Cancer gene profiling for response prediction.

Preoperative treatment strategies are now recommended for a variety of human cancers. Unfortunately, the response of individual tumors to a preoperative treatment is not uniform, and ranges from complete regression to resistance. This poses a considerable clinical dilemma, because patients with a priori resistant tumors could either be spared exposure to radiation or DNA-damaging drugs, i.e., they could be referred to primary surgery or dose-intensified protocols could be pursued. Because the response of an individual tumor as well as therapy-induced side effects represent the major limiting factors of current treatment strategies, identifying molecular markers of response or for treatment toxicity have become exceedingly important. However, complex phenotypes such as tumor responsiveness to multimodal treatments probably do not depend on the expression levels of just one or a few genes and proteins. Therefore, methods that allow comprehensive interrogation of genetic pathways and networks hold great promise in delivering such tumor-specific signatures, because expression levels of tens of thousands of genes can be monitored simultaneously. During the past few years, microarray technology has emerged as a central tool in addressing pertinent clinical questions, the answers to which are critical for the realization of a personalized genomic medicine, in which patients will be treated based on the biology of their tumor and their genetic profile (1-4).