Identifying children with subclinical neurocognitive decline and susceptibility to oxidative damage during the early months of therapy for ALL This proposal is in response to Provocative Question 7, “How can prediction models be developed and used to identify patients at highest risk of treatment related complications?” We have previously identified genetic variants conferring susceptibility to oxidative stress that are associated with inferior cognitive function after treatment for acute lymphoblastic leukemia (ALL). In preparation for a clinical trial testing whether antioxidant therapy can protect against treatment-induced neurocognitive decline, we will test the hypothesis that leukemia therapy induces oxidative damage and measurable neurocognitive decline among susceptible individuals during the first months of ALL therapy. The proposed prediction model will therefore identify patients with childhood ALL at a time when a proactive intervention might prevent permanent treatment-induced cognitive deficits. We will test this hypothesis in three related Specific Aims: (1) Prospectively demonstrate that subclinical treatment-induced changes in cognitive function can be detected in the first three months of treatment for ALL and predict dysfunction 1 year after treatment among children being treated for ALL on Dana Farber Cancer Institute ALL Consortium protocol 16-001 at eight sites in the United States and Canada; (2) Prospectively demonstrate relationships between treatment-induced changes in neurocognitive functioning and targeted polymorphisms in genes conferring susceptibility to oxidative stress. and (3) Prospectively identify relationships between gene variants and changes in biomarkers indicative of oxidative damage within the central nervous system. Neurocognitive function will be assessed using a well-validated computer-based instrument (Cogstate) at multiple time points during the two years of therapy and one year after completion of treatment, allowing detection of subclinical changes in neurocognitive abilities from baseline. Latent Class Mixture Modeling will be used to resolve distinct patterns of change in performance over time, and patterns of changes from baseline during the first months of treatment will be linked to deficits persisting among survivors more than one year after completion of treatment. This project will therefore identify patients with treatmentrelated changes in neurocognitive function (Aim 1) at a point in therapy when an intervention might prevent further decline. In addition, the project complements our laboratory efforts to understand the pathophysiology of treatment-related cognitive decline, by testing the relationship between variants in genes related to oxidative stress and cognitive decline (Aim 2) and biomarkers of oxidative damage (Aim 3). Thus, the results obtained from this proposal are expected to have a positive impact because they will provide the foundation to improve the therapeutic index of cancer therapy, and potentially guide clinical trials of protective interventions, aimed at reducing the permanent burdens of curative treatment for leukemia.