Skip to main content
Download PDF

DNA methylation in primary myelofibrosis is partly associated with driver mutations and distinct from other myeloid malignancies

Clinical Epigenetics volume 17, Article number: 72 (2025) Cite this article

Abstract

Background

Primary myelofibrosis (PMF) is a clonal blood disorder characterized by mutually exclusive driver mutations in JAK2, CALR, or MPL genes. So far, it is largely unclear if the driver mutations have a specific impact on DNA methylation (DNAm) profiles and how epigenetic alterations in PMF are related to other myeloid malignancies.

Results

When we compared DNAm profiles from PMF patients we found very similar epigenetic modifications in JAK2 and CALR mutated cases, whereas MPL mutations displayed less pronounced and distinct patterns. Furthermore, induced pluripotent stem cell (iPSC) models with JAK2 mutations indicated only a moderate association with PMF-related epigenetic changes, suggesting that these alterations may not be directly driven by the mutations themselves. Additionally, PMF-associated epigenetic changes showed minimal correlation with allele burden and seemed to be largely influenced by shifts in the cellular composition. PMF DNAm profiles compared with those from other myeloid malignancies—such as acute myeloid leukemia, juvenile myelomonocytic leukemia, and myelodysplastic syndrome—showed numerous overlapping changes, making it difficult to distinguish PMF based on individual CpGs. However, a PMF score created by combining five CpGs was able to discern PMF from other diseases.

Conclusion

These findings demonstrate that PMF driver mutations do not directly evoke epigenetic changes. While PMF shares epigenetic alterations with other myeloid malignancies, DNA methylation patterns can distinguish between PMF and related diseases.

Background

Malignancies are marked not only by somatic mutations but also by epigenetic modifications, though the interplay between the two remains largely unexplored. Myeloproliferative neoplasms (MPNs) offer a unique opportunity to investigate whether specific mutations lead to distinct epigenetic changes, as they feature unique driver mutations in Janus kinase 2 (JAK2), particularly at the amino acid position 617 (valine to phenylalanine, V617F), calreticulin (CALR), and myeloproliferative leukemia protein (MPL) [1]. MPN are categorized into various subentities, such as essential thrombocythemia (ET), polycythemia vera (PV) and primary myelofibrosis (PMF), which harbor these mutations at different frequencies [2], with JAK2 V617F mutations occurring in almost all PV cases but only approximately 50–60% ET and PMF cases, CALR mutations occurring in 25–35% of ET and PMF but not PV cases, and MPL mutations occurring in 5–8% of ET and PMF but not PV cases [3].

DNA methylation (DNAm) at CG dinucleotides (CpGs) is an epigenetic mechanism that modulates chromatin structure, transcription, and splicing [4]. Prior research already demonstrated aberrant DNAm in MPN [5]. The different entities of MPN were shown to have similar DNAm changes, which increase during progression and may play an important role in the pathogenesis and leukemic transformation [6]. Recently, it has been suggested that DNAm could serve as a biomarker for the fibrotic progression in PMF [7]. However, it is so far unclear whether distinct driver mutations are associated with specific epigenetic modifications. Understanding these connections could yield valuable insights into disease mechanisms and identify potential therapeutic targets.

Epigenetic aberrations are also implicated in other myeloid disorders, including acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), and juvenile myelomonocytic leukemia (JMML) [8]. However, a comprehensive study that compares these myeloid malignancies with respect to their DNAm profiles is yet elusive. Moreover, many previous studies have not adequately addressed how changes in cellular composition of a given blood sample may affect aberrant DNAm patterns in malignancies [9].

In this study, we have therefore systematically compared the DNAm profiles associated with different driver mutations in PMF and subsequently assessed how these profiles differ from those of other myeloid malignancies.

Methods

Blood samples

Peripheral blood mononuclear cells (PBMCs) of 28 patients diagnosed with PMF, 2 patients diagnosed with ET and 10 healthy donors were used for the study. A detailed description of the sample cohort is shown in Supplemental Table S1. PBMCs were isolated from peripheral blood via Ficoll density gradient centrifugation and DNA was isolated with the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany).

Analysis of mutational burden

We employed a clinically validated amplicon-based next-generation sequencing (NGS) panel (Truseq Custom Amplicon Kit, Illumina, San Diego, USA) to analyze the coding regions of 32 genes commonly associated with hematologic malignancies [10]. Variants were manually reviewed, applying a bidirectional frequency cutoff of > 1% for driver mutations and > 5% for additional mutations.

Analysis of DNA methylation profiles

Genomic DNA of the above-mentioned samples was subjected to bisulfite conversion and analyzed using the Illumina human EPIC methylation microarray version 2 (EPIC v2; GSE277841). We also examined DNAm profiles of induced pluripotent stem cell (iPSC) lines of three PV donors that are either WT, homozygous or heterozygous for the JAK2 V617F mutation (supplemental methods). iPSC clones and their iPSC-derived hematopoietic progenitors (iHPCs) were analyzed with EPIC version 1 (EPIC v1; GSE277890). Additionally, we included DNAm profiles of an earlier study on 4 PV, 4 MF, and 4 healthy control samples hybridized on the 450 k BeadChip for our validation cohort (GSE277866). All Illumina BeadChip microarrays were analyzed at Life and Brain (Bonn, Germany). Further details on preprocessing and DNAm analysis are provided in the supplemental methods.

