AGA Abstracts
syndrome cases. Aims: To determine the sensitivity and specificity of: 1), MSI, IHC or both for identifying carriers of mismatch repair gene mutations among persons with CRC; and 2) BRAF mutation or MLH1 promoter methylation testing to identify sporadic CRCs in cases with abnormal MLH1 IHC. Methods: Systematic searches were completed querying MEDLINE, EMBASE, Health Technology Assessment, and Cochrane Central Register Cochrane Database until November 2013. We included all fully published randomized trials or observational studies assessing: 1) MSI and/or IHC tumor testing in adults with CRC; 2) BRAF mutation or MLH1 promoter methylation testing to identify sporadic CRCs in cases with abnormal MLH1 IHC. Diagnostic testing meta-analyses were conducted with results reported as odd-ratios (OR) with 95% confidence intervals (CI). Results: For topics 1 and 2, respectively, 1374 and 859 articles were retrieved by the search, and 41 and 25 articles were included. Table 1 lists the number of studies and patients that were ultimately included per topic. The pooled estimates (and 95% CI) for the sensitivity and specificity of tumor testing for identifying germline mutations were 0.93 (0.87;0.96) and 0.79 (0.70;0.86) for MSI; 0.91 (0.85;0.95) and 0.83 (0.77;0.88) for IHC; and 0.97 (0.90;0.99) and 0.75 (0.62;0.84) for MSI and IHC (Table 1). The pooled estimates for the sensitivity and specificity of further testing of tumors with abnormal MLH1 IHC for identifying sporadic tumors were 0.76 (0.60;0.87) and 0.96 (0.60;0.99) for BRAF mutation testing; and 0.75 (0.59;0.86) and 0.94 (0.79;0.98) for MLH1 promoter methylation testing. The latter results demonstrated substantial heterogeneity with a very wide prediction range (Table 1). Conclusions: Our findings can inform population-wide screening strategies as well as the care of individual patients. Screening CRCs for MSI or abnormal IHC should identify most CRCs due to Lynch syndrome. MLH1 methylation testing and BRAF mutation testing appear to be comparable for identifying sporadic CRCs with hypermethylation, but neither is perfectly specific. Even though these tumor tests have good test performance characteristics, normal tumor tests should probably not dissuade clinicians from considering germline testing in appropriate patients with clinically-suspected Lynch syndrome. Tumor tests to screen for Lynch syndrome
Mo1993 Identification of Individuals at High Risk of Gastric Cancer for Targeted Endoscopic Screening Zhu Feng, Li Lin Lim, Calvin J. Koh, David E. Ong, Lee Guan Lim, Khek-Yu Ho, Chia Chung-King, Christopher J. Khor, Choon Jin Ooi, Kwong Ming Fock, Jimmy B. So, Wee Chian Lim, Khoon-Lin Ling, Tiing Leong Ang, Andrew S. Wong, Jaideepraj Rao, Andrea Rajnakova, Ming Teh, Manuel Salto-Tellez, Supriya Srivastava, Yik Ying Teo, Khay Guan Yeoh Background: Endoscopic screening for gastric cancer is useful for the detection of early gastric neoplasia; however there is scant data to guide the selection of individuals at increased risk. The Gastric Cancer Epidemiology and Molecular Genetics Program (GCEP) is a prospective multicentre study initialized in 2004 with the aims of identifying predictive risk factors for gastric cancer in the Singapore Chinese population. The study was presented at DDW 2013 (Lim et al, Gastroenterology 2013;144(5):S-95-S-96). Objectives: To identify individuals at high risk of gastric cancer in the Singapore Chinese population. Method: Chinese subjects aged >50 years were recruited into this cohort study and endoscopy surveillance was offered for a minimum of 5 years. All subjects gave informed consent. Approval from institutional review boards was obtained. The main outcome measure is the detection of early gastric neoplasia (EGN), including high grade dysplasia or adenocarcinoma. Risk factors (RF) were identified by univariate analysis, and those satisfying p<0.15 were entered in backward regression model. Receiver operating characteristic (ROC) analysis was performed for the risk prediction model. Results: 3033 Chinese with mean age 59±7 years were recruited. The study is still in progress and will be completed by 2015. So far, 21 cases of de novo EGN have been detected during prospective surveillance. 2649 subjects with complete data results were analysed. Eight risk factors including age, education, smoking, alcohol consumption, Helicobacter pylori (Hp) seropositivity, serum pepsinogen index (PGI), atrophic gastritis, and intestinal metaplasia (IM) were selected for backward regression analysis. The most parsimonious model in risk prediction included four RF, age >70, smoking, serum PGI and IM with adjusted odds ratios (95%CI) of 3.17 (1.19-8.47), 3.49 (1.45-8.41), 3.80 (1.33-10.91), 4.167 (1.20-14.50) respectively. The area under curve (AUC) in identifying EGN was 0.76 (95%CI 0.64-0.87). 64% of the cohort and 91% of EGN had at least one of these 4 RF. The cohort was grouped into high risk (HR), moderate risk (MR) and average risk (AR) based on RF >2, 1 and 0 respectively. The prevalence of EGN in the HR, MR and AR groups was 2.3%, 0.5% and 0.2% respectively. The subjects in the HR group had 11.33fold (95%CI 2.55-50.40) increased prevalence of EGN compared to those with no RF. Conclusions: In our study, individuals with >2 of 4 RF (age >70, smoking, serum PGI, IM) comprised 19% of the cohort and were at 11-fold increased risk of EGN. These criteria could be useful to risk-stratify high risk individuals for endoscopic surveillance to detect EGN. Table 1. Distribution of subjects, prevalence of EGN and odds ratios among risk groups.
