Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Apr 22:13:e93019.
doi: 10.7554/eLife.93019.

Convergent epigenetic evolution drives relapse in acute myeloid leukemia

Kevin Nuno #  1   2   3   4 , Armon Azizi #  2   3   4   5 , Thomas Koehnke  2   3   4 , Caleb Lareau  6   7 , Asiri Ediriwickrema  1   2   3   4 , M Ryan Corces  1   2   3   4   8   9   10 , Ansuman T Satpathy  6   7   11   12 , Ravindra Majeti  2   3   4
Affiliations

Convergent epigenetic evolution drives relapse in acute myeloid leukemia

Kevin Nuno et al. Elife. .

Abstract

Relapse of acute myeloid leukemia (AML) is highly aggressive and often treatment refractory. We analyzed previously published AML relapse cohorts and found that 40% of relapses occur without changes in driver mutations, suggesting that non-genetic mechanisms drive relapse in a large proportion of cases. We therefore characterized epigenetic patterns of AML relapse using 26 matched diagnosis-relapse samples with ATAC-seq. This analysis identified a relapse-specific chromatin accessibility signature for mutationally stable AML, suggesting that AML undergoes epigenetic evolution at relapse independent of mutational changes. Analysis of leukemia stem cell (LSC) chromatin changes at relapse indicated that this leukemic compartment underwent significantly less epigenetic evolution than non-LSCs, while epigenetic changes in non-LSCs reflected overall evolution of the bulk leukemia. Finally, we used single-cell ATAC-seq paired with mitochondrial sequencing (mtscATAC) to map clones from diagnosis into relapse along with their epigenetic features. We found that distinct mitochondrially-defined clones exhibit more similar chromatin accessibility at relapse relative to diagnosis, demonstrating convergent epigenetic evolution in relapsed AML. These results demonstrate that epigenetic evolution is a feature of relapsed AML and that convergent epigenetic evolution can occur following treatment with induction chemotherapy.

Keywords: acute myeloid leukemia; cancer biology; epigenetics; genetics; genomics; human; relapse.

Plain language summary

Acute myeloid leukemia (or AML for short) is a type of blood cancer characterized by abnormally high production of immature white blood cells. Despite advances in AML treatment, many patients relapse after an initially successful first round of treatment. As a result, understanding the factors contributing to relapse is essential for developing effective treatments for the disease. Like most cancers, AML can evolve because of changes to the DNA sequence in cells that cause them to grow uncontrollably or resist treatment. Alongside these genetic mutations, AML cells also undergo ‘epigenetic’ changes, where regions of the DNA are modified and genes can be switched on or off without altering the DNA sequence. Previous research has demonstrated that epigenetic changes contribute to the development of AML, however, it was not clear if these changes could also make cells resistant to treatment without acquiring new DNA mutations. Nuno, Azizi et al. addressed this question by analyzing the epigenetic states of AML cells from 26 patients at the time of their diagnosis and after treatment when the disease had relapsed. Analysis revealed that almost half of the patients with AML experienced a relapse without acquiring new DNA mutations. Instead, these AML cells developed specific epigenetic changes that helped them to resist cancer treatment. Moreover, studying individual AML cells from different patients showed that the cells became more epigenetically similar at relapse, suggesting that they converge towards a more treatment-resistant disease. Future experiments will determine exactly how these epigenetic changes lead to treatment resistance. Currently, most of the drugs used to treat AML are either chemotherapies or ones that target specific DNA mutations. The findings of Nuno, Azizi et al. suggest that drugs targeting specific epigenetic changes may be more effective for some patients. Further studies will be needed to determine which patients may benefit and which epigenetic drugs could be useful.