To investigate the granulocyte bias observed by the epigenetic deconvolution, we utilized a dataset containing triple negative MPN samples (GSE156546: n = 25) [27] and a dataset with different subsets of hematopoietic stem and progenitor cells (GSE63409: n = 30) [11]. To mitigate the impact of variations in cellular composition, we retrieved 289 DNAm profiles of sorted human hematopoietic cell types from the GEO database, including B cells (n = 60), CD4 T cells (n = 63), CD8 T cells (n = 56), granulocytes (n = 34), monocytes (n = 61), NK cells (n = 6), and dendritic cells (n = 9; Supplemental Table S2). Pairwise comparisons of mean DNAm values across all cell types were conducted, focusing on CpGs that did not exceed a beta value threshold of 0.1 in any of these comparisons (393,675 CpGs). For further analysis, we concentrated on CpGs that were also detected in all datasets of myeloid malignancies (216,532 CpGs).

For comparison with other myeloid malignancies, we used DNAm datasets of MDS (GSE221745: n = 5; GSE152710: n = 73), JMML (GSE237299: n = 41), and AML (GSE212937: n = 5; GSE62298: n = 68), as well as, healthy control datasets of either peripheral blood (GSE141682: n = 42; GSE221745: n = 5) or bone marrow (GSE221745: n = 7; GSE124413: n = 40; GSE152710: n = 10 samples). For validation, we used available datasets of PMF (GSE152519: n = 35; GSE118241: n = 22), secondary MF (GSE118241: n = 17), ET (GSE156546: n = 32), AML (GSE159907: n = 316), pediatric AML (GSE133986: n = 64) and healthy controls (GSE118241: n = 6; GSE156546: n = 2; Supplemental Table S3).

Additional methods

Methods for preprocessing and further analysis of DNAm profiles, correlation with gene expression data, generation and characterization of hematopoietic differentiation of iPSC lines, targeted bisulfite amplicon sequencing, and colony-forming unit (CFU) assays are detailed in the supplemental methods.

Results

Aberrant DNA methylation in primary myelofibrosis

We examined DNAm profiles of peripheral blood mononuclear cells of PMF patients with JAK2 V617F mutation (n = 10), CALR mutation (n = 10), MPL mutation (n = 10, including two ET samples), alongside healthy controls (n = 10; Supplemental Table S1). Multidimensional scaling plots (MDS plots) clearly separated DNAm profiles of PMF and healthy samples, whereas PMF samples with different driver mutations revealed some differences but were less clearly separated (Fig. 1a). While samples with additional mutations tended to cluster further away from controls, they did not exhibit a clear separation of specific mutations.

Fig. 1
figure 1

Aberrant DNA methylation in primary myelofibrosis. a Multidimensional scaling plot of DNA methylation profiles in PMF patients with different driver mutations (JAK2, CALR, MPL) and healthy controls (812,274 CpGs). b Scatter plot of mean methylation beta values of PMF patients and healthy controls. Significant hypo- and hypermethylated CpGs are indicated in blue and red (mean DNAm difference > 20%; adjusted p-values < 0.05). Gray numbers indicate all CpGs exceeding the mean DNAm difference > 20%, irrespective of statistical significance. c, d) Enrichment analysis of significant hyper- and hypomethylated CpGs in PMF patients in c genomic regions and d) CpG islands (Hypergeometric distribution: * = p < 0.05, # = p < 10–10, + = p < 10–20, and $ = p < 10–100. e Differential mean DNAm of CpGs adjacent to all significant hypermethylated and hypomethylated CpGs (1 kb window). f Comparison of DNA methylation and gene expression changes (GSE26049) between PMF patients and healthy controls, with genes showing significant differences in both categories highlighted

Full size image

Comparing PMF (n = 28) and control samples, we identified 1,736 hypermethylated and 3,641 hypomethylated CpGs in PMF (adj p < 0.05; difference of mean DNAm > 20%; Fig. 1b; Supplemental Figure S1a,b). The most significantly hypermethylated CpGs were frequently linked to genes involved in the pathogenesis of hematological diseases, including RUNX1 [12], BRD4 [13], SRSF2 [14], SETBP1 [15] and TNFSF10 [16]. Overall, the genes associated with differentially methylated CpGs were enriched in the Gene Ontology (GO) categories for immune response (Supplemental Figure S1c) and were predominantly located in intergenic regions (Fig. 1c), rather than with CpG islands commonly associated with promoter regions (Fig. 1d).

To assess whether aberrant DNAm affects single CpG sites or broader differentially methylated regions (DMRs), we analyzed DNAm in the surrounding of the 1,736 hyper- and 3,641 hypomethylated CpGs. A significant gain or loss of DNAm was observed within a 500 bp window around these CpGs (Fig. 1e). We also compared DNAm changes to gene expression profiles from a publicly available dataset with nine PMF samples (GSE26049). While no clear overall association was found, some candidate genes exhibited concordant changes in DNAm and gene expression, such as hypermethylated and downregulated ZFP36L1, linked to myelofibrosis progression [17] and INPP5D known to be downregulated by JAK2 V617F [18]. Conversely, hypomethylated and upregulated genes included LAPTM4B [19], and NEDD4 [20], both associated with oncogenic potential (Fig. 1f). To further validate these findings, we compared gene expression profiles of CD34 + sorted cells from PMF patients and controls (GSE53482) and found a similar association (Supplemental Figure S1d).