*studies with data available for carriers and non-carriers were analyzed. Mo1995 Healthcare Utilization Was an Independent Predictor of Outcome in Colorectal Cancer Patients From Northern Iran Shahryar Semnani, Zahra Noorafkan, Mohammad Aryaie, SeyedMehdi Sedaghat, Fatemeh Ghasemi-Kebria, Gholamreza Roshandel Introduction: Increasing trends in the incidence rates of colorectal cancer (CRC) have suggested it as an important health problem in the developing world. Healthcare utilization was considered as an indicator of outcome in different conditions including cancers. We aimed to assess healthcare utilization (HU), its determinants, as well as its relationship with survival in CRC patients from Northern Iran. Methods: This cross-sectional study was conducted on new pathologically-confirmed primary CRC cases, diagnosed between 2006 and 2008 in Golestan province of Iran. Data on demographic factors and characteristics of tumor including stage were obtained from the Golestan population-based cancer registry. Stage of tumor was identified using instructions of summary staging system, developed by the surveillance epidemiology and end results (SEER) program. Information on HU was collected using a valid questionnaire, considering 8 diagnostic and 4 therapeutic services. The validity and reliability of the questionnaire was approved in a previous study from our region. The results were categorized as good and poor HU. Multivariate logistic regression analysis was used to assess the relationship between HU and other variables. Cox-regression analysis was performed to assess the effects of HU and other variables on patients' survival. Crude and adjusted hazard ratios (HR) as well as 95% CIs were calculated. Results: In total, 227 new cases of CRC were enrolled. The mean (SD) of participants' age was 53.62 (15.87) years with interquartile range of 44-66 years. 138 (60.79%) of participants were male. The stage of tumor was identified in 209 cases (92.1%), of which, 42 (20.1%), 105 (50.2%) and 62 (29.7%) were localized, regional and distant metastasis, respectively. HU could be assessed in 218 subjects (96%). Living in rural areas was the strongest variable related to poor HU (adjusted OR=2.65; 95%CI: 1.30-5.40). The median survival time was 40.5 months. The 1, 3 and 5- year survival rates were 71%, 52% and 44%, respectively. Figure 1 shows the survival curve in CRC patients according to levels of HU. Cox regression analysis showed a significant lower survival rate in patients with poor HU (adjusted HR= 2.3; 95%CI: 1.46-3.64). Conclusion: HU was an independent predictor of survival in our CRC patients. Patients' place of residence was a significant determinant of HU. Regarding its effects on patients' outcome, HU and its determinants should be considered in designing CRC controlling programs in our population as well as other high-risk areas.
Mo1994 Colorectal Tumor Testing to Identify Persons With Mutations in Mismatch Repair Genes (Lynch Syndrome): Systematic Review and Meta-Analysis Uri Ladabaum, James Ford, Myriam Martel, Alan N. Barkun Background: Lynch syndrome confers a lifetime colorectal cancer (CRC) risk approaching 80%. Intensive surveillance can mitigate this risk substantially, but many persons with Lynch syndrome remain undiagnosed. Routine colorectal tumor testing for microsatellite instability (MSI) or abnormal immunohistochemistry (IHC) is promoted as a strategy to identify candidates for Lynch syndrome genetic testing. Sporadic CRCs with MSI or abnormal MLH1 IHC due to somatic MLH1 promoter methylation must be distinguished from Lynch
AGA Abstracts
S-762