Gene expression signatures for prognosis in NSCLC, coupled with signatures of oncogenic pathway deregulation, provide a novel approach for selection of molecular targets.

7020 Background: Gene microarray analysis can identify signatures that reflect unique aspects of individual tumors and provide precise prognostic information. Previously, we identified gene expression signatures reflecting the deregulation of oncogenic signaling pathways. In this study, we coupled gene expression data with the ability to identify the state of critical regulatory pathways within an individual tumor to determine prognosis. METHODS: We prepared RNA from 101 stage I non-small cell lung cancer (NSCLC) tumor samples (51 squamous, 50 adenocarcinomas) for gene expression analysis with the Affymetrix U133 GeneChip. Each group consisted of 25 patients who died within 2 years of resection and 25 patients with a >5year survival. We developed predictive models that accurately distinguished patients with good vs. poor prognosis. We validated the model with a leave-one-out cross validation, with distinct training and validation sample sets. These data were used to predict the status of Ras, Src, β-cat, E2F & Myc pathways and then analyzed by hierarchical clustering to identify patterns of pathway deregulation. Results were expressed as a probability of pathway activation and Kaplan-Meier survival analysis was performed stratifying for pathway status. RESULTS: The predictive model had 80% accuracy in distinguishing patients with respect to survival. Kaplan-Meier analysis revealed that patient subgroups defined by distinct patterns of pathway deregulation exhibited statistically significant differences in disease-free survival. Tumors with deregulated Ras and Myc pathways had much worse prognosis than those with only deregulated Ras (69% vs 20% 2-yr survival, p<0.05). CONCLUSIONS: The capacity to stratify NSCLC patients according to individual risks using genomic-based prognostic tools provides opportunity for personalized treatment decisions. The use of gene expression data to predict the status of oncogenic signaling pathways provides an opportunity to better characterize the oncogenic process, and may provide a path to selecting targeted therapeutics. Investigations are underway for EGFR, HER2-neu and VEGF pathways. No significant financial relationships to disclose.

Duke Scholars

Author Mike West Statistical Science

Author David Harold Harpole Jr. Surgery, Cardiovascular and Thoracic Surgery