Abstract
BACKGROUND: Pancreatic cancer is characterized by its unsatisfying early detection rate, rapid disease progression and poor prognosis. Further studies on molecular mechanism and novel predictive biomarkers for pancreatic cancer based on a large sample volume are required. METHODS: Multiple bioinformatic analysis tools were utilized for identification and characterization of differentially expressed genes (DEGs) from a merged microarray data (100 pancreatic cancer samples and 62 normal samples). Data from the GEO and TCGA database was utilized to validate the diagnostic and prognostic value of the top 5 upregulated/downregulated DEGs. Immunohistochemical assay (46 paired pancreatic and para- cancerous samples) was utilized to validate the expression and prognostic value of COL11A1, GJB2 and CTRL from the identified DEGs. RESULTS: A total number of 300 DEGs were identified from the merged microarray data of 100 pancreatic cancer samples and 62 normal samples. These DEGs were closely correlated with the biological characteristics of pancreatic cancer. The top 5 upregulated/downregulated DEGs showed good individual diagnostic/prognostic value and better combined diagnostic/prognostic value. Validation of COL11A1, GJB2 and CTRL with immunohistochemical assay showed consistent expression level with bioinformatics analysis and promising prognostic value. CONCLUSIONS: Merged microarray data with bigger sample volume could reflect the biological characteristics of pancreatic cancer more effectively and accurately. COL11A1, GJB2 and CTRL are novel predictive biomarkers for pancreatic cancer.