Similar epigenetic effects from JAK2 and CALR mutations

Next, we categorized PMF samples based on their specific driver mutation. Comparing healthy samples (n = 10) with MPN samples harboring specific driver mutations (n = 10 each), we identified 2,770 hyper- and 7,611 hypomethylated CpGs for JAK2, 2,361 and 9,426 for CALR, and 238 and 282, respectively, for MPL mutations (all adj p < 0.05; difference of mean DNAm > 20%; Fig. 2a-c). Thus, MPL mutations revealed fewer modifications than those with JAK2 and CALR mutations, and this was also observed when we excluded the two ET samples with MPL mutation (882 hyper- and 598 hypomethylated CpGs). Notably, a direct comparison between JAK2 and CALR samples revealed only three significant CpGs, indicating minimal differences in their impact on DNAm profiles (Fig. 2d). Comparison of JAK2 versus MPL (Fig. 2e) and CALR versus MPL (Fig. 2f) revealed a greater number of significant CpGs (Supplemental Table S4), suggesting that MPL mutations have distinct epigenetic consequences compared to JAK2 or CALR mutations (Fig. 2g).

Fig. 2
figure 2

Changes in DNA methylation according to driver mutations. a-c Scatter plots illustrating the mean DNAm beta values in PMF patients compared to healthy controls, stratified by driver mutation: a) JAK2 mutation, b) CALR mutation, and c) MPL mutation. d-f Additional scatter plots comparing the mean DNAm beta values of PMF patients based on their driver mutation: d) CALR versus JAK2; e) MPL versus JAK2; and f) MPL versus CALR. Significant hypo- and hypermethylated CpGs are indicated in blue and red, respectively (mean DNAm difference > 20%; adjusted p-values < 0.05). g Comparison of significantly differentially methylated CpGs between JAK2 and MPL versus CALR and MPL, with gene names for overlapping CpGs highlighted

Full size image

Patient specific DNAm patterns in a JAK2 iPSC model

To further explore the link between JAK2 V617F mutations and epigenetic aberrations, we analyzed DNAm profiles in iPSC lines derived from three PV patients previously generated and independent of the above PMF samples, for each of them wild type (WT) JAK2, heterozygous (het), and homozygous (hom) JAK2 V617F mutations [21, 22]. The iPSC model offers the advantage of working with clonal and homogenous cell populations, where all cells within a given cell population harbor either heterozygous or homozygous JAK2 V617F mutations or no JAK2 mutation. In the pluripotent state, no significant DNAm differences were observed between WT and either het or hom iPSC clones (adj p < 0.05; Fig. 3a, b). However, focusing on CpGs exhibiting significant hyper- or hypomethylation in the blood samples of JAK2 V617F positive patients versus healthy controls (2,198 versus 5,322 CpGs; lower CpG numbers than in Fig. 2a due to the different EPIC array version), we noted a modest yet significant hypomethylation in JAK2 V617F iPSCs compared to WT iPSCs (Fig. 3c, d; Supplemental Figure S2a, b).

Fig. 3
figure 3

iPSCs with JAK2 V617F fail to recapitulate disease-associated changes. a, b Scatter plots showing mean DNAm beta values of a) wild type (WT) versus JAK2 V617F heterozygous (het) iPSCs, and b) WT versus JAK2 V617F homozygous (hom) iPSCs. The numbers of CpGs with > 20% DNAm difference are indicated, but none reached statistical significance. c, d To determine if DNAm changes in PMF patients with JAK2 V617F are reflected in iPSCs with or without JAK2 V617F, we focused on CpGs that were significantly differentially methylated in JAK2 V617F PMF versus healthy control (from Fig. 2a). Average DNAm changes were then analyzed in these CpGs in iPSCs with either c) WT versus heterozygous JAK2 V617F, or d) WT versus homozygous JAK2 V617F. e, f Following differentiation of iPSC lines into hematopoietic progenitor cells (iHPCs), scatter plots depict mean DNAm beta values for e) WT versus JAK2 V617F heterozygous iHPCs, and f) WT versus JAK2 V617F homozygous iHPCs (none of the CpGs reached statistical significance). Gray numbers indicate all CpGs exceeding the mean DNAm difference > 20%, irrespective of statistical significance. g, h The CpGs with significant differences in JAK2 V617F PMF versus healthy controls were reanalyzed in iHPCs: g) heterozygous and h) homozygous JAK2 V617F iHPCs exhibited an overall decrease in DNAm at CpGs that gained or lost methylation in JAK2 V617F PMF. Statistical significance was evaluated using one-way ANOVA

Full size image

To determine whether the impact of JAK2 V617F mutations was masked in the pluripotent state, we differentiated these clones into hematopoietic progenitor cells (iHPCs; Supplemental Figure S3a). After 16 days, all clones generated non-adherent cells exhibiting typical hematological morphology, with flow cytometry confirming upregulation of various hematopoietic markers (Supplemental Figure S3b, c). Heterozygous and homozygous JAK2 V617F iHPCs showed a bias toward CD235a/glycophorin A positive erythroid cells in line with our previous studies [21, 22]. DNAm profiles validated that iHPCs had exited the pluripotent state and were aligned toward mesodermal lineage (Supplemental Figure S3d, e) [23]. When comparing iHPCs to iPSCs, we identified 3,201 hypermethylated and 25,507 hypomethylated CpGs (adj p < 0.05; difference of mean DNAm > 20%; Supplemental Figure S3f). These CpGs were enriched in genes related to Gene Ontology categories including blood vessel development, regulation of signaling and migration, indicating that the differentiation captured epigenetic changes of hematopoietic development (Supplemental Figure S3g). However, even in iHPCs there were no significant differences between WT and JAK2 V617F mutated clones. This lack of distinction might stem from inherent variability during the differentiation process (Fig. 3e, f).

