State-transition and leukemia potential dynamics to inform disease evolution and adaptive therapy

PROJECT SUMMARY Acute myeloid leukemia (AML) is a group of aggressive and highly heterogeneous malignancies with poor overall survival. Despite advances in identification of molecular prognostic factors, it remains challenging to predict or tailor optimal individualized treatment options. We recently reported the application of a state-transition model to view AML initiation and progression as trajectories of the transcriptome in an AML state-space characterized by a leukemogenic potential. We successfully constructed a health-to-leukemia transcriptome state-space using time-sequential RNA-seq data collected from a murine genetic model of AML driven by the CBFB-MYH11 (CM) leukemogenic fusion gene that is created by inv(16)(p13.1q22), a cytogenetic/molecular subtype accounting for approximately 8-10% of AML patients. Analysis of transcriptome trajectories in the leukemogenic potential allowed us to mathematically identify state-transition critical points associated with key leukemogenic events and to accurately predict disease development and outcome. We now propose to utilize the transcriptome movement in the leukemia potential as a dynamic biomarker that can be used to design adaptive treatment approaches to overcome treatment resistance and identify new therapeutic approaches. We have recently developed a microRNA-126 inhibitor (miRisten) which effectively inhibits a highly treatment resistant leukemia stem cell population in several leukemia models. Our preliminary time-series RNA-seq data pre- and post-chemotherapy show that the transcriptome trajectory can accurately predict therapy response in murine models of AML. The central hypothesis and theoretical concept of this proposal is that the dynamics of the transcriptome and the leukemia potential can be used to predict therapy response and guide optimization of adaptive therapeutic approaches to mitigate treatment resistance. Specifically, we will model the transcriptome dynamics following treatment with anti-leukemia therapies, estimate state-transition critical points, and therapeutic force to optimize therapy dose and combinations. We propose the following specific aims: Aim 1. Quantify leukemia potential dynamics driven by different oncogenic signals in murine models. Aim 2. Evaluate the effects of treatment on the leukemia potential to design and test adaptive therapies in murine AML models. Aim 3. Construct a human AML transcriptome state-space to identify opportunities for adaptive treatment approaches. Impact. Through integration of experimental and clinical data, this work will accelerate the implementation of personalized therapy, inform future clinical trial designs, and improve outcomes of AML patients.