Whole-genome association studies of complex human diseases represent a new paradigm in the post-genomic era. We developed a sequential method to identify genetic loci that can predict risk of developing esophageal squamous cell carcinoma (ESCC). First, using the generalized linear model (GLM) with adjustment for potential confounders and multiple comparisons, we identified 37 SNPs associated with disease. Second, using the 37 SNPs identified from the GLM in a Principal Components Analysis (PCA), the first principal component correctly predicted 46 of 50 cases and 47 of 50 controls. Among all the SNPs selected from GLMs for the three modes of transmission, 39 could be mapped to one of 33 genes. Many of these genes are known to be involved in various cancers, including GASC1, previously shown to be amplified in esophageal squamous cell carcinomas, and EPHB1 and PIK3C3. In conclusion, we have demonstrated the feasibility of the Affymetrix 10K SNP array in genome-wide association studies of common cancers and identified new candidate loci to study in ESCC. To identify genes involved in the etiology of upper gastrointestinal (GI) cancers and develop diagnostic markers for them, we performed a genome-wide gene expression analysis for three GI cancers, including 54 pairs of esophageal squamous cell carcinoma (ESCC), 62 pairs of cardia cancer (CC), and 72 pairs of body of stomach cancer (BC). We observed a hierarchical structure that is consistent with anatomical structure and histopathology of the three subtypes of GI cancers using principal components analysis (PCA). We identified 2132 genes that showed significant changes between normal and tumor cells in all three GI cancers. We also identified 1326, 1253, and 470 genes that were altered in tumors uniquely for EC, BC, and CC cancer etiology, respectively. We found that 12 gene otology (GO) groups are significantly enriched for the genes that are commonly altered in the three GI cancers. Those are the genes that are generally thought to be important for cancer development such as genes involved in the cell cycle, mitosis, and DNA replication. In contrast, six GO groups that are significantly enriched for the genes with differentiation expression in the BC/CC branch belong to more specialized biological processes such as rhodopsin-like receptor activity and G-protein coupled receptor protein signaling pathway. Our study provides a novel integrated approach to investigate cancer etiology at systems biology level.
Showing the most recent 10 out of 11 publications
