BACKGROUND: Gastric cancer and esophageal cancer are the second and sixth most common causes of cancer death worldwide. Both of these upper gastrointestinal tract (UGI) cancers have a very poor prognosis, largely because symptoms usually do not occur until late in the disease. Significant reduction in UGI cancer mortality will probably require development of new prevention strategies based on identification of new modifiable risk factors and development of new methods to diagnose and treat more cases at earlier, more curable stages of disease. PURPOSE: The aims of this project are (1) to examine hypotheses relating to the etiology and prevention of upper gastrointestinal cancers, and (2) to develop successful clinical strategies for the early detection and treatment of these cancers. METHODS: Both etiologic and early detection studies are most efficiently done in high-risk populations, so many of the studies in this project are performed in such populations. (1) Etiologic studies: (a) China: Between 1985 and 1991, we conducted two randomized nutrition intervention trials (NIT), in Linxian, China, a county where cumulative death rates due to esophageal squamous cell carcinoma (ESCC) and gastric cardia adenocarcinoma (GCA) exceed 20%. These trials, with a combined enrollment of nearly 33,000 people, evaluated the effect of supplementation with several vitamin/mineral combinations on UGI cancers and found significant reductions in total mortality and gastric cancer mortality among those taking a combination of selenium, beta-carotene and vitamin E. Since 1991, we have followed the NIT participants as a cohort, and have performed nested studies evaluating the association between baseline characteristics and later development of UGI cancers. In addition, we have performed other etiologic studies specifically designed to evaluate individual exposures such as polycyclic aromatic hydrocarbons (PAHs), acetaldehyde, human papillomavirus (HPV), and Helicobacter pylori (H. pylori). (b) Iran: We have recently performed two etiologic studies in Golestan Province, Iran, another population with very high rates of ESCC. Between 2003 and 2007, the Golestan Case-Control Study enrolled 300 ESCC cases, 571 neighborhood controls and 300 clinic controls, and between 2004 and 2008 the Golestan Cohort Study enrolled 50,000 participants. Comparing results from high-risk populations in Iran and China, which are quite distinct geographically, ethnically and culturally, should give us insight into which environmental and genetic risk factors are most important for the development of this disease. (c) Brazil: We have begun preliminary studies evaluating PAH exposure in a high ESCC risk population in southern Brazil; (d) Kenya: We have begun a feasibility study for a case-control study of ESCC in western Kenya, which has high rates of this disease and the unusual occurrence of 8% of the cases in individuals younger than 30 years old. (2) Early Detection of Esophageal Cancer: This part of the project includes four studies: (a) the Cytology Sampling Study (CSS), to develop practical and accurate primary screening tests for esophageal squamous dysplasia and early ESCC; (b) the Mucosal Staining Study (MSS), to develop methods for endoscopic localization of squamous and glandular dysplasia of the esophagus and stomach; (c) the Endoscopic Therapy Pilot Study (ETPS), to evaluate new techniques for endoscopic treatment of high-grade squamous dysplasia and early ESCC; and (d) the Chemoregression Study (CRS), to test chemoprevention strategies to reduce progression and increase regression of low-grade esophageal squamous dysplasia. PROGRESS: (1) Etiologic studies: (a) China: Recent follow-up through 10 years after the end of the Linxian General Population Nutrition Intervention Trial has shown continued significant reductions in total mortality and gastric cancer mortality among those taking a combination of selenium, beta-carotene and vitamin E, especially among those who were less than the median age (55 years) at the beginning of the trial. Results from nested studies in the NIT cohort have shown: (i) a strong association between low serum selenium levels and increased ESCC and gastric cardia cancer risk; (ii) no relation between serum carotenoids and risk of UGI cancers; (iii) a strong association between low serum alpha-tocopherol levels and increased ESCC risk; (iv) a strong association between low tissue zinc levels and increased ESCC risk; (v) increased risk for both cardia and non-cardia gastric cancer among subjects with positive serology for H pylori; (vi) no association between positive serology for HPV 16, HPV 18, or HPV 73 and ESCC, gastric cardia cancer or non-cardia gastric cancer; (vii) no relation between fumonisin exposure and ESCC risk; (viii) an association between tooth loss and risk of UGI cancers; (ix) an association between self-reported goiter and non-cardia gastric cancer; and (x) associations between several genetic polymorphisms and risk of ESCC and/or gastric cardia cancer. We also found no evidence of HPV DNA or HPV oncogene activity in fresh-frozen tumor samples from 272 consecutive ESCC resection specimens in Linxian. (b) Iran: Results from the Golestan Case-Control Study have shown an significant associations between case status and tobacco use, opium use or both; poor oral hygiene; drinking hot-temperature black tea; and various measures of low socioeconomic status. Comparison of normal esophageal tissue biopsies from cases and controls has shown a striking dose-response relationship between PAH exposure in the esophageal tissue (measured by intensity of immunohistochemical staining with anti-PAH antibodies) and case status, consistent with a causal role for PAH exposure in ESCC carcinogenesis. The accrual for the Golestan Cohort Study has also now been completed, and the subjects are being followed annually for vital status and cancer endpoints. (c) Brazil: Recent results from southern Brazil have show an association between urinary 1-hydroxypyrene glucuronide (a PAH metabolite) and consumption of mate, a local tea which has consistently been associated with ESCC risk. Laboratory analysis of dry mate leaves and mate beverages have shown high levels of PAHs in both. (d) Kenya: We have recently begun a feasibility study for the Thinker case-control study of ESCC in western Kenya. (2) Early Detection of Esophageal Cancer: (a) CSS: We recently analyzed a CSS study in which 720 asymptomatic Linxian adults were examined by one of two esophageal balloon samplers, followed by blood collection and endoscopy. The sensitivity/specificity of the two samplers for identifying biopsy-proven esophageal squamous dysplasia or cancer were 39%/85% and 46%/84%, insufficient for general use as a primary screening test for these lesions. (b) CRS: We recently completed the analysis of our first chemoregression study. This study randomized 267 Linxian patients with biopsy-proven mild or moderate squamous dysplasia to one of four treatment groups: selenomethionine 200ug once/day, celecoxib 200mg twice/day, both agents, or neither. Endoscopy was performed at baseline (T0) and after 10 months of intervention (T10). The primary endpoint was change in each patients worst dysplasia grade between T0 and T10 (regression/stable disease/progression). 238 patients (89%) completed the trial. Analysis showed that celecoxib treatment did not influence dysplasia grade (p=0.78). Selenomethionine had no effect in patients who began the trial with moderate dysplasia (p=1.00), but showed a significant improvement in dysplasia grade in patients who began with mild dysplasia (p=0.02).