Date of Award

Spring 5-15-2018

Author's School

Graduate School of Arts and Sciences

Author's Department

Biology & Biomedical Sciences (Molecular Genetics & Genomics)

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Cancer is a heterogeneous group of diseases that currently takes over half a million lives per year in the United States alone. Our understanding of cancer has improved dramatically over the last forty years, beginning with the discovery that cancer is a disease of the genome. Currently, the set of somatic mutations found in malignancy are largely known. The specific somatic mutations driving an individual’s disease can be readily assessed at clinical presentation. Additionally, the functional consequences for many of these mutations are known as well as their role in tumorigenesis. Despite this understanding, a cure for cancer remains elusive.

Acute myeloid leukemia (AML) is a particularly deadly example, which currently kills about 10,000 people per year and has a 5-year survival rate of only 25%. While the current outlook for these patients is grim, much is known about the disease, which will fuel future improvements in detection and therapy. Existing research has identified the spectrum of somatic mutations driving most cases of AML and has elucidated the oligoclonal nature of the disease. Following treatment, relapse often arises from a minor clone that was inconspicuous at presentation, but resistant to treatment. The current gold standard for assessing response to treatment is multiparameter flow cytometry (MPFC), which identifies persistent leukemic cells marked by a patient-specific leukemia-associated immunophenotype. Unfortunately, MPFC is only useful in a subset of patients and not sensitive to the clonal diversity present in many tumors. Conversely, virtually every case of AML is marked by leukemia-specific somatic mutations that theoretically distinguish every leukemic cell from its normal counterparts.

These limitations of MPFC and the general need for improved residual disease detection were early motivations for this thesis work: to develop a sequencing-based modality for rare leukemic-clone detection. Previous efforts to develop a sequencing-based platform for residual disease detection had largely failed because of the intrinsic error rate of next-generation sequencing (NGS) technology, which precludes the detection of leukemic clones less common than 1:20 cells (0.025 variant allele fraction for heterozygous mutations). For comparison, MPFC is sensitive and prognostic to a detection limit of 1:10,000 cells. To address this limitation, we developed methods for targeted error-corrected sequencing that mitigated the effect of sequencing errors. After an extensive development and validation process, we applied this technology to study two fundamental questions in AML and hematopoiesis in general.

First, we applied our error-corrected sequencing methods to study leukemogenesis in therapy-related AML (t-AML). This aggressive form of leukemia arises months to years following treatment with chemotherapy or radiation for a primary malignancy. The prevailing notion was that antecedent therapy introduced somatic mutations in hematopoietic stem and progenitor cells (HSPCs) that directly caused the development of t-AML. We used error-corrected sequencing to demonstrate that leukemogenic TP53 mutations were present at low frequency months to years before the diagnosis of t-AML and in some cases preceded the initial chemotherapy exposure. These findings redefined the etiology of t-AML. Instead of being introduced by chemotherapy, these TP53 mutations likely arose stochastically in HSPCs throughout the patient’s lifetime and were selected for by cytotoxic therapy, eventually spawning malignancy.

Second, we applied error-corrected sequencing to further our understanding of benign clonal hematopoiesis in healthy individuals over time. Recent work had identified benign hematopoietic clones harboring leukemia-specific somatic mutations in the blood of healthy individuals. The prevalence of this phenomenon increased as a function of age; while rare below 50, clones were detected in up to 10% of individuals by 70 years-old. These findings were made with conventional NGS and, likewise, did not detect rare clonal mutations in fewer than 1:20 cells. We sought to characterize the prevalence, stability and mutation spectrum of benign hematopoietic clones below this threshold. Using our error-corrected sequencing approach, we demonstrated that approximately 95% of disease-free individuals have hematopoietic clones harboring leukemia-associated mutations by 50-60 years of age. We also demonstrated that these clonal mutations were stable over time and originated in long-lived HSPCs.

These findings demonstrate the utility of our error-corrected sequencing platform to identify and characterize previously undetectable leukemia-associated somatic mutations. We applied these techniques to unveiled new insights into clonal HSPC biology and the development of t-AML. Future work will apply this technology as a sequencing-based modality for residual disease detection in pediatric AML. We believe this technology will improve the detection of residual leukemia, identify the step-by-step molecular perturbations driving relapse, inform therapeutic selection, and improve clinical outcomes and survival.


English (en)

Chair and Committee

Todd E. Druley

Committee Members

Donald F. Conrad, Timothy J. Ley, Daniel C. Link, Matthew J. Walter,


Permanent URL: