Novel Gene Expression Analyses to Accelerate Precision Pediatric Oncology Research

Cancer is the second leading cause of death in the United States. While there have been medical advances in treating cancer, the standard of care has not changed significantly in recent decades. Chemotherapy, radiation, and surgery are the clinician’s first line of defense against cancer progression, but new therapeutic strategies such as precision oncology are being developed that personalize cancer therapy to individuals. Precision oncology has primarily relied on coding mutations as biomarkers of response to therapies. Numerous challenges have arisen in the incorporation of transcriptome analysis into precision oncology workflows. One such challenge is in the necessary consideration of relative rather than absolute gene expression level, requiring differential expression analysis across samples. However, expression programs related to the cell-of-origin and tumor microenvironment effects confound the search for cancer-specific expression changes. To address these challenges, we developed an unsupervised clustering approach for discovering differential pathway expression within cancer cohorts using gene expression measurements. The hydra approach uses a Dirichlet process mixture model to automatically detect multimodally distributed genes and expression signatures. This led to the identification of recurrent tumor microenvironment signatures across pediatric cancers as well as a relationship between transposable element expression and immune infiltration. I then developed the vaccinaTE software toolkit to further characterize transposable elements as potential immunotherapy targets. Using RNA-seq and mass spectrometry analysis, I found expression and MHC-bound peptides uniquely mapping to transposable element loci. This led to the creation of a novel process for prioritizing TE vaccine targets as well as a microarray technology for personalizing TE vaccine therapy. To address the need for accurate preclinical models to accelerate drug development for pediatric cancers, I then created a Bayesian hierarchical modeling framework for evaluating patient-derived xenografts. I generated a database of PDX-specific pathway expression to facilitate validation studies that attempt to target differentially expressed pathways. This thesis has sought to improve the treatment of pediatric cancers through the identification of tumor subtypes that respond to specific therapies, identify novel immunotherapy targets based on tumor microenvironment states, and use gene expression analysis to optimize preclinical validation experiments. These methods have been developed for pediatric cancers but can be modified for adult cancers as well as other diseasesfor which gene expression data is available