Focusing on the CpGs with notable DNAm changes in JAK2 V617F patients, we found that hypomethylated CpGs in JAK2 V617F PMF also exhibited hypomethylation in iHPCs carrying the JAK2 V617F mutation. However, also the hypermethylated CpGs in patients showed moderate hypomethylation in iHPCs (Fig. 3g, h). A direct comparison of DNAm changes associated with JAK2 V617F in iHPCs and PMF patients revealed a moderate but significant association between hypomethylated regions (Supplemental Figure S2c, d). This could be linked to a partial recapitulation of MPN phenotype observed in the JAK2 V617F iPSC clones. Overall, our findings suggest that the iPSC model does not accurately reflect the DNA methylation changes seen in PMF, indicating that the epigenetic alterations may not be directly driven by the JAK2 V617F mutation alone.

Epigenetic age is accelerated in primary myelofibrosis

Subsequently, we investigated if the acceleration of epigenetic age predictions in PMF varied among samples with different driver mutations. Building on our earlier research using bisulfite amplicon sequencing (BA-seq) of three age-associated regions of PDE4C, FHL2, and CCDC102B, which indicated an overall acceleration of epigenetic age in MPN [24], we employed two epigenetic signatures [25, 26] to further validate that PMF exhibits significantly accelerated epigenetic age, consistent across all three driver mutations (Fig. 4a–d). The difference between predicted and chronological age (delta-age) did not correlate with mutation burden.

Fig. 4
figure 4

Age-associated DNAm changes in primary myelofibrosis. a–d The correlation between epigenetic age predictions with chronological age and the difference between predicted and chronological age (delta-age) was determined with epigenetic clocks developed by a,b) Horvath [26] and c,d) Han et al. [25]. Statistical significance was determined using an unpaired t-test. e–h The DNAm at an age-associated region in PDE4C was analyzed by bisulfite amplicon sequencing (BA-seq). The heatmaps exemplify the presence of methylated (red) and non-methylated (blue) CpGs within the PDE4C amplicon, covering 26 neighboring CpGs. The age-associated CpG of the aging signature is indicated by arrow. Exemplary heatmaps are depicted for e) a healthy donor, and f) a PMF patient blood sample of the same age. The frequency of reads is clustered according to their DNAm patterns. The same analysis was performed in colony-forming units (CFUs) on day 14 that were either g) a wild type (WT), or h) harbored the JAK2 V617F mutation. Unlike the blood samples from PMF patients or controls, the CFUs exhibited a distinct DNAm pattern that appears to reflect the clonal characteristics of the colony-initiating cells

Full size image

To delve deeper into the heterogeneity of epigenetic aging, we revisited the BA-seq data of amplicons within the three age-associated regions. Notably, the DNAm at neighbouring CpGs within individual reads of amplicons appeared to be independently regulated (Fig. 4e, f; Supplemental Figure S4a), corroborating previous findings in healthy samples [25]. However, in a clonal disease context, we expected to see a dominant pattern reflective of the tumor-initiating cells. Consequently, we examined BA-seq DNAm patterns at the same age-associated CpGs in single-cell-derived colony-forming units (CFUs; Fig. 4g, h; Supplemental Figure S4b). In fact, we observed prominent patterns in CFUs, regardless of whether they are derived from cells with or without JAK2 V617F mutation, indicating that age-associated DNAm patterns remain largely preserved at least during the CFU formation.

Disease-associated DNAm and cellular composition

We hypothesized that the allele burden of driver mutations serves as a proxy for the fraction of malignant cells, potentially correlating with PMF-associated aberrant DNAm. In blood samples from PMF patients, allele frequencies ranged from 26 to 93%, although these values were likely different in the PBMCs due to depletion of granulocytes. Multidimensional scaling analysis revealed minimal clustering of DNAm profiles according to allele burden (Fig. 5a), suggesting that aberrant DNAm may not be homogeneous across malignant clones. Furthermore, only few individual CpGs exhibited moderate correlation with the allele burden of JAK2 V617F, CALR, or MPL mutations (Supplemental Table S5). For example, cg14658896_BC21 (R = 0.62) and cg16965444_BC21 (R = 0.56) displayed an association with allele burden, irrespective of the specific driver mutation (Fig. 5b, c).

