With increased malignancy, lung cancer can be classified into adenocarcinoma (ADC), squamous cell carcinoma (SQC), large cell carcinoma (LCC), and the small cell subtype (SCLC); yet, elucidations to this augmented malignancy has not been addressed. In this study, we elucidated the molecular diversity among these subtypes by investigating large-scale sequencing datasets. Among genes upregulated from normal, ADC, SQC, LCC to SCLC, six hub genes were found closely correlated with adverse clinical outcome and were testified on cellular or tissue level with quantitative RT-PCR. Cox regression model was then built to generate a risk signature. The possible linkages among these genes were also explored. Transcript levels of BUB1, E2F1, ESPL1, GTSE1, RAB3B, and U2AF2 were found significantly elevated from normal, ADC, SQC, LCC to SCLC. Overexpression of one or multiple of these genes was correlated with adverse overall survival (OS) and relapse-free survival (RFS) in the whole patient cohort or groups stratified according to clinical variables, while most of all six genes were independent prognostic factors. When used as a six-gene risk signature, patients with high signature score displayed more unfavorable clinical variables and poorer outcome. Tight regulative relationships were found within these genes, while BUB1 and E2F1 were likely to be the drivers. We considered the augmented malignancy from non-small cell lung cancer (NSCLC) to SCLC might be due to the elevation of these six genes. We believe these genes were powerful cancer prognostic markers and potential therapeutic targets in lung cancer; moreover, changes of their level might be correlated with lung cancer phenotype plasticity.
Keywords: Cancer phenotype plasticity; Expression profiling; Lung cancer; Prognostic markers; Risk signatures.