PubMed Disclaimer

Conflict of interest statement

KN, AA, TK, AE, MC No competing interests declared, CL consultant for Cartography Biosciences, AS founder of Immunai and Cartography Biosciences; receives research funding from Allogene Therapeutics and Merck Research Laboratories, RM on the Advisory Boards of Kodikaz Therapeutic Solutions, Orbital Therapeutics, and is an inventor on a number of patents related to CD47 cancer immunotherapy licensed to Gilead Sciences. Co-founder and equity holder of Pheast Therapeutics, MyeloGene, and Orbital Therapeutics

Figures

Figure 1.
Figure 1.. Meta-analysis of genomic change in relapsed AML.
(a) Oncoprint of recurrently mutated genes from published relapsed AML dataset. Genetic alterations are color coded according to dynamics between disease state: stable at relapse, gained at relapse, or lost at relapse. The number of gained, lost, and stable mutations as well as percentage of patients with mutations in each gene are displayed on the right and left sides of the plot respectively. (b) Ternary plot of recurrently mutated genes from dataset analysis depicting mutational dynamics between diagnosis and relapse (restricted to genes with at least 20 events in our cohort). The sides of the plot correspond to the percentage of mutations displaying the indicated dynamic (gained, lost, or stable at relapse), the size of each dot indicates the number of mutations detected for that gene. (c) Schematic of clonal ‘bins’ each patient was categorized into based on mutational data. Bar chart on the right depicts the proportion of patients assigned to each clonal bin. (d) Survival analysis of patients based on sample clonality data gathered for each patient group. ‘Stable’ clonality refers to patients with equivalent mutations detected at diagnosis and relapse, ‘Unstable’ refers to patients that had any change in mutations at relapse (log-rank test p<0.05).
Figure 1—figure supplement 1.
Figure 1—figure supplement 1.. Plots depicting change in variant allele frequency for detected ‘pre-leukemic’ epigenetic modifier mutations (a), RAS pathway mutations (b), or WT1 mutations (c) in meta-analysis relapsed AML cohort.
Lines connect samples from the same patient between disease timepoints, color coded according to mutation dynamic (red = stable, blue = gained at relapse, yellow = lost at relapse) (a, c) or according to RAS family member (b).
Figure 1—figure supplement 2.
Figure 1—figure supplement 2.. Gene-specific clonal analysis and multivariate analysis of relapse-free-survival based on clonality.
(a) Time to relapse analysis of meta-analysis cohort patients according to WT1 mutation status between diagnosis and relapse (blue = gained WT1 mutation at relapse, red = WT1 mutation stable at relapse). (b) Plot depicting recurrent mutations of meta-analysis cohort AML patients. Fraction of each bar graph depicts the proportion of mutants that fall in the designated clonal group colored according to key shown at right. (c) Multivariate Cox analysis of survival data from meta-analysis patient cohort data accounting for mutation clonality, age, and sex.
Figure 2.
Figure 2.. Chromatin accessibility change in stable and non-stable relapsed AML.
(a) Schematic of relapse sample acquisition and preparation. (b) Oncoprint of recurrently mutated genes in samples analyzed from Stanford University patient cohort (n=26). (c) Bar chart depicting fractions of patients from Stanford patient cohort in each clonal bin. (d). Violin plot depicting global chromatin accessibility similarity of diagnosis/relapse pairs based on ATAC-seq data. Each dot represents a value calculated for each patient based on overall chromatin accessibility similarity between diagnosis and relapse samples. (e) Volcano plot depicting differentially accessible genes at relapse in patients within the Stable clonal bin. (f) Heatmap depicting the accessibility of the top differential genes at diagnosis and relapse in mutationally stable samples Left: Accessibility displayed for stable samples. Right: accessibility displayed for non-stable samples. (g). Gene set enrichment analysis (GSEA) of differentially open or closed accessible genes calculated with ATAC-seq gene scores in mutationally stable samples. Top enriched or de-enriched pathways crossing significance threshold of padj <0.05 are shown.
Figure 2—figure supplement 1.
Figure 2—figure supplement 1.. Relapse-free-survival and differential chromatin analysis of the Stanford AML cohort.
