Evolution Of Acute Myelogenous Leukemia Stem Cell Properties Following Treatment and Progression

Acute Myelogenous Leukemia (AML) is a disease that clinically evolves over time as many patients who are responsive to therapy upfront acquire resistance to the same agents when applied in the relapse setting. The stem cell model for AML has been invoked to explain primary resistance to standard therapy; the leukemia stem cell (LSC) population representing a therapy-refractory reservoir for relapse. There have been no prospective efforts to formally assess the evolution of the LSC population during patients’ clinical course. We performed a prospective characterization of specimens from a well-defined cohort of patients with AML at diagnosis and relapse to assess the frequency and phenotype of functionally defined LSCs. Methods Primary bone marrow and peripheral blood samples were collected on IRB approved protocols from patients with newly diagnosed AML undergoing induction therapy. Twenty-five patients who relapsed after achieving a complete remission were selected for further study. Screening studies identified seven patients whose pre-therapy samples demonstrated sustained engraftment of NSG mice following transplantation. Pre-therapy and post-relapse LSC frequencies were assessed using xenotransplantation limiting dilution analyses (LDA). We assessed the frequencies of CD45RA, CD32, TIM-3, CD96, CD47, and CD97 expressing populations that have been previously published to possess LSC activity. Functionally validated pre-therapy and post-relapse LSC populations were identified using fluorescent labeled cell sorting and NSG xenotransplantation. LSC activity was confirmed for each population using secondary xenotransplantation. Gene expression analysis of highly enriched LSC populations from pre-therapy and post-relapse samples was performed using ABI TILDA qPCR analyses following pre-amplification. Results We demonstrated by LDA an 8 to 42-fold increase in LSC frequency between diagnosis and relapse in paired primary patient samples. The increase in LSC activity was not associated with an increase in frequency for phenotypically-defined populations previously reported to possess LSC activity. Rather, we found that LSC activity expanded at relapse to immunophenotypic populations of leukemic cells that did not possess LSC activity prior to treatment. Moreover, in all patients, the number of phenotypically distinct LSC populations (as defined by CD34 and CD38 or CD32 and CD38) detectable at relapse was dramatically expanded. Further, while the majority of the LSC populations’ gene expression profile remained stable between diagnosis and relapse, a subset of genes were enriched in defined LSC populations at relapse including IL3-receptor alpha and IL1-RAP, both previously demonstrated to play a role in LSC biology. Conclusions This study is the first to characterize the natural evolution of LSCs in vivo following treatment and relapse. We demonstrate an increase in LSC activity and greatly increased phenotypic diversity of the LSC population, suggesting a loss of hierarchical organization following relapse. These findings demonstrate that treatment of AML patients with conventional chemotherapy regimens can promote quantitative and qualitative expansion of the LSC compartment. Further, the data indicate that surface antigen immune-phenotype is not predictive of function in relapse and suggest a major limitation to efforts targeting specific surface antigens in the relapse setting. Understanding the mechanisms by which LSC expansion occurs and how to target it will likely improve our currently poor treatment options for patients who relapse. Disclosures: Becker: Millenium: Research Funding.