Project Summary The treatment of cancer has dramatically increased in complexity over the last several decades. Five-year relative cancer survival for all cancers has improved from 49% in the 1970’s to 68% in the 2000’s. This improved outlook is the result of careful and systematic evaluation of many drug combinations. However, cancer remains the second-leading cause of death in the United States and the leading cause of years of potential life lost. The explosion of treatment options for any given cancers has introduced a complexity of choice which is troubling for most clinicians. As just one example, there are at least 70 distinct regimens that have been evaluated in randomized trials for the postoperative (adjuvant) treatment of breast cancer. Given the number of possible drug combinations for nearly every subtype of cancer, a full comparison of all treatment options in the clinical trial setting can never be done. Adding to this complexity is that the very definition of cancer is changing, as genomic information regarding prognosis and treatment selection is brought into the clinic. This deluge of information has exceeded the cognitive capacity of cancer clinicians. One solution to this information problem is to increasingly rely on expert-driven guidelines or proscribed pathways. However, many cancer scenarios are already complex enough that guidelines do not, for example, provide ranked recommendations for treatment options. With few exceptions, published guidelines also rarely incorporate tumor biology. We have previously introduced new information theoretic methods to quantitatively compare the value of treatment regimens that have never been compared directly, as well as network analytic methods to begin to tie tumor biology into the clinical setting. Our group also has experience with advanced data extraction techniques and experience developing standardized software applications that can be utilized by clinicians and researchers. In response to PAR-15-332 (Early Stage Development of Informatics Technologies for Cancer Research and Management), we will: 1) Produce a comprehensive ontology of chemotherapy regimen concepts; 2) Determine a treatment preference hierarchy based on information theoretic network analysis, and present the results to clinicians and cancer researchers; and 3) Present preference hierarchy-modulated genomic treatment options. Our software will be evaluated on two common cancers: breast cancer and multiple myeloma and will be made freely available for non-commercial uses. We believe that the software developed as a result of this work will be widely applicable to a variety of cancer types, to the benefit of clinicians and the patients that they care for.

Project Narrative The treatment of cancer has dramatically increased in complexity over the last several decades, with associated improvements in survival; however, cancer remains the second-leading cause of death in the United States and the leading cause of years of potential life lost. The very large number of treatment options will become even more complicated as cancer genomics is added to traditional clinical information, and cancer clinicians (oncologists) need new approaches to interpret, organize, and transform this data into knowledge-driven clinical decisions. We will develop new methods and software to analyze complex biological and treatment networks; the tools developed as a result of this work will be evaluated for breast cancer and multiple myeloma, and should be applicable to a wide variety of cancer types.