Prediction of Clinical Outcome in Multiple Lung Cancer Cohorts by Integrative Genomics: Implications for Chemotherapy Selection

¹JE2492, Faculty of Medicine Paris-Sud, Bicêtre, France; ²Assistance Publique Hôpitaux de Paris; ³Faculty of Medicine Paris Descartes, Paris, France; ⁴Yong Loo Lin School of Medicine, National University of Singapore; ⁵Genome Institute of Singapore; ⁶Duke-NUS Graduate Medical School, Singapore, Republic of Singapore; and ⁷Department of Cancer Biology, Wake Forest University School of Medicine, Winston-Salem, North Carolina

^* To whom correspondence should be addressed. E-mail: philippe.broet{at}inserm.fr.

	Abstract

The role of adjuvant chemotherapy in patients with stage IB non–small-cell lung cancer (NSCLC) is controversial. Identifying patient subgroups with the greatest risk of relapse and, consequently, most likely to benefit from adjuvant treatment thus remains an important clinical challenge. Here, we hypothesized that recurrent patterns of genomic amplifications and deletions in lung tumors could be integrated with gene expression information to establish a robust predictor of clinical outcome in stage IB NSCLC. Using high-resolution microarrays, we generated tandem DNA copy number and gene expression profiles for 85 stage IB lung adenocarcinomas/large cell carcinomas. We identified specific copy number alterations linked to relapse-free survival and selected genes within these regions exhibiting copy number–driven expression to construct a novel integrated signature (IS) capable of predicting clinical outcome in this series (P = 0.02). Importantly, the IS also significantly predicted clinical outcome in two other independent stage I NSCLC cohorts (P = 0.003 and P = 0.025), showing its robustness. In contrast, a more conventional molecular predictor based solely on gene expression, while capable of predicting outcome in the initial series, failed to significantly predict outcome in the two independent data sets. Our results suggest that recurrent copy number alterations, when combined with gene expression information, can be successfully used to create robust predictors of clinical outcome in early-stage NSCLC. The utility of the IS in identifying early-stage NSCLC patients as candidates for adjuvant treatment should be further evaluated in a clinical trial. [Cancer Res 2009;69(3):1055–62]

Key Words: Lung cancer, genomics, survival