(a) Time to relapse analysis of Stanford cohort patients according to clonality assessed by driver mutation genotyping. (b) Volcano plot of differential accessibility between diagnosis and relapse across all ATAC-seq peaks in genomically stable samples. Peaks considered significantly up or down-regulated if adjusted p-value < 0.05 and absolute log2 fold change >0.25. (c) Plot depicting fold change in accessibility of tiled 100 kb bins on chromosome 6 across stable samples at relapse. Dots indicate single bin fold change at relapse and red line shows the running fold change average across bins. Genes that were found to be significantly differential in accessibility are labeled at the bottom of the plot. (d) Heatmap of motif accessibility at diagnosis and relapse for genomically stable samples. Only the top 40 differentially accessible motifs are shown.
Figure 2—figure supplement 2.
Figure 2—figure supplement 2.. Projection of bulk AML ATAC samples to healthy hematopoietic single-cell manifold demonstrates shifts in epigenetic differentiation states at relapse.
(a) Healthy cell reference manifold projection of single-cell ATAC-seq data from Granja et al., 2019. Healthy hematopoietic cell categories color coded according to key shown below. (b) Bulk AML ATAC-seq data ‘pseudo’-single cell profiles projected to healthy hematopoietic reference umap. Projected cells are colored either by sorted cell population (left) or time of acquisition (right). (c) Projection of bulk AMLs from two patients using pseudo-single cell approach. Diagnosis samples shown in blue, relapse samples shown in red. (d) Plot depicting the change in projection between diagnosis and relapse across all samples. Arrows connect the average projection location of diagnosis cells to the average projection location of relapse cells for each patient. (e) Distribution of the closest mapped cell type across all AML Blast (non-LSC enriched) samples. Diagnosis and relapse distributions were compared using a fisher’s exact test.
Figure 3.
Figure 3.. Chromatin change in relapsed AML LSCs.
(a) Cell sorting and library preparation scheme for comparison of chromatin accessibility profiles of LSC vs. non-LSC enriched subpopulations in relapsed AML cohort. (b) Box and whisker plot depicting chromatin accessibility correlation coefficient for non-LSCs vs. LSCs. Lines connect LSC similarity and non-LSC similarity values for each patient. (c) Volcano plot of differentially accessible peaks between diagnosis and relapse in non-LSCs (left) or in LSCs (right). The top 200 differential peaks by log2FC in LSCs (dark purple) and non-LSCs (light purple) are highlighted in each plot. Cutoffs for adjusted p value of<0.05 are depicted as dashed lines in both plots. (d) Bar plot comparing fraction of significantly differentially accessible peaks in non-LSCs that were also differentially accessible in LSCs and vice versa. (e) Gene set enrichment analysis (GSEA) of differentially open or closed accessible genes calculated using gene scores from ATAC-seq data from LSC samples. A positive NES indicates that the pathway was upregulated at relapse relative to diagnosis based on the gene accessibility data. Top pathways crossing significance threshold of padj <0.05 are shown. (f) GSEA plot of exemplar gene pathway enriched in bulk Stable ATAC-seq samples (HALLMARK_MYC_TARGETS_V2) applied to either non-LSC gene score data set (left) or LSC gene score data set (right).
Figure 3—figure supplement 1.
Figure 3—figure supplement 1.. Comparison between AML relapse signature and LSC signature.
Relapse signatures were derived by comparing relapse timepoints to diagnosis timepoints in stable AML patients. LSC signatures were derived by comparing LSCs to non-LSC populations at diagnosis across patients with sortable LSC populations. Relapse chromatin changes were compared to LSC chromatin changes using gene accessibility scores. (A) Heatmap of the most differential gene accessibility scores in relapse samples at diagnosis and relapse timepoints (left), and LSC vs Non-LSC populations (right). (B) Dot-plot depicting the correlation between relapse signature gene accessibility score fold changes and LSC signature gene accessibility score fold changes.
Figure 4.
Figure 4.. Single-cell ATACseq analysis of relapsed AML.
(a) Scheme for single-cell ATAC-seq sample preparation and cell clustering analysis (b) UMAP projection of all single-cell ATAC-seq patient samples from Stanford diagnosis/relapse cohort (samples color-coded according to scheme at right). (c–d) Cell clustering analysis of Stanford patient samples SU360 (c) and SU892 (d). At left, UMAP plots of individual samples from each patient with color-coded cell clusters shown; at right, non-hierarchical clustering heatmap of determined cell clusters. (e) UMAP projections of single-cell ATAC-seq from patient samples showing heat signature for Stable sample relapse ATAC-seq signature (from Figure 2). Inset UMAP shows cells colored according to diagnosis or relapse samples. Overall relapse score for each sample is quantified in plots at the bottom.
Figure 4—figure supplement 1.
Figure 4—figure supplement 1.. Single-cell SNN clusters and between-cluster epigenetic similarities.
(a) UMAP plots of clustering analyses performed for the four patient samples indicated (diagnosis sample at left, relapse at right). (b) Hierarchical clustering analysis of scATAC-seq data from diagnosis and relapse cell clusters from patients SU142 (left) and SU484 (right).
Figure 4—figure supplement 2.
Figure 4—figure supplement 2.. Projection of AML scATAC samples to healthy hematopoietic single-cell manifold demonstrates shifts in epigenetic differentiation states at relapse.
(a) Schematic of strategy for healthy cell LSI projection analysis of single-cell ATAC-seq diagnosis/relapse data (b) Healthy cell reference of single-cell ATAC-seq data from Granja et al Nat. Granja et al., 2019. Healthy hematopoietic cell categories color coded according to key shown at bottom. (c–f) UMAP projections of single-cell ATAC-seq samples onto healthy cell reference manifolds for indicated patients. Barplot of nearest cell type classification fractions across all cells at diagnosis and relapse is shown for each patient.
Figure 5.
Figure 5.. Mitochondrial-based clonal tracing paired with single-cell chromatin accessibility in stable AMLs.
(a) Scheme for sample processing and analysis of mitochondrial single-cell ATAC-seq (mtscATAC-seq). (b) Heatmap of mitochondrial variant heteroplasmy values across all single cells for each genomically stable sample. Only variants passing filtering criteria are shown. Cells are ordered based on the cluster they are assigned to. (c) Fishplots of inferred mitochondrial clone evolution at diagnosis and relapse based on mitochondrial cluster frequencies in (B). (d) Heatmap of epigenetic similarity across all major mitoclones for each sample (left). Boxplot of inter-clone epigenetic similarity at diagnosis and relapse (means compared using unpaired student’s t-test) (right).
Figure 5—figure supplement 1.
Figure 5—figure supplement 1.. Mitochondrial-based clonal tracing paired with single -cell chromatin accessibility in a genetically non-stable AML.
(a) Comparison of relapse vs diagnosis accessibility fold change across all peaks between mitoclone 1 and mitoclone 2 in SU360 (R=0.41, p<2.2e-16). (b) Heatmap of mitochondrial variant heteroplasmy values across all single cells for SU484. Only variants passing filtering criteria are shown. Cells are ordered based on the cluster they are assigned to. (C) Fishplot of inferred mitochondrial clone evolution at diagnosis and relapse based on mitochondrial cluster frequencies in (B). (D) Heatmap of epigenetic similarity across all major mitoclones (left). Boxplot of inter-clone epigenetic similarity at diagnosis and relapse (means compared using unpaired student’s t-test) (right). (e) LSI projection of cells containing either 7069T>C or 15,713T>C mitochondrial variants to the healthy hematopoietic reference umap (left). Barplot depicting frequency of cells from diagnosis or relapse timepoints containing either 7069T>C or 15,713T>C variants (right).
Figure 5—figure supplement 2.
Figure 5—figure supplement 2.. Clone-specific epigenetic relapse scores at diagnosis and relapse timepoints.
(a–d) Box plots depicting relapse signature ATAC-seq score calculated for single cells for the indicated patients according to timepoint, diagnosis (left) shown in blue, relapse (right) shown in red. (e–h) Box plots depicting relapse signature score calculated for single cells sorted according to mitoclone designation using mtscATAC-seq for indicated patients. Clones are defined and colored according to the keys shown at right for both diagnosis and relapse samples.
Author response image 1.
Author response image 1.. Volcano plots of differential analysis between mutant vs wildtype of relapse vs diagnosis fold change.
Mutations analyzed include (A) FLT3, (B) DNMT3A, (C) NPM1, (D) TET2. Volcano plots are displayed with unadjusted (left) and adjusted (right) p values. In all analyses but one, p-value adjustment resulted in no significant gene loci chromatin evolution differences. In TET2-mutant samples, 5 loci remained differential after multiple-hypothesis testing correction.
See this image and copyright information in PMC

