Single-cell Multiomics Analysis of Myelodysplastic Syndromes and Clinical Response to Hypomethylating Therapy

Abstract

Alterations in epigenetic marks, such as DNA methylation, represent a hallmark of cancer that has been successfully exploited for therapy in myeloid malignancies. Hypomethylating agents (HMA), such as azacitidine, have become standard-of-care therapy to treat myelodysplastic syndromes (MDS), myeloid neoplasms that can evolve into acute myeloid leukemia. However, our capacity to identify who will respond to HMAs, and the duration of response, remains limited. To shed light on this question, we have leveraged the unprecedented analytic power of single-cell technologies to simultaneously map the genome and immunoproteome of MDS samples throughout clinical evolution. We were able to chart the architecture and evolution of molecular clones in precious paired bone marrow MDS samples at diagnosis and posttreatment to show that a combined imbalance of specific cell lineages with diverse mutational profiles is associated with the clinical response of patients with MDS to hypomethylating therapy.

Significance: MDS are myeloid clonal hemopathies with a low 5-year survival rate, and approximately half of the cases do not respond to standard HMA therapy. Our innovative single-cell multiomics approach offers valuable biological insights and potential biomarkers associated with the demethylating agent efficacy. It also identifies vulnerabilities that can be targeted using personalized combinations of small drugs and antibodies.

FIGURE 1

Mutational landscape of patients with the studied MDS. A, Summary of the study workflow. B, Oncoprint indicating the mutations present in the patient cohort, at diagnosis and after AZA treatment, colored by coding impact. Multi-hit means presence of more than one mutation of same coding impact (missense, nonsense, or frameshift) in the same gene. C, UpSet plot illustrating the exclusive intersection of mutated pathways at diagnosis in patients with MDS. The patients are colored according to the response status (green, responders; magenta, nonresponders).

FIGURE 2

Distribution of mutant clones and CNVs at diagnosis in patients with the studied MDS. A, Proportion of mutant clones over all mutant cells at diagnosis in each patient, colored by clone abundance. Upper bar plot illustrates number of clones at diagnosis and the middle bar indicates the response status (green, responders; magenta, nonresponders). Note that patient #10 is not included because no mutant clones were found using our gene panel. B, Number of mutant clones at diagnosis (left); Shannon diversity index computed among mutant clones at diagnosis (middle); predominant mutant clone size with respect to other mutant clones at diagnosis (right). R, responders; NR, nonresponders. C, Percentage of patients with mutations in each gene in the predominant clone at diagnosis (left); Percentage of patients with mutations in each gene in the first clone at diagnosis (right). D, Proportion of responders and nonresponder patients with CHIP mutations (TET2, DNMT3A, and ASXL1) in first (left) and predominant (right) clones at diagnosis. E, Oncoprint of CNVs in the patient cohort at diagnosis. F, Median per amplicon ploidy of the mutant clones for patients with CNVs (patients #2, #3, #5, #8, #9, and #13).

FIGURE 3

Clonal evolution of responder and nonresponder patients with MDS upon AZA treatment. A, Clonal phylogenies of patients #3 and #14 (responders) and #4 and #8 (nonresponders) at diagnosis. Dot size represents clone size. B, Fishplots of patients #3 and #14 (responders) and #4 and #8 (nonresponders), illustrating the clonal distribution at diagnosis and after AZA treatment. C, Proportion of mutant cells (left) and predominant mutant clone (right) in responder patients at diagnosis and after AZA treatment. Dx, diagnosis; Aza, after AZA treatment.

FIGURE 4

Characterization of BM cell populations of patients with MDS based on scProt-seq data. A, UMAP of BM cells from all MDS samples colored by abundance of the depicted proteins. B, UMAP of BM cells from all MDS samples colored by cell type annotation. C, Cell type proportions at diagnosis and after AZA treatment in responders. *, scCODA FDR <0.1. D, UMAPs at diagnosis and after AZA for all the responder patients. Dx, diagnosis; Aza, after AZA treatment.

FIGURE 5

Distribution of mutant cells within the BM compartments defined by the immunophenotype. A, Percentage of mutant progenitor, immature erythroid and myeloid cells (Pro_Ery_Mye) compared with mutant lymphoid cells in each patient at diagnosis (top); percentage of mutant progenitor, immature erythroid and myeloid (bottom, left); or T, B, and NK cells (bottom, right) in each patient at diagnosis. B, UMAP of BM cells from all samples colored by number of mutations per cell (left); cell type proportions according to the number of mutations per cell (right). C, UMAP colored according to gene mutational status. WT, wild-type; MUT, mutant.

FIGURE 6

Impact of AZA treatment on mutant cell populations in responder patients. A, Mutant cell type proportions at diagnosis and after AZA treatment in responders. *, scCODA FDR <0.1. B, UMAPs at diagnosis and after AZA in all the HMA responder patients, colored by number of mutations per cell, highlight the three mutant populations significantly associated with response. C, UMAPs highlighting mutant HSPCs in patient #2 (responder), mutant immature granulocytes in patient #9 (responder) and mutant immature monocytes in patient #7 (responder), colored by clone. Dx, diagnosis; Aza, after AZA treatment.