Tracing the Roots of Cancer Evolution

What is the cell of origin of tumors? This fundamental question has challenged cancer research for decades. Almost 30 years ago, in a review published in Science, Peter C. Nowell (1) proposed that tumors originate from a single cell and that “an induced change provides it with a selective growth advantage.” Subsequently, “acquired genetic lability permits stepwise selection of variant sublines and underlies tumor progression.” This groundbreaking theory, now known as the cancer clonal evolution model, was partly based on the observation he made with David Hungerford that in patients with chronic myelogenous leukemia (CML), essentially all tumor cells carried a chromosomal translocation (encoding BCR–ABL) known as the Philadelphia chromosome (1). Cancer as a clonal disease is now a widely accepted concept. However, the cell of origin of many cancers is still largely unknown, and the exact initial incident that triggers the cascade of events leading to malignant transformation is yet to be identified.

Hematologic malignancies provide great models to study the cell-of-origin question, largely owing to our relatively comprehensive knowledge of the hierarchical differentiation path from hematopoietic stem cell (HSC) to various lineages of blood cells where these malignancies arise. In addition, new technologies, such as multicolor flow cytometry and xenotransplantation in immune-deficient mouse strains, have helped hematologists pinpoint the rare cell populations that are responsible for the propagation of disease. For example, it is now evident that leukemia stem cells (LSC) exist in acute myelogenous leukemia (AML), and that these cells are capable of initiating and sustaining leukemic clones in vivo (2, 3). However, several key questions remain. First, the exact cell of origin of these LSCs is still up for debate. Two hypotheses have emerged. One proposes that LSCs originate from normal HSCs, given the phenotypic (cell-surface markers) and functional (self-renewal) similarities between these two cell populations. The other suggests that the initial transformation event occurs at lineage-restricted committed progenitor cells because of the phenotypic heterogeneity among patients with AML. Evidence from animal experiments has supported both models. For example, Huntly and colleagues (4) demonstrated that expression of the oncogenic AML fusion protein MOZ–TIF2 in common myeloid progenitors and granulocyte–monocyte progenitors (GMP) can recapitulate AML in mice. In contrast, expressing the CML oncogenic translocation product BCR–ABL in the same progenitor cells failed to induce myeloproliferative disease (4). Therefore, it seems that in CML, normal HSC is the disease cell of origin, whereas in AML, committed progenitor cells can regain self-renewal ability through oncogenic events and become LSCs. Another remaining question is whether the cell of origin is determined by the initial oncogenic event. The study by Huntly and colleagues (4) suggested that different oncogenic events have various degrees of transformation capabilities, and such differences may explain the different cells of origin in different leukemias. In a mouse model of MLL–AF9 leukemia, the dosage of MLL–AF9 expression could affect the transformation susceptibility of different cell types. Progenitor cells, such as GMPs, could be transformed only when a very high dose of MLL–AF9 was expressed (5). Hence, understanding the initiating event is as important as the identification of cancer's cell of origin. A third unresolved question is: What is the sequence of events that leads to the clonal evolution from a normal HSC/progenitor to a preleukemic cell and eventually leukemia? According to the Nowell model, LSCs would acquire additional genetic abnormalities in a stepwise fashion that facilitate the progress of disease development and subsequent relapses. Indeed, whole-exome sequencing (WES) and whole-genome sequencing of more than 200 adult de novo AMLs have discovered that, on average, each AML genome carries 13 mutations located within genes (6), among which only five are recurrently mutated in AML. However, sequencing studies on bulk tumors can only infer clonal evolution based on the mutation allele frequency and could not provide information such as at which exact stage along the LSC differentiation these mutations are acquired. Moreover, biologic experiments are needed to differentiate driver mutations from passenger mutations that have no contribution to leukemogenesis. Without such knowledge, it is difficult to develop targeted therapies to eradicate LSCs.

