The IL-18-IFNγ axis predicts response to immunotherapy

ABSTRACT CAR T-cells have revolutionized the treatment of pediatric leukemia. However, ~50% of responsive patients eventually relapse and in ~50-70% of patients, this therapy induces devastating side effects (i.e. neurotoxicity (NTX) and cytokine release syndrome (CRS)), potentially leading to long term neurological damage and death. There are no clinical biomarkers to predict survival or toxicity despite a crucial need. SCRI has made a substantial investment in CAR T therapy and is a world leader in pediatric CAR T cell clinical trials. This proposed project will build upon this program by helping to predict and mitigate the devastating, life-threatening side effects caused by this revolutionary therapy and may aide in the prediction of non-response cases. Data from our pilot experiments utilizing samples from patients on CAR-T trials at SCRI suggests that pre-monocyte activation status and cytokine profiles from the monocyte fraction of a patient can predict CAR T toxicity across both blood and solid tumors, in particular we see trends in the IL-18- IFN𝛾𝛾 axis. We plan to act on this encouraging pilot data in this proposal to develop a series of validated monocytic biomarkers (cytokine, flow and transcriptional) and machine learning algorithms which can be utilized prior to patients going on trial in order to save time and resources for patients who are set to fail treatment and allow medical teams to be more prepared to mitigate toxicity in those who are more prone to it. The proposal further expands to the development a humanized mouse model that can adopt the healthy donor and pediatric patient immune systems recapturing natural variability in human immunity to investigate causality of the IL-18- IFN𝛾𝛾 axis in driving both therapeutic toxicity. The project combines existing but unused large bank of SCRI CAR T patient samples/data with monocytic omics profiling to develop predictive biomarkers for therapeutic toxicity (AIM 1). We aim to further extend the impact of this work to mechanistically understand why certain patients’ cells are working against them through the use of novel humanized mouse models (AIM2). This is built upon to develop therapeutics or strategies to improve safety and efficacy of CAR-T treatments. Importantly, our team of experts in clinical CAR T (Dr. Rebecca Gardner, Dr. Navin Pinto), machine learning (Dr. Bobbie-Jo Webb-Robertson), humanized mice (Dr. James Keck), and myeloid tumor biology (Dr. Heather Gustafson) pose this grant for optimal success.