Fig. 5
figure 5

DNA methylation changes are largely attributed to the cellular composition. a The multidimensional scaling plot demonstrates that PMF sample did not cluster by mutation allele frequency. b, c Correlation between DNAm and allele burden (across all driver mutations, given that they hardly affected DNAm) for the top two candidate CpGs: b) cg14658896_BC21 and c) cg16965444_BC21. d An epigenetic deconvolution algorithm [51] was applied to estimated fractions of granulocytes, monocytes, B cells, NK cells, CD4 and CD8 T cells. e, f To determine how many significant DNAm changes in PMF versus controls is attributed to CpGs that have high variation between leukocyte subsets we compared scatter plots e) before and f) after exclusion of CpGs with more than 10% DNAm between any of the leukocyte subsets (only CpGs measured across all datasets are shown). Significant hypo- and hypermethylated CpGs are indicated in blue and red (mean DNAm difference > 20%; adjusted p-values < 0.05). Gray numbers indicate all CpGs exceeding the mean DNAm difference > 20%, irrespective of statistical significance

Full size image

So far, it is largely unclear how much of the aberrant DNAm in myeloid malignancies can be attributed to the changes in the cellular composition of blood. While precise flow cytometric analysis or manual cell differential counts were not available for our samples, we used a deconvolution algorithm for cell type-specific DNAm signatures [27]. This analysis revealed higher predictions for granulocytes in PMF samples, as compared to controls, albeit PBMCs were used for both groups (Fig. 5d). To further explore this granulocyte bias, we alternatively applied the deconvolution algorithm to triple negative MPN samples (GSE156546) [28] and subsets of hematopoietic stem and progenitor cells (GSE63409) [11]. These results suggested that myeloid bias of predictions in MPN and the increased fraction of progenitor cells might contribute to higher estimates of granulocytes in the deconvolution algorithm (Supplemental Figure S5).

Given the significant differences in cellular composition between PMF patients and healthy donors, we subsequently concentrated on CpGs exhibiting stable DNAm levels across all healthy leukocyte subsets. This approach was taken to minimize the impact of disparate cellular compositions. Analyzing 289 DNAm profiles from sorted cell types (Supplemental Table S2), we excluded CpGs with over 10% DNAm variation in pairwise comparisons. This process yielded 393,675 CpGs with consistent DNAm across various healthy donor cell types. Comparing PMF and control samples, we found 978 CpGs significantly hypermethylated and 1,659 hypomethylated in PMF (adj p < 0.05; mean DNAm difference > 20%; Fig. 5e). However, after excluding CpGs with high-variation between leukocyte subsets, only 60 were significantly hypermethylated and 141 hypomethylated (Fig. 5f), indicating that while many PMF-associated epigenetic aberrations are linked to cellular composition, some remain distinctive.

Comparative epigenetic analysis with other myeloid malignancies

To examine how DNAm profiles in PMF patients compare with other myeloid malignancies, we analyzed datasets from bone marrow samples of myelodysplastic syndrome (MDS) patients [29, 30], peripheral blood of juvenile myelomonocytic leukemia (JMML), and peripheral blood from acute myeloid leukemia (AML) patients [31, 32]. To account for potential biases related to tissue type, we compared these against corresponding healthy controls (Supplemental Table S3). Multidimensional scaling showed that healthy samples clustered closely together, while JMML formed a distinct cluster. PMF, MDS, and AML profiles did not separate clearly; however, PMF samples tended to cluster closer to healthy samples, followed by MDS and AML (Fig. 6a,b). A similar pattern was observed when we excluded the CpGs with high variability between leukocyte subsets (Supplemental Fig. 6a, b).

Fig. 6
figure 6

Comparison of myeloid malignancies after exclusion of cell type-specific CpGs. a, b Multidimensional scaling plots of DNAm profiles (216,532 CpG sites) in patients with PMF, MDS, JMML, AML, and healthy controls (peripheral blood (PB) and bone marrow (BM)): a) first versus second dimension; b) third versus fourth dimension. c Scatter plots showing mean DNAm beta values of healthy control versus PMF, MDS, JMML, or AML, accounting for potential differences between peripheral blood and bone marrow by using appropriate control sets. Significant hypo- and hypermethylated CpGs are indicated in blue and red (mean DNAm difference > 20%; adjusted p-values < 0.05). Gray numbers indicate all CpGs exceeding the mean DNAm difference > 20%, irrespective of statistical significance. d, e Venn diagrams illustrating CpGs that are overlapping d) hyper- or e) hypomethylated in the above-mentioned comparisons of four myeloid malignancies after the exclusion of cell type-specific CpGs

Full size image

Subsequently, we analyzed pairwise comparisons of DNAm changes between individual diseases and compounding controls (either BM or PB). Without filtering for cell type-specific CpGs, we identified many significant CpGs for each disease (adj p < 0.05; difference of mean DNAm > 20%; Supplemental Fig. 6c), and there was a considerable overlap in CpGs that were hyper- or hypomethylated across PMF, MDS, JMML, and AML (Supplemental Fig. 6d, e). However, when we excluded CpGs with high variability between cell types, the differential DNAm signatures for each disease became significantly reduced and overlap among myeloid malignancies was minimal (Fig. 6c–e). Only one CpG site (cg04470072_TC11) was hypermethylated across all these malignancies. Thus, while the DNAm changes in comparison with healthy blood samples can be largely attributed to the cellular composition, certain epigenetic aberrations may be indicative of specific diseases.

