Gene expression profiling: Does it add predictive accuracy to clinical characteristics in cancer prognosis?

摘要: It is widely accepted that gene expression classifiers need to be externally validated by showing they predict the outcome well enough on other patients than those from whose data classifier was derived. Unfortunately, gain in predictive accuracy as compared established clinical prognostic factors often not quantified. Our objective illustrate application of appropriate statistical measures for this purpose. In order compare accuracies a model based only and plus classifier, we compute decrease inaccuracy proportion explained variation. These have been obtained three studies published classifiers: survival lymphoma patients, breast cancer diagnosis lymph node metastases head neck cancer. For our results indicate varying possibly small added variation due classifiers. Therefore, future should routinely demonstrated measures, such ones recommend.