Gene Expression Profiling Advance for Lung Cancer Prognosis

Mayo Clinic Cancer Center researchers have identified two non-cell type-specific gene expression biomarker profiles that provide improved survival prediction for non-small-cell lung cancer. Their findings will be published in the Feb. 20 issue of the Journal of Clinical Oncology.

Gene expression profiling allows identification of a tumor’s molecular signature and can lead to better subclassification of patients. In 2006, Mayo Clinic Cancer Center researchers reported that gene expression profiling methods were not better than existing conventional methods of assessing survival in lung cancer patients. They based that conclusion on their evaluation of literature published since this promising technology started about 10 years ago. Today’s study findings appear to change this conclusion.

“While we have seen a number of exciting results from gene expression profiling in lung cancer outcome prediction, none provided value-added, unique information,” says Zhifu Sun, M.D., Mayo Clinic cancer researcher and the study’s lead author. “By saying that, we mean that none of the previously identified genetic biomarkers provided more information than conventional methods, and their clinical applications were limited. Our goal was to identify a biomarker or panels of markers that would add to our current knowledge of patient management, allowing much more refined prognostic subclassification, and eventually leading to better personalized treatment options.”

When the human genome project was completed in 2003, interest in gene expression profiling rapidly developed, as scientists sought to find the molecular roots of disease. Dr. Sun says while clinical application from gene expression profiling is still in its infancy, particularly in lung cancer, this finding is a big step forward.

Before this study’s findings, gene expression profiling in lung cancer provided no additional information than that gleaned from conventional methods that factor in age, gender, stage, cell type and tumor grade, performance status or treatment. The studies did not consider these factors in marker selection; therefore, molecular markers or signatures identified were often surrogates of what the research community already knew. Dr. Sun and the Mayo research team took a different approach. They looked at adenocarcinoma and squamous cell carcinoma separately and first selected two nonoverlapping sets of survival-related gene signatures from two large microarray datasets. This was done by adjusting for the conventional predictors. The signatures were then evaluated in two large independent datasets of non-small-cell lung cancer. Each signature was evaluated in the same cell type it was derived from, and also validated for survival prediction in other cell types.

The team found two survival-related 50-gene signatures, one for each of the two cancer types. These were nonoverlapping and largely unique in gene content compared to previously identified predictive gene expression signatures for lung cancer developed by other researchers.