ABSTRACT Chromosome aberrations are a hallmark of acute myeloid leukemia and offer mechanistic and prognostic insights into disease. As such, a combination of cytogenetic assays are routinely applied as a part of the AML diagnostic workflow. While offering invaluable information on disease severity, most chromosome aberrations fall into the “cytogenetic abnormalities not classified” or “complex karyotype” categories. A range of studies have shown that, while ambiguous, these variants have prognostic value, suggesting the existence of cryptic variants of significance or complex epistases that drive the AML phenotype. However, there is currently no system for translating genome-wide chromosomal aberration information into patient risk. To improve the predictive potential of chromosome aberration profiles, we propose the development of a risk-prediction metric that will add new prognostic value to AML studies. Specifically, we will produce a method which will establish a patient risk metric that can help guide treatment decisions for patients traditionally judged as of intermediate risk. This development will employ our scalable cytogenomic tools and novel machine learning analytics to generate a large collection of cytogenomic datasets and analyze them to identify patterns linked to AML phenotypes. Once completed, we will have a combined kit and software solution that will not only improve upon existing cytogenetic applications in AML, but will offer new prognostic insights beyond what is possible with current tools. This product will deliver high-resolution view of the chromosome aberration landscape in AML and an offer a data-driven interpretation of how variants will impact disease severity.

NARRATIVE Chromosomal aberrations are essential biomarkers used to guide treatment decisions for acute myeloid leukemia patients. In this proposal we describe the use of a comprehensive method to assay chromosome aberrations at high resolution and use this information to more accurately predict the severity of disease. This information will potentially be helpful guiding decision making when making treatment decisions for patients poorly stratified by current methods.