Establishing an epigenetic signature for PMF

To determine if PMF possesses unique DNAm patterns that could aid in diagnosis, we performed pairwise comparisons between PMF and other myeloid malignancies, identifying significant differences in 387 CpGs for MDS, 700 for JMML, and 689 for AML (Supplemental Fig. 7a–c). Among these, 17 CpGs were overlapping hypermethylated and 36 hypomethylated which are specific for PMF (Supplemental Fig. 7d, e). Next, we analyzed if these CpGs were also differentially methylated in PMF versus heathy controls (Fig. 6d, e). In fact, six hypomethylated CpGs (none of the hypermethylated CpGs) were overlapping in these comparisons: cg02210934 (no gene), cg02739280 (NAV2), cg21708058 (TACC1), cg07197092 (no gene), cg08069247 (HABP2), and cg04902833 (C17orf99; Supplemental Fig. 8a). However, none of these CpGs could reliably discern all PMF from other samples. Furthermore, cg04902833 (C17orf99) was excluded from further analysis, because it was also hypomethylated in our control samples, indicating that it might be affected by batch effects or microarray versions.

Since single CpG analysis was insufficient for effective differentiation, we combined the remaining five CpGs into a PMF score (calculated as 5 minus the sum of the beta values, with higher scores indicating a stronger association with PMF). This score was not intended for clinical application, but rather to test if DNAm patterns can be used to discern MPN from other myeloid malignancies. In fact, the 5 CpG signature could effectively distinguished PMF samples from all controls and most myeloid malignancies (Fig. 7a). To further validate the score, we utilized other available datasets of PMF [7, 17], secondary myelofibrosis [17], ET [28], AML [33], pediatric AML [34], additional healthy controls, and own yet unpublished 450 k profiles of post PV-MF, PV, and healthy controls (Supplemental Table S3). All five CpGs of the PMF score showed clear offsets in MPN samples, except for the ET samples (Supplemental Fig. 8b). The PMF score was higher (> 1) in almost all PMF, MF and PV samples (Fig. 7b). Importantly, the PMF score did not correlate with the allele burden of driver mutations, reinforcing our earlier finding that epigenetic aberrations are heterogeneous within the malignant clone (Fig. 7c). Furthermore, PMF score did not reveal significant differences in our samples with JAK2 V617F, CALR, and MPL mutation. Notably, in comparison with a public dataset of triple negative (TN) PMF samples, the PMF score was lower in TN than in JAK2 V617F or CALR mutated samples (Fig. 7d).

Fig. 7
figure 7

A five-CpG signature can discern PMF from other malignancies and controls. a, b The DNAm levels at five CpGs (cg02210934, cg02739280, cg21708058, cg07197092, and cg08069247) were combined into a simple PMF score (PMF score = 5—sum of the five DNAm values). The PMF score is provided for datasets of the a) training set, and b) independent validation set. The datasets of PMF, other myeloid malignancies and healthy samples used from both our and public datasets are indicated with GSE numbers. c The PMF score did not correlate with mutation allele burden. d Box plot demonstrating that the driver mutations did not have significant impact on the PMF score. However, triple negative (TN) samples of publicly available PMF datasets (GSE152519) revealed a lower PMF score. Statistical significance was determined using an unpaired t-test

Full size image

Discussion

The results of this study indicate that the driver mutations in PMF have surprisingly little impact on the disease-associated epigenetic modifications. Significant DNAm differences between PMF with JAK2 and CALR mutations were scarce, consistent with a recent study in ET patients [28]. In contrast, the observed differences in MPL mutations might partly be attribute to lower allele burden, inclusion of two ET samples, and inclusion of one sample with a low additional allele frequency of an additional CALR mutation in this group. Moreover, the MPL comprised nine of ten male samples—since there are also moderate gender-related DNAm differences on autosomes this might also contribute to the discrepancy to the JAK2 and CALR groups [35, 36]. The molecular link between the genomic and epigenetic modifications remains unclear. All three driver mutations activate the JAK-STAT signaling pathway, and while JAK2 and CALR mutations were suggested to cause distinct mitotic defects leading to chromosomal instability [37], MPL mutation directly affect the structure and function of thrombopoietin receptor [38, 39].

When we investigated the impact of the JAK2 V617F mutation in iPSCs, we observed no significant epigenetic changes. The moderate association of JAK2 V617F PMF-associated DNAm changes in iHPCs might also be attributed to the pathognomonically skewed megakaryocytic and erythroid differentiation of iPSCs with the JAK2 V617F mutation [21, 22]. It is conceivable that time of hematopoietic differentiation in vitro does not suffice to evoke mutation-associated DNAm changes—particularly given that MPN seems to usually develop over decades. Thus, the iPSC model may not be ideally suited to recapitulate the complex epigenetic modifications that arise over many years during development of the disease—at least, it did not reflect an immediate sequel of MPN-associated mutations on corresponding DNAm changes.

Additionally, the PMF-associated DNAm changes did not correlate with mutational allele burden, indicating that the epigenetic modifications are not homogenous in the entire malignant clone. Only very few CpGs correlated with allele burden, and it might be speculated that these arise earlier after the driver mutation and hence better reflect fraction of the malignant clone. Furthermore, the epigenetic makeup can also be altered in the non-malignant hematopoietic compartments or in the bone marrow microenvironment [40].

