Predicting lung cancer survival

15 Aug 2008

Scientists have compiled the largest ever study to validate prognostic models of lung cancer patient survival. In addition to providing the largest available set of microarray data with extensive pathological and clinical annotation for lung tumours, this study is a potential milestone for the clinical application of predictive models of patient survival.

Gene expression signatures can predict survival of patients with lung cancer but it is crucial to establish their performance across different subject populations and laboratories before they are used in the clinic.

Online in Nature Medicine, David Beer and colleagues report the largest, most comprehensive validation study to characterise the performance of several prognostic models based on gene expression for 442 lung tumours. They examined whether measurements of gene expression alone or combined with clinical variables such as tumour stage, age or gender could be used to predict overall survival in lung cancer patients.

Several of the prognostic models they examined produced risk scores that substantially correlated with actual patient survival. Most models performed better with clinical data, supporting the combined use of clinical and molecular information when building prognostic models for early-stage lung cancer.