ABSTRACT
In lung cancer, integrating translational data from various histologies obtained in different patients under different conditions can increase their robustness. This is a meta-analysis of cDNA array data obtained in 688 tumor patients (541 non-small cell lung cancer, 33 small cell lung cancer and 114 others) and 205 controls. 1,206 genes were found to be dysregulated in one of the 12 transcriptomics studies available. 748 results (62%) were obtained only once and might be questioned. 38% of observations could be reproduced twice or more. 346 genes were reported twice, 80 three times, 27 four and 5 five times. A common set of genes dysregulated in lung cancer was obtained, including BPA1, DUSP6, ASCL1, RNAS1 and S100P. p63 and CK 5/6 p63 are useful for differentiating adenocarcinoma and small cell lung cancer from squamous cell carcinoma. TFF-3 and MUC1 are over-expressed in adenocarcinoma. INSM1, SGNE1 and H2AFZ are typical for small cell lung cancer. Using a meta-analysis approach, it was possible to detect a robust set of genes differentially expressed in lung cancer and to determine a limited number of key genes linked to subtypes in lung cancer molecular pathology.