In this issue of Cancer Discovery, Damm and colleagues (7) sought to address these important questions in the context of chronic lymphocytic leukemia (CLL). CLL is the most common adult leukemia in the Western world. This disease is characterized by the clonal expansion of CD5⁺CD23⁺ B cells in blood, bone marrow, and secondary lymphoid tissues. Roughly half of the patients with CLL carry mutations within the immunoglobulin heavy-chain variable-region genes (IGHV), which is frequently associated with an indolent disease. Patients with unmutated IGHVs often present with more aggressive disease. Previously, xenotransplantation studies have demonstrated that HSCs isolated from patients with CLL were primed toward lymphoid lineage and were prone to develop features of CLL, suggesting that normal HSCs may be the cell of origin of CLL (8). In this study, the authors used cell-surface markers to isolate populations of immature progenitor cells (CD34⁺) and mature T cells (CD3⁺), monocytes (CD14⁺), and normal and tumor B cells (CD19⁺) from 24 patients with CLL, and then surveyed the mutational landscapes in these cell populations by WES. Strikingly, they observed that in 21 of 24 patients, a subset of the CLL mutations seen in the CD19⁺ B cells could be detected in either or both the immature CD34⁺ cells and the CD14⁺ monocytes in the myeloid lineage from the same patient, suggesting that CLL pathogenesis involves immature progenitor cells (7). Moreover, the authors sorted CD34⁺CD19⁻ progenitor cells from 18 patient samples and cultured them in myeloid conditions. These cells yielded myeloid colonies, of which colonies from 13 patients had detectable CLL mutations, demonstrating that these CLL mutation–carrying immature progenitor cells still had multilineage potential. Mutations detected in the progenitors of CLL patients targeted several well-known CLL oncogenes, including NOTCH1, SF3B1, TP53, XPO1, BRAF, and MLL2. In addition, Damm and colleagues (7) also found mutations in NFKBIE and EGR2 in the progenitor cells of several patients with CLL. Of note is that those patients with EGR2 mutations were associated with higher CD38⁺ expression (a poor prognostic marker), shorter time to re-treatment, and low 5-year overall survival, further suggesting that mutations in EGR2 may be a potential initiating event in progenitor cells, resulting in preleukemic cells that would eventually develop into CLL. Indeed, when these authors expressed EGR2 mutants in a murine multipotent hematopoietic cell line, EML, they observed that these cells exhibited slower growth and faster loss of the B220⁺ and Gr1⁺ population than EML cells transduced with wild-type EGR2. How these changes affect the development of CLL requires further investigation, potentially using a xenotransplantation model. Finally, using RNAseq, they identified a gene expression signature in CLL patients with EGR2 mutations. In particular, the upregulated signature of EGR2-mutated CLL samples was enriched of genes induced by B-cell receptor (BCR) stimulation in normal B cells, suggesting a deregulation of BCR signaling in these patients (7).

Taken together, results from this study confirm that the cell of origin of CLL is among the hematopoietic progenitor cells that still have the potential to differentiate into different lineages of mature blood cells, and identify early mutation events in these progenitor cells that may lead to the subsequent transformation of CLL. Recently, similar studies also uncovered preleukemic HSCs of AML and the founding mutations associated with these cells, including ASXL1, DNMT3A, IDH2, and IKZF1 (9, 10). In addition, Shlush and colleagues (10) elegantly showed that the initiating DNMT3A mutation resulted in an expanded pool of HSCs and downstream progenitors, within which additional mutations such as NPM1c were acquired, probably at the GMP and/or multilymphoid progenitor stages, to drive progression to AML. Collectively, these studies raise several important issues in the treatment of leukemias. First, in addition to the eradication of leukemia blasts by conventional chemotherapy, new effective therapies are needed to selectively kill preleukemic HSCs and progenitor cells that are the real source of leukemias. Although this remains a major clinical challenge, the newly discovered features of these rare cell populations, that is, upregulated BCR signaling gene expression in EGR2-mutated CLLs and expanded HSC pools in DNMT3A-mutated AMLs, provide opportunities for the development of innovative targeted therapy in the near future. Second, the detection of minimal residual disease after initial treatment and the monitoring of relapse disease development need to be redesigned to trace these preleukemic cells that often escape and survive chemotherapy. Multicolor flow sorting and ultrasensitive digital PCR may need to be adopted in the clinical setting to detect rare preleukemic cells in peripheral blood samples routinely. Third, in this personalized precision-medicine era, for those patients with leukemia without a known oncodriver mutation, it may be beneficial to include xenotransplantation animal models to pinpoint the disease-driving mutation in the preleukemic stem cells.

Looking back to the Nowell model, it is amazing how accurate his prediction of the clonal evolution course of cancer was. Thirty years later, we finally have the right tools and assays to fully address the question of cancer's cell of origin. It is foreseeable that an explosion of research on this topic will occur in the next several years and significantly advance our understanding of cancer and improve how we treat this devastating disease.

No potential conflicts of interest were disclosed.