Update of

Similar articles

See all similar articles

Cited by

References

    1. Abdel-Wahab O, Adli M, LaFave LM, Gao J, Hricik T, Shih AH, Pandey S, Patel JP, Chung YR, Koche R, Perna F, Zhao X, Taylor JE, Park CY, Carroll M, Melnick A, Nimer SD, Jaffe JD, Aifantis I, Bernstein BE, Levine RL. ASXL1 mutations promote myeloid transformation through loss of PRC2-mediated gene repression. Cancer Cell. 2012;22:180–193. doi: 10.1016/j.ccr.2012.06.032. - DOI - PMC - PubMed
    1. Akbani R, Akdemir KC, Aksoy BA, Albert M, Ally A, Amin SB, Arachchi H, Arora A, Auman JT, Ayala B, Baboud J, Balasundaram M, Balu S, Barnabas N, Bartlett J, Bartlett P, Bastian BC, Baylin SB, Behera M, Belyaev D, Benz C, Bernard B, Beroukhim R, Bir N, Black AD, Bodenheimer T, Boice L, Boland GM, Bono R, Bootwalla MS, Bosenberg M, Bowen J, Bowlby R, Bristow CA, Brockway-Lunardi L, Brooks D, Brzezinski J, Bshara W, Buda E, Burns WR, Butterfield YSN, Button M, Calderone T, Cappellini GA, Carter C, Carter SL, Cherney L, Cherniack AD, Chevalier A, Chin L, Cho J, Cho RJ, Choi YL, Chu A, Chudamani S, Cibulskis K, Ciriello G, Clarke A, Coons S, Cope L, Crain D, Curley E, Danilova L, D’Atri S, Davidsen T, Davies MA, Delman KA, Demchok JA, Deng QA, Deribe YL, Dhalla N, Dhir R, DiCara D, Dinikin M, Dubina M, Ebrom JS, Egea S, Eley G, Engel J, Eschbacher JM, Fedosenko KV, Felau I, Fennell T, Ferguson ML, Fisher S, Flaherty KT, Frazer S, Frick J, Fulidou V, Gabriel SB, Gao J, Gardner J, Garraway LA, Gastier-Foster JM, Gaudioso C, Gehlenborg N, Genovese G, Gerken M, Gershenwald JE, Getz G, Gomez-Fernandez C, Gribbin T, Grimsby J, Gross B, Guin R, Gutschner T, Hadjipanayis A, Halaban R, Hanf B, Haussler D, Haydu LE, Hayes DN, Hayward NK, Heiman DI, Herbert L, Herman JG, Hersey P, Hoadley KA, Hodis E, Holt RA, Hoon DSB, Hoppough S, Hoyle AP, Huang FW, Huang M, Huang S, Hutter CM, Ibbs M, Iype L, Jacobsen A, Jakrot V, Janning A, Jeck WR, Jefferys SR, Jensen MA, Jones CD, Jones SJM, Ju Z, Kakavand H, Kang H, Kefford RF, Khuri FR, Kim J, Kirkwood JM, Klode J, Korkut A, Korski K, Krauthammer M, Kucherlapati R, Kwong LN, Kycler W, Ladanyi M, Lai PH, Laird PW, Lander E, Lawrence MS, Lazar AJ, Łaźniak R, Lee D, Lee JE, Lee J, Lee K, Lee S, Lee W, Leporowska E, Leraas KM, Li HI, Lichtenberg TM, Lichtenstein L, Lin P, Ling S, Liu J, Liu O, Liu W, Long GV, Lu Y, Ma S, Ma Y, Mackiewicz A, Mahadeshwar HS, Malke J, Mallery D, Manikhas GM, Mann GJ, Marra MA, Matejka B, Mayo M, Mehrabi S, Meng S, Meyerson M, Mieczkowski PA, Miller JP, Miller ML, Mills GB, Moiseenko F, Moore RA, Morris S, Morrison C, Morton D, Moschos S, Mose