The AML Salad Bowl

Abstract

Tumorsarise from single cellsbut become genetically heterogeneous through continuousacquisition of somaticmutations as they progress. In this issue of Cancer Cell, Klco and colleagues used whole genome sequenceanalysis to demonstrate the correlation of genetic clonal architecture with functional heterogeneity in acute myeloid leukemia.

A major tenet of cancer biology is that tumorsare clonal reflecting their origins from single cells. This is best illustrated by early seminal cytogenetic studies demonstrating that all cells of a patient's leukemia, for example, may harbour a specific chromosomal aberration. However, it has also been recognized for decades thatleukemias and solidtumors are heterogeneous in their genetic composition such that despite sharing a specific chromosomal aberration, not all cells of a given cancer demonstrate a completely identical cytogenetic profile. This intratumoralgenetic heterogeneity extends to the level of individual genes and DNA mutations as shown by next-generation sequencing technologies, and is fully expected based on the fact that tumor (and normal) cells acquirenew mutations with each cell division. At a practical level, somatically acquired mutations that accumulate at defined frequencies candistinguish individual cells or tumorsubclones,and serve as a clock to mark and track theirdivergence from a common ancestor cell. The complexity of clonal architecture has been shown inhematological malignancies including acute lymphoblastic leukemia (ALL) (Anderson et al., 2011) andacute myeloid leukemia (AML) (Ding et al., 2012),as well as other cancer types such as breast carcinoma (Shah et al., 2009), and is likely a universal feature of all cancers. It is also known that subpopulations of cells in an individual tumor can be morphologically or functionally distinct, e.g. display sensitivity or resistance to therapeutic agents. However, the relationship between intratumoral genetic heterogeneity and cancer cell function has not been well defined. Nevertheless, clonal evolution has major implications for understanding the cellular hierarchies and inter-relationships in tumors, as well as for the development and application of targeted therapies in the rapidly unfoldingera of personalized medicine. In this issue, Klco and colleaguesexplored the correlation of clonal architecture with functional heterogeneity in AML (Klco et al., Cancer Cell 2014). Rather than a melting pot blend of operationaland genomicdiversity, the data support that AML comprises a salad bowl of distinctsubclones whose functional differences may be genetically determined. Whole genome (andcapture-based targeted) sequences were analyzedto determine the somatic mutations present in unfractionated bone marrow cells of patients at presentation with de novo AML encompassing a range of morphological and genetic subtypes. The spectrum of mutations and their fractional representation was used to define the founding clone, from which all leukemic cells were descended, and also identified leukemic cell subpopulations possessing the “signature” variants of the founding clone as well as additional subclonalsequence variants that arose during tumorevolution. Sequence analysis of single cells purified by cell sorting in several AMLs verified the identity of subclonal genotypes and the allele fractions deduced from unfractionated bone marrow samples.

The genetically defined subclones were evaluated under various biological and experimental conditions. The clonal architecturepresent in the bone marrow wasconsistentlydetected in the peripheral blood, indicating no major differences in trafficking propertiesamong different AML subclones unlike the regional intratumoraland metastatic variation reported in solid tumors (Navin et al., 2011). Mutations found in AML blast cells wereoftenpresent in morphologically more mature myelomonocytic cells, demonstrating maintenance of at least minimaldifferentiation potential despite the presence of AML driver genes that otherwise antagonize maturation. Somatic mutations in rare peripheral blood B and T lymphocytes suggested the acquisition of some mutations in leukemic multi-potential hematopoietic stem-progenitor cells or even in pre-leukemic hematopoietic stem cellsconsistent with recent observations (Shlush et al., 2014). In some cases, morphologic or phenotypic features, as well as in vitro growth properties,correlated with distinct subclonessuggesting functional variation in differentiation potentialthat may be genetically determined.

The in vivo functional heterogeneityof cells comprising leukemia samples at disease presentation was interrogated by transplantation into immune-compromised mice. Unexpectedly, none of the resulting xenografts displayed a subclonal architecture that was identical to that of the transplanted AML. Rather, subclones showed variable engraftment potential, and single subclonesgenerally predominated in the engrafted micedespite the presence of multiple subclones in the injected sample. Relapsing AML subcloneswere not predicted by engraftment outcome or by the presence of recurring AML mutations. Thus, in many cases there was no apparent relationship between the engrafting cells and the evolutionary hierarchy of the leukemiasubclones in the patient. However, these resultsshould be interpreted with caution. Although the functional heterogeneity among AML clones was clearly demonstrated, the engrafted subclone in some cases was dictated by the recipient mouse strain used. This underscores that xeno-engraftment can be affected by a variety of technical factors that were not optimized such as mouse strain and preconditioning, route of injection, number of injected cells, and time for engraftment/disease assessment. The application of next-generation sequencing techniquesin future studies shouldilluminate the relativeinfluence of these variousfactors on the clonal compositions and clinical significance of engrafting leukemiacells.

AML cells capable of engrafting in xenograft assays, and thus establishing disease in mice, are operationally defined as leukemia-initiating cells or leukemia stem cells (LSCs) (Lapidot et al., 1994). Previous studies have shown that LSCs defined by this experimental approach may be phenotypically heterogeneous (Goardon et al., 2011). The studies of Klco et al. extend this to suggest that LSCs may also be genetically heterogeneous. Importantly, the foundingAML clone defined genetically may not necessarily be the same as the LSC clone defined functionally by xenotransplantation. This likely reflects thatxeno-transplant models exert selective growth pressuresthrough the mouse micro-environmentor lack of immune-system that are not equivalent to those encountered by leukemia cells in the patient. The authors' results highlight the clonal and functional diversity of LSCs, and suggest that future studies should include an integration of genetic and functional data as part of their characterization.

The authors correlated the presence of specific subclonesmarked by their respective mutational spectra with functional readouts, but the mutations that mechanistically account for the observed functional differences are unknown. It will be important to define the genetic (or epigenetic) determinants that underliethe observed biology, particularly the genes or pathways that may promote engraftment in various immune-compromised mouse models and relapse in patients.

Xenograft modelsalso serve an instrumental role in preclinical stages of cancer therapeutics development. Indeed, almost every FDA-approved anti-cancer drug in the modern era has been tested in such models. However, they are known to have limitations for predicting clinical responses in solid tumors(Sharpless and DePinho, 2006), and the results of Klco and colleagues underscore potential constraints of the approach to evaluate drug efficacy in AMLalthough previous studies have demonstrated their value in predicting early relapse in ALL (Meyer et al., 2011). Clonal analysis through whole genome sequencing provides a powerful approach for assessing the fidelity of xenografts in support of ongoing efforts to devise novel targeted therapeutics directed at the pathways and mutant factors that sustain critical functions offounding clones and/or cancer stem cells.

Functional and genetic heterogeneity of primary AML.

A founding AML clone arises with accumulation of a set of acquired “signature” mutations. During tumor progression, the founding clone evolves into genetically distinct subclones through the acquisition of new mutations. The complete spectrum of mutations defines morphological and phenotypic properties of the subclones that correlate with distinct functional characteristics such as in vitro growth, engraftment in immune-deficient mice, or relapse in patients.