Previous studies suggested that alterations in the DNAm landscape play an important role in the pathogenesis and leukemic transformation of MPN [6]. This might be caused by additional mutations in epigenetic writers, such as ASXL1, DNMT3A, SRSF2, and TET2 [39, 41,42,43]. Either way, the PMF-associated epigenetic changes were observed across different secondary mutations and many of these mutations are also frequently observed in other malignancies. This suggests that secondary mutations in epigenetic writers are not the sole drivers of the epigenetic changes seen in PMF.

Given that different cell types exhibit distinct DNAm profiles, it is crucial to consider cellular heterogeneity when interpreting DNAm data in hematological diseases [44,45,46]. We excluded CpGs that displayed variability among different leukocyte subsets, demonstrating that previously noted differences in comparisons of diseased versus healthy blood largely stem from cellular composition. Furthermore, epigenetic aberrations have hardly been compared across different myeloid malignancies [47,48,49]. Our integrative analysis across multiple studies demonstrated that also such comparisons are largely affected by differences in the cellular composition.

We subsequently followed the question if PMF-associated DNAm patterns are disease specific and how they are related to other myeloid malignancies. None of the individual CpGs could reliably discern PMF from healthy controls as well as from other myeloid malignancies, underscoring the need to combine multiple CpGs into comprehensive epigenetic signatures. Furthermore, there was a gradual overlap of aberrant DNAm in PMF with all other hematopoietic malignancies. Our pairwise comparisons indicate that there are significance differences in the methylome of PMF, MDS, AML, and JMML samples. However, such analysis can be biased by many parameters, including differences in the cellular composition, allele burden, gender, array types and batch variation between different studies. Thus, further analysis is needed to elaborate the characteristic epigenetic phenotype of hematopoietic malignancies.

As a proof of concept, we have exemplarily generated a 5 CpG PMF score to discern PMF from other hematopoietic malignancies. However, it could not discern PMF and PV samples. Either way, our score has not been validated and optimized for clinical application. Looking ahead, such scores could ultimately be used to support stratification of specific MPN entities or diagnosis of triple negative cases. Further refinements may even reveal the transition of MPN into a secondary AML. Small epigenetic signatures may facilitate a clinical translation, because they enable targeted analysis, using methods such as pyrosequencing, BA-seq, or digital PCR, to facilitate fast and cost-effective assessments for clinical application [50]. However, given the heterogeneity of aberrant methylation patterns that seem to evolve rather independent from driver mutations, it appears to be necessary to consider larger signatures to reliably discern hematopoietic diseases.

Conclusions

The results of this study demonstrate that epigenetic patterns can differentiate myeloid malignancies, with the observed differences not directly linked to specific driver mutations but rather influenced by cellular composition and overlapping across various myeloid diseases.

Limitations

Our study is limited in that we did not analyze the methylome of sorted and defined hematopoietic subpopulations—ideally sorted as healthy and malignant cell fractions directly from the bone marrow. In fact, our results demonstrate that many epigenetic aberrations in myeloid malignancies can be attributed to cell type-specific DNA methylation changes. The use of PBMCs, absence of conventional leukocyte counts, and the relatively small samples sizes are further limitations. While PMF with JAK2 and CALR mutation have similar epigenetic aberrations, it needs to be further explored why our MPL samples had less pronounced aberrations. At this point, we could not identify reliable epigenetic patterns for specific driver mutations or MPN entities that could be applied for clinical application.

Data availability

The DNAm data on PMF (EPIC v2) and MPN (450 k) are accessible in the Gene Expression Omnibus (GEO) under super series GSE277889 with sub series GSE277841, and GSE277866. Additionally, DNAm profiles for JAK2 iPSCs and iHPCs (EPIC v1) can be found under GSE277890.

References

  1. Tefferi A, Lasho TL, Finke CM, Knudson RA, Ketterling R, Hanson CH, et al. CALR vs JAK2 vs MPL-mutated or triple-negative myelofibrosis: clinical, cytogenetic and molecular comparisons. Leukemia. 2014;28(7):1472–7.

    Article  CAS  PubMed  Google Scholar 

  2. Tefferi A, Pardanani A. Myeloproliferative neoplasms: a contemporary review. JAMA Oncol. 2015;1(1):97–105.

    Article  PubMed  Google Scholar 

  3. Klampfl T, Gisslinger H, Harutyunyan AS, Nivarthi H, Rumi E, Milosevic JD, et al. Somatic mutations of calreticulin in myeloproliferative neoplasms. New Engl J Med. 2013;369(25):2379–90.

    Article  CAS  PubMed  Google Scholar 

  4. Lev Maor G, Yearim A, Ast G. The alternative role of DNA methylation in splicing regulation. Trends Genet. 2015;31(5):274–80.

    Article  CAS  PubMed  Google Scholar 

  5. Nischal S, Bhattacharyya S, Christopeit M, Yu Y, Zhou L, Bhagat TD, et al. Methylome profiling reveals distinct alterations in phenotypic and mutational subgroups of myeloproliferative neoplasms. Can Res. 2013;73(3):1076–85.

