Andrew Josef Ewald - US grants

PROJECT SUMMARY The overwhelming majority of deaths from cancer are attributable to metastasis, rather than growth of the primary tumor. In breast cancer, metastatic recurrence can occur years to decades after apparently successful surgery. Current methods do not allow individualized assessment of metastatic recurrence risk nor do they offer effective therapies for metastatic breast cancer patients. Breast cancer presents a unique research opportunity because the long interval between surgery and recurrence offers the potential to improve patient outcomes if effective anti-metastatic therapies could be developed. However, few drug discovery efforts to date have focused on the metastatic process specifically. The challenges we address are developing and applying methods to identify the basic mechanisms of metastasis, then prioritizing and validating genes and proteins as potential therapeutic targets. Our approach combines advances in experimental (Ewald) and computational (Bader) methods that we have developed to interrogate the metastatic process and to systematically dissect the genetic basis of human disease. Experimentally, we will use a pipeline that relies on organoids from primary human breast cancer tissue to model several distinct steps of metastasis: invasion into the surrounding matrix, dissemination of cancer cell clusters, and outgrowth of these clusters molecular models of distant organs. Computationally, we have developed and applied powerful methods to connect quantitative traits to their genetic basis across multiple complex human disease. We will now apply these computational methods to dissect the molecular basis of breast cancer metastasis. The central insight of our proposal is that the known heterogeneity of breast tumors, while confounding to other methods, enables our quantitative trait loci approach. We will exploit this heterogeneity with computational methods that have the potential to identify the molecular differences between primary human breast tumor organoids that demonstrate metastatic vs. non-metastatic cell behaviors (Aim 1). We will use network analysis techniques to prioritize these as targets, and then use a combination of mammalian genetic engineering and small molecule perturbations to validate targets first in the organoid system and then in accepted mouse PDX models for metastatic growth (Aim 2). Finally, we will combine our novel target based approaches with chemical and genetic perturbagens from the CTD2 Network and broader drug discovery efforts (Aim 3). In this way, we can build on existing knowledge to accelerate our progress towards improved patient outcomes. Success of this program will provide clinically actionable targets for preventing metastatic recurrence or treating patients with established breast cancer metastases. Importantly, our approaches can provide a general platform for dissecting metastasis across epithelial cancers.

Project Summary Liver cancer is a major global health problem, responsible for the 3rd most cancer deaths worldwide. Diagnosis often occurs at late stages, at which point liver tumors have complex tumor/stroma interactions across multiple spatial and temporal scales. The resulting multiscale interactions drive tumor progression and therapeutic response. The proposed project will develop new mathematical/computational techniques to model molecular, cellular, tumor, and organ scales to elucidate the mechanisms driving liver cancer progression and to predict the response to targeted therapeutics. The investigator team is uniquely suited to develop the proposed multiscale models of hepatocellular carcinoma (HCC), the most common type of liver cancer. The expertise of the four PIs/PDs is synergistic, combining a state of the art multiscale computational models of cancer (Dr. Popel) with molecular and cellular features inferred from bioinformatics analysis (Dr. Fertig) using state of the art 3D in vitro organoid models (Dr. Ewald) and in vivo mouse models of HCC (Dr. Tran). The well-integrated experimental/computational design of the proposal will result in new algorithms for predictive computational modeling of therapeutic response in HCC. We include extensive experimental studies for model development, parameter tuning, and validation. Specific Aim 1 will infer bioinformatically the signaling pathways important in crosstalk between cancer and stromal cells, integrate models of intracellular signaling and 3D extracellular ligand transport and biochemical reactions and embed them into the cell fate decision rules of an agent-based model of cellular agents resulting in a multiscale hybrid model. The model will be parameterized with phospho- proteomic data under relevant ligand stimulations identified by the bioinformatics analysis and with growth, invasion, proteomic, and genomic data from co-cultured cancer and stromal cells and organoids; independent data will be used for model validation. We will use this model to predict outcomes in a 3D in vitro organoid model of HCC. Specific Aim 2 will extend and adapt this hybrid model to model the tumor microenvironment and to account for the drug pharmacokinetic and pharmacodynamic, the 3D geometry of the liver, molecular interactions in vivo and cellular composition inferred from bioinformatics analysis. Finally, Specific Aim 3 will develop new bioinformatics analysis algorithms to initialize the model with distribution of cellular agents and molecular states from The Cancer Genome Atlas (TCGA) genomic and proteomic data to predict the efficacy of targeted therapeutics in the diverse genetic backgrounds of human liver cancer. The project will develop innovative computational techniques to integrate features at both the molecular and cellular scales from genomics and proteomics analysis with multiscale computational models to predict therapeutic response. The resulting computational algorithms will address the IMAG cutting edge challenge of fusing data-rich and data- poor scales for predictive multiscale computational modeling of biological systems.