Cancer's ability to evolve in response to the pressures exerted by therapy is the single most common cause of therapeutic failure and poor clinical outcomes across most cancer types. We studied the evolution of chronic lymphocytic leukemia (CLL), which epitomizes the challenge posed to modern oncology by cancer evolution: despite effective therapies, the disease invariably recurs. We have measured intra-tumoral diversity, using genomic and epigenomic data, and have shown that evolution with therapy is virtually universal (Cell, 2013; Cancer Cell, 2014; Nature, 2015; Nature Communications, 2016). A major, emerging theme in oncology is that tumor evolution to resistance may also follow non-genetic routes. Specifically, cell persistence, lineage plasticity, and micro-environmental interactions are all non-genetically determined phenotypes, which are heritable and enable tumors to evolve and evade therapeutic attack. In CLL, we have shown that these mechanisms are highly active in resistance to ibrutinib therapy, a leading targeted agent (Nature Communications, in press). Unlike genetic resistance, we currently lack even a rudimentary framework on how to therapeutically target these mechanisms. Therefore,we aim to dissect how epigeneticand micro-environmental heterogeneity produce the cancer's ability to evolve and relapse. First,through the application of novel statistical inference to serial DNA methylation (DNAme) patient sample profiling, we will define positively selected DNAme changes ? epidrivers ? that result in cell persistence. We will validate candidate epidrivers through precision epigenetic editing, in order to causally link epidrivers with the persistence phenotype. We will integrate the DNAme profiles with transcriptional and clonal identity by applyingour innovative, multi-modality, single-cell platform. Second,histone code disruption and aberrant transcription factor (TF) expression have heritable evolutionary potential, and may be causes of lineage plasticity. To identify epidrivers of lineage plasticity, we will perform combinatorial histone mapping in serial samples from patients with Richter's transformation. We will complementthe patient sample studies with Screen-seq ? a method to link TF and histone modifiers with the abilityto evolve to high-grade lymphoma under therapeutic pressure. Third, our results show that cells with resistance mutations enable the growth of cells without the mutation through micro-environmental collaboration. To interrogate the co-evolution of CLL and its micro-environment, we will apply single-cell droplet RNAseq to serial lymph node samples from patients undergoing ibrutinib therapy. Our multi-modality, novel single-cell droplet platform will allow us to capture clonal identity, together withwhole transcriptome profiling, across CLL clones and neighboring, interacting immune cells. Collectively, this novel set of tools will map the epigenetic and environmental evolutionary potential of CLL and nominate targets for therapeutic intervention to overcome the central obstacle of tumor evolution.
Even as cancer treatments become more effective, the ability of malignant cells to adapt to therapy through an intensive evolutionary process is the leading obstacle to achieving a cure. Our proposed project seeks to tackle the non-genetic mechanisms driving cancer evolution, which remains a major, underexplored, but slowly emerging theme across the field of oncology. We will do so by developing experimental models and measurements in patient leukemia samples that will enable precision customization of anti-cancer therapy to prevent non-genetic evolution to treatment resistance.