Classification of Pediatric Tumors Using DNA Microarrays

Abstract

Expression profiling is of particular interest in pediatric tumors for both clinical and scientific reasons. Clinically, pediatric tumors are often histologically primitive, leading to difficulties in diagnosis, and tumors of similar morphology may have markedly different behavior and treatment responses. Despite the improving trend in survival among pediatric patients with cancer over the last 25 yr (1), several fundamental problems remain in the management of pediatric cancers. First, the choice of chemotherapeutic agents for treatment of cancer is primarily empirical in nature, based on their efficacy in clinical trials and not on targeting specific genes, proteins, or pathways known to be active in that cancer. Second, the majority of these drugs target all dividing cells, including those in normal bone marrow and mucosa, which often leads to severe dose-limiting toxicity. Third, other idiopathic sometimes fatal toxicities, such as cardiomyopathy, may occur as maximal tolerated doses are reached. Fourth, there is currently no cure for the 35% of patients with the most aggressive disease, including those with metastatic disease and those with poor prognostic molecular markers, such as gene amplification (e.g., MYCN in neuroblastoma). Finally, despite very careful clinical and pathological prognostic stratifications, 30% of patients with apparently “low-risk” cancers will die from their cancer, while a similar percentage in the “high risk” groups will survive. Therefore, there is a need for more accurate markers of prognosis and treatment response to be delineated. For these reasons, there has been an increasing emphasis on using global genomic approaches to determine the biological and molecular features of high risk cancers, correlating these with diagnosis and prognosis, and identifying new targets for therapy (Fig. 1).