC. These data were used as input for 5 different, previously published models, i.e. the PREMM, Leiden, Edinburgh, UK-Amsterdam Plus and UK-Alternative model. External validity was assessed by discriminative ability (Area under the receiver operating characteristics curve, AUC) and calibration (Hosmer-Lemeshow goodness-of-fit test). For further insight, predicted probabilities were categorized as 5% or less, 5.1% to 10%, 10.1% to 20%, 20.1% to 40% and more than 40%. Results: Of the 321 families, 66 were diagnosed with a germline mutation (25 MLH1, 23 MSH2 and 18 MSH6). All models discriminated well between high risk and low risk families (AUC: 0.82-0.84), but the calibration was poor for the Leiden, UK-Amsterdam Plus and UK-Alternative model. With a 5% probability cut off, the UK- Alternative Model had the highest sensitivity for detecting a mutation (100%), while the Leiden model had the lowest sensitivity (73%). The sensitivities of the PREMM, Edinburgh and UK-Amsterdam Plus model were 98%, 99% and 82% respectively. However, at this 5% cut off, specificities were very different (UK-Alternative: 87%, Leiden: 20%, PREMM: 78%, Edinburgh: 91% and UK-Amsterdam Plus: 45%). Conclusions: Mutation prediction models have a high accuracy in predicting the presence of a germline mutation in the mismatch repair genes. The poor calibration of some models however hampers direct applica- tion of probability cut-offs in clinical practice. T2042 Mismatch Repair Gene Product Expression and Colorectal Cancer in Hispanics Wilfredo E. De Jesus-Monge, Ronghua Zhao, Carmen Gonzalez-Keelan, Stanley R. Hamilton, Miguel Rodriguez-Bigas, Marcia R. Cruz-Correa Background & Aims: Colorectal cancer (CRC) is a leading cause of morbidity and mortality, and ethnic variations in incidence and mortality have been described. Alterations in nucleotide mismatch repair (MMR) genes are responsible for CRC account for approximately 10-15% of all cases. Methylation of the MLH1 MMR gene leading to transcriptional silencing and mutations and deletions in the MLH1 and MSH2 genes account for over 95% of MMR- defective CRC. Immunohistochemistry (IHC) for MMR proteins in these tumors usually