    Article  CAS  Google Scholar 

  6. Pérez C, Pascual M, Martín-Subero JI, Bellosillo B, Segura V, Delabesse E, et al. Aberrant DNA methylation profile of chronic and transformed classic Philadelphia-negative myeloproliferative neoplasms. Haematologica. 2013;98(9):1414–20.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Lehmann U, Stark H, Bartels S, Schlue J, Büsche G, Kreipe H. Genome-wide DNA methylation profiling is able to identify prefibrotic PMF cases at risk for progression to myelofibrosis. Clin Epigenetics. 2021. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13148-021-01010-y.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Zhao A, Zhou H, Yang J, Li M, Niu T. Epigenetic regulation in hematopoiesis and its implications in the targeted therapy of hematologic malignancies. Signal Transduction Targeted Therapy. 2023. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/s41392-023-01342-6.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Loyfer N, Magenheim J, Peretz A, Cann G, Bredno J, Klochendler A, et al. A DNA methylation atlas of normal human cell types. Nature. 2023;613(7943):355–64.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Kirschner M, Maurer A, Wlodarski MW, Ventura Ferreira MS, Bouillon AS, Halfmeyer I, et al. Recurrent somatic mutations are rare in patients with cryptic dyskeratosis congenita. Leukemia. 2018;32(8):1762–7.

    Article  CAS  PubMed  Google Scholar 

  11. Jung N, Dai B, Gentles AJ, Majeti R, Feinberg AP. An LSC epigenetic signature is largely mutation independent and implicates the HOXA cluster in AML pathogenesis. Nat Commun. 2015;6:8489.

    Article  CAS  PubMed  Google Scholar 

  12. Adnan Awad S, Dufva O, Ianevski A, Ghimire B, Koski J, Maliniemi P, et al. RUNX1 mutations in blast-phase chronic myeloid leukemia associate with distinct phenotypes, transcriptional profiles, and drug responses. Leukemia. 2021;35(4):1087–99.

    Article  CAS  PubMed  Google Scholar 

  13. Dawson MA, Gudgin EJ, Horton SJ, Giotopoulos G, Meduri E, Robson S, et al. Recurrent mutations, including NPM1c, activate a BRD4-dependent core transcriptional program in acute myeloid leukemia. Leukemia. 2014;28(2):311–20.

    Article  CAS  PubMed  Google Scholar 

  14. Kim E, Ilagan JO, Liang Y, Daubner GM, Lee SCW, Ramakrishnan A, et al. Mutations contribute to myelodysplasia by mutant-specific effects on exon recognition. Cancer Cell. 2015;27(5):617–30.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Makishima H, Yoshida K, Nguyen N, Przychodzen B, Sanada M, Okuno Y, et al. Somatic SETBP1 mutations in myeloid malignancies. Nat Genet. 2013;45(8):942–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. De Carvalho DD, Binato R, Pereira WO, Leroy JM, Colassanti MD, Proto-Siqueira R, et al. BCR-ABL-mediated upregulation of PRAME is responsible for knocking down TRAIL in CML patients. Oncogene. 2011;30(2):223–33.

    Article  PubMed  Google Scholar 

  17. Martinez-Calle N, Pascual M, Ordonez R, Eneriz ESJ, Kulis M, Miranda E, et al. Epigenomic profiling of myelofibrosis reveals widespread DNA methylation changes in enhancer elements and ZFP36L1 as a potential tumor suppressor gene that is epigenetically regulated. Haematologica. 2019;104(8):1572–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Glück M, Dally L, Jücker M, Ehm P. JAK2-V617F is a negative regulation factor of SHIP1 protein and thus influences the AKT signaling pathway in patients with Myeloproliferative neoplasm (MPN). Int J Biochem Cell Biol. 2022;149: 106229.

    Article  PubMed  Google Scholar 

  19. Li L, Wei XH, Pan YP, Li HC, Yang H, He QH, et al. A novel cancer-associated gene motivates multidrug resistance through efflux and activating PI3K/AKT signaling. Oncogene. 2010;29(43):5785–95.

    Article  CAS  PubMed  Google Scholar 

  20. Jayaprakash S, Hegde M, BharathwajChetty B, Girisa S, Alqahtani MS, Abbas M, et al. Unraveling the potential role of NEDD4-like E3 ligases in cancer. Int J Mol Sci. 2022;23(20):12380.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Flosdorf N, Böhnke J, de Toledo MAS, Lutterbach N, Lerma VG, Grasshoff M, et al. Proinflammatory phenotype of iPS cell-derived JAK2 V617F megakaryocytes induces fibrosis in 3D in vitro bone marrow niche. Stem Cell Rep. 2024;19(2):224–38.

    Article  CAS  Google Scholar 

  22. Satoh T, Toledo MAS, Boehnke J, Olschok K, Flosdorf N, Gotz K, et al. Human DC3 antigen presenting dendritic cells from induced pluripotent stem cells. Front Cell Dev Biol. 2021;9: 667304.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Schmidt M, Zeevaert K, Elsafi Mabrouk MH, Goetzke R, Wagner W. Epigenetic biomarkers to track differentiation of pluripotent stem cells. Stem Cell Rep. 2023;18(1):145–58.

    Article  CAS  Google Scholar 

  24. Vieri M, Tharmapalan V, Kalmer M, Baumeister J, Nikolic M, Schnitker M, et al. Cellular aging is accelerated in the malignant clone of myeloproliferative neoplasms. Blood Cancer J. 2023;13(1):164.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Han Y, Franzen J, Stiehl T, Gobs M, Kuo CC, Nikolic M, et al. New targeted approaches for epigenetic age predictions. BMC Biol. 2020;18(1):71.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Horvath S. DNA methylation age of human tissues and cell types. Genome Biol. 2013