Biomedical Informatics in Cancer Research

view, showing that multiple molecular pathways must be affected for cancer to develop, but with different specific proteins in each pathway mutated or differentially expressed in a given tumor (The Cancer Genome Atlas Research Network 2008; Parsons et al. 2008). Different studies demonstrated that while widespread mutations exist in cancer, not all mutations drive cancer development (Lin et al. 2007). This suggests a need to target only a deleterious subset of aberrant proteins, since any tre- ment must aim to improve health to justify its potential side effects. Treatment for cancer must become highly individualized, focusing on the specific aberrant driver proteins in an individual. This drives a need for informatics in cancer far beyond the need in other diseases. For instance, routine treatment with statins has become widespread for minimizing heart disease, with most patients responding to standard doses (Wilt et al. 2004). In contrast, standard treatment for cancer must become tailored to the molecular phenotype of an individual tumor, with each patient receiving a different combination of therapeutics aimed at the specific aberrant proteins driving the cancer. Tracking the aberrations that drive cancers, identifying biomarkers unique to each individual for molecular-level di- nosis and treatment response, monitoring adverse events and complex dosing schedules, and providing annotated molecular data for ongoing research to improve treatments comprise a major biomedical informatics need.

In the past two decades, the large investment in cancer research led to identification of the complementary roles of genetic mutation and epigenetic change as the fundamental drivers of cancer. With these discoveries, we now recognize the deep heterogeneity in cancer, in which phenotypically similar behaviors in tumors arise from different molecular aberrations. Although most tumors contain many mutations, only a few mutated genes drive carcinogenesis. For cancer treatment, we must identify and target only the deleterious subset of aberrant proteins from these mutated genes to maximize efficacy while minimizing harmful side effects.

Together, these observations dictate that next-generation treatments for cancer will become highly individualized, focusing on the specific set of aberrant driver proteins identified in a tumor. This drives a need for informatics in cancer research and treatment far beyond the need in other diseases. For each individual cancer, we must find the molecular aberrations, identify those that are deleterious in the specific tumor, design and computationally model treatments that target the set of aberrant proteins, track the effectiveness of these treatments, and monitor the overall health of the individual. This must be done efficiently in order to generate appropriate treatment plans in a cost effective manner. State-of-the-art techniques to address many of these needs are being developed in biomedical informatics and are the focus of this volume.