The Hematologist

July-August 2019, Volume 16, Issue 4

Seeking Insight Into the AML Subpopulations and Microenvironment Through Single-cell RNA-Seq

Sameer S. Khatri, MD Hematopathology Fellow
Brigham and Women's Hospital, Boston, MA
Annette S. Kim, MD, PhD Associate Professor of Pathology
Harvard Medical School; Brigham and Women's Hospital, Boston, MA

Published on: June 03, 2019

van Galen P, Hovestadt V, Wadsworth li MH, et al. Single-cell RNA-seq reveals AML hierarchies relevant to disease progression and immunity. Cell. 2019;176:1265-1281.

The most recent World Health Organization and European Leukemia Network classification systems use cytogenetic and molecular findings to distribute cases of acute myeloid leukemia (AML) into favorable, intermediate, or adverse risk categories, which in turn influence treatment strategies for patients.1,2 Despite advances in treatment stratification, relapse after initial remission is common in all categories, and a subset of patients also remain refractory to induction chemotherapy. This has led to an interest in better understanding tumor biology to inform development of novel therapies, with a resultant resurgence of interest in transcriptomics. The somatic mutation literature has shown that heterogeneity within tumors plays an important role in recurrence and resistance.3 However, comprehensive molecular profiling in AML that includes transcriptomics has not yet been widely adopted despite the publication of several AML risk stratification models that incorporate combinations of somatic mutations, cytogenetics, and gene expression profiling.4 These prior studies used various methods of assessing gene expression profiles of entire AML samples without single-cell resolution.

In their recent article, Dr. Peter van Galen and colleagues used tandem single-cell transcriptomics and genotyping to assess bone marrow (BM) mononuclear cells from 16 patients with AML at diagnosis and during treatment, and BM aspirates from five healthy donors. Adapting Seq-well technology, the investigators performed high-throughput single-cell RNA sequencing (scRNA-seq) as well as single-cell genotyping by short-read and long-read sequencing to detect and phase AML mutations in individual cells. A machine learning classifier distinguished malignant cells from normal cells based on mutation status and classified cell types on the “hematopoietic stem cell (HSC) to myeloid differentiation” axis according to their similarity to normal BM cell types.

The analysis revealed six distinct malignant AML cell phenotypes within the AML cases. The proportion of these cell types varied across the AML cases studied as well as over the clinical course for each AML. The gene signature profiles of these six phenotypes were used to score bulk expression profiles of 179 diagnostic AML aspirates from the Cancer Genome Atlas (TCGA). The TCGA cohort could be clustered into seven different malignant tumor cell composition profile groups that, not surprisingly, demonstrated correlation with both characteristic genetic lesions (i.e., acute promyelocytic leukemia mapped to a high granulocyte-macrophage progenitor signature while CBFB-MYH11 cases mapped to high monocyte and dendritic cell-like clusters) and former French-American-British morphologic subtypes. Additionally, maturational dyssynchrony was found by transcriptional profiling that is reminiscent of the aberrant immunophenotypes of leukemic blasts and maturing cells seen typically by flow cytometry.

Interestingly, within the somatic mutational subtypes, the investigators found differential behavior between NPM1 subgroups — one with a strong HSC/progenitor phenotype in cases with co-occurring NPM1 and FLT3-ITD mutations and a second with a more differentiated monocyte- to dendritic cell–like signatures with ITD-negative cases. Even within the FLT3 mutated cases, different cellular hierarchies were identified. Genotyping of one case showed three subclones: “A” with a FLT3 p.A680V mutation, “B” with an additional FLT3-ITD mutation on the opposite allele, and “C” with a FLT3 p.N841K mutation only. Interestingly, most cells in subclones “A” and “B” expressed signature genes associated with progenitor-like cells, and the majority of subclone “C” cells expressed genes associated with differentiated monocyte-like or conventional dendritic-like cells. In these examples, the more progenitor-like phenotype typically was associated with an adverse prognosis.

Another area of recent investigative interest has been the study of antitumor immune responses within AML. Although therapeutic trials of immunomodulating treatments in AML have been limited to date,5,6 studies have indicated a role of T cell dysregulation in impairing antileukemic immune responses.7 Dr. Van Galen and colleagues also looked at the T cell signatures in the AML and control samples. They found that AML BMs contained fewer T cells and cytotoxic T lymphocytes but relatively more T-regulatory cells (also confirmed by immunohistochemical analysis of CD25 and FOXP3). Additionally, the investigators performed in vitro assays of T cell activation that showed a strong inhibitory effect associated with the presence of leukemic cells from an acute myelomonocytic leukemia cell line, MUTZ-3. Cultures enriched with monocyte-like (CD14+) MUTZ-3 cells were also particularly suppressive of T cell activation, decreasing T cell activation by 10-fold (p<0.0001). Review of the transcriptome data from the AML cases in this series showed that monocyte-like AML cells preferentially expressed certain genes important in immune modulation, supporting the notion that these cell types may be driving the immune-suppressive biology of AMLs.

In summary, this study provides comparative transcriptomic and genomic assessment of single cells, confirming the differential transcriptome between AML and control marrows, between different cytogenetic and morphologic subtypes of AML, and between different somatic mutational subtypes of AML. Taking advantage of the single-cell phenotyping capability of this method, the authors also examined the microenvironment of AML, demonstrating the role of monocyte-like cells in inhibiting immune response to AML and different T cell subset distributions in AML, paving the way for future studies observing immunosuppression in AML. Despite these findings and several other gene expression profiling AML stratification models,4 transcriptomics has yet to find a place in routine clinical practice. Additional work is needed to parse out the clinical interplay between more traditional prognostic markers and the adverse prognosis associated with progenitor-like signatures in clonal and subclonal populations, the effect of immunosuppressive monocytes, and the T cell subset milieu. Additionally, by using the presence of mutations, including genes such as DNMT3A and TET2, to distinguish the malignant clonal cells from the background normal cells, the authors did not discriminate between clonal hematopoiesis and leukemic hematopoiesis. These complexities will require further elucidation prior to clinical adoption. However, it is clear that the use of single-cell profiling provides the added granularity needed to take our understanding of AML biology to the next level.

References

  1. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision of the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127:2391-2405.
  2. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129:424-447.
  3. Welch JS, Ley TJ, Link DC, et al. The origin and evolution of mutations in acute myeloid leukemia. Cell. 2012;150:264-278.
  4. Wang M, Lindberg J, Klevebring D, et al. Validation of risk stratification models in acute myeloid leukemia using sequencing-based molecular profiling. Leukemia. 2017;31:2029-2036.
  5. Davids MS, Kim HT, Bachireddy P, et al. Ipilimumab for patients with relapse after allogeneic transplantation. N Engl J Med. 2016;375:143-153.
  6. Daver NG, Basu S, Garcia-Manero G, et al. Phase IB/II study of nivolumab with azacytidine (AZA) in patients (pts) with relapsed AML. J Clin Oncol. 2017; doi: 10.1200/JCO.2017.35.15_suppl.7026.
  7. Williams P, Basu S, Garcia-Manero G, et al. The distribution of T-cell subsets and the expression of immune checkpoint receptors and ligands in patients with newly diagnosed and relapsed acute myeloid leukemia. Cancer. 2019;125:1470-1481.

Conflict of Interests

Dr. Khatri and Dr. Kim indicated no relevant conflicts of interest. back to top