LE, Muller FL, Mungall AJ, Murawa D, Murawa P, Murray BA, Nezi L, Ng S, Nicholson D, Noble MS, Osunkoya A, Owonikoko TK, Ozenberger BA, Pagani E, Paklina OV, Pantazi A, Parfenov M, Parfitt J, Park PJ, Park WY, Parker JS, Passarelli F, Penny R, Perou CM, Pihl TD, Potapova O, Prieto VG, Protopopov A, Quinn MJ, Radenbaugh A, Rai K, Ramalingam SS, Raman AT, Ramirez NC, Ramirez R, Rao U, Rathmell WK, Ren X, Reynolds SM, Roach J, Robertson AG, Ross MI, Roszik J, Russo G, Saksena G, Saller C, Samuels Y, Sander C, Sander C, Sandusky G, Santoso N, Saul M, Saw RPM, Schadendorf D, Schein JE, Schultz N, Schumacher SE, Schwallier C, Scolyer RA, Seidman J, Sekhar PC, Sekhon HS, Senbabaoglu Y, Seth S, Shannon KF, Sharpe S, Sharpless NE, Shaw KRM, Shelton C, Shelton T, Shen R, Sheth M, Shi Y, Shiau CJ, Shmulevich I, Sica GL, Simons JV, Sinha R, Sipahimalani P, Sofia HJ, Soloway MG, Song X, Sougnez C, Spillane AJ, Spychała A, Stretch JR, Stuart J, Suchorska WM, Sucker A, Sumer SO, Sun Y, Synott M, Tabak B, Tabler TR, Tam A, Tan D, Tang J, Tarnuzzer R, Tarvin K, Tatka H, Taylor BS, Teresiak M, Thiessen N, Thompson JF, Thorne L, Thorsson V, Trent JM, Triche TJ, Tsai KY, Tsou P, Van Den Berg DJ, Van Allen EM, Veluvolu U, Verhaak RG, Voet D, Voronina O, Walter V, Walton JS, Wan Y, Wang Y, Wang Z, Waring S, Watson IR, Weinhold N, Weinstein JN, Weisenberger DJ, White P, Wilkerson MD, Wilmott JS, Wise L, Wiznerowicz M, Woodman SE, Wu CJ, Wu CC, Wu J, Wu Y, Xi R, Xu AW, Yang D, Yang L, Yang L, Zack TI, Zenklusen JC, Zhang H, Zhang J, Zhang W, Zhao X, Zhu J, Zhu K, Zimmer L, Zmuda E, Zou L. Genomic classification of cutaneous melanoma. Cell. 2015;161:1681–1696. doi: 10.1016/j.cell.2015.05.044. - DOI - PMC - PubMed
    1. Assi SA, Imperato MR, Coleman DJL, Pickin A, Potluri S, Ptasinska A, Chin PS, Blair H, Cauchy P, James SR, Zacarias-Cabeza J, Gilding LN, Beggs A, Clokie S, Loke JC, Jenkin P, Uddin A, Delwel R, Richards SJ, Raghavan M, Griffiths MJ, Heidenreich O, Cockerill PN, Bonifer C. Subtype-specific regulatory network rewiring in acute myeloid leukemia. Nature Genetics. 2019;51:151–162. doi: 10.1038/s41588-018-0270-1. - DOI - PMC - PubMed
    1. Azizi A. Azizi/Nuno eLife 2024. swh:1:rev:f82a1cb04d6292ecdcf35c9e9c6d692c0fab84b2Software Heritage. 2024 https://archive.softwareheritage.org/swh:1:dir:2cdb2d4fed638289596f0e0cc...
    1. Chan SM, Majeti R. Role of DNMT3A, TET2, and IDH1/2 mutations in pre-leukemic stem cells in acute myeloid leukemia. International Journal of Hematology. 2013;98:648–657. doi: 10.1007/s12185-013-1407-8. - DOI - PMC - PubMed

Publication types

Associated data