PROJECT SUMMARY/ABSTRACT Relapsed ALL is associated with poor outcome and remains the leading cause of cancer-related death among all childhood cancer. Current therapies are toxic and can result in high incidence of late effects such as infertility and heart failure. Thus, it has been a standard practice to allocate newly diagnosed patients to therapies based on predicted risk of relapse. The presence of residual cancer cells after induction chemotherapy, known as minimal residual disease (MRD), is a highly significant prognostic variable. However, many patients not considered to be “high risk” still experience relapse. There is an unmet need to develop novel risk models with enhanced accuracy to enable allocation of patients to risk-adapted therapies to reduce the likelihood of future relapse. Our prior genomics studies on relapsed ALL in protein-coding regions (~2% of the human genome) have revealed novel insights on the drivers of resistance to therapy. While these findings have potential for developing novel molecular risk models, significant knowledge gaps remain. First, more than 50% of relapsed cases lack any known resistance drivers. Second, the known resistance drivers are derived from retrospective studies of relapsed specimens and it is unclear how to apply such information prospectively from initial diagnosis to decrease the likelihood of relapse. Our goal is to develop novel molecular risk models by tracking resistance drivers at diagnosis. Our central hypothesis is that resistance drivers, when detected at diagnosis, will be informative for allocation of patients to risk-adapted therapies. We will test our hypothesis in three aims. In Aim 1, we will identify comprehensive resistance drivers from both protein-coding and non-coding regions by leveraging a large cohort of 669 relapsed childhood ALL cases from a recently completed cooperative clinical trial with genome and transcriptome sequencing data available. We hypothesize that unexplored non- coding regions will harbor novel resistance drivers and that the large cohort size will empower the discovery of rare resistance drivers. In Aim 2, we will backtrack the resistance drivers at diagnosis by using ultra-deep sequencing coupled with state-of-the-art computational error suppression that will enable detection of rare variants with frequency as low as 0.01%. In Aim 3, we will investigate if resistance drivers pre-exist in an independent cohort of patients at diagnosis. We will develop novel molecular risk models by comparing prevalence profiles of resistance drivers detected at initial diagnosis between patients who have relapsed and those who are cured. Successful completion of our project aims will deliver 1) comprehensive knowledge of drivers of resistance to therapy, 2) the full spectrum of pre-existing resistance drivers at diagnosis, and 3) novel molecular risk models for decreasing the risk of relapse. Our deliverables will form the basis for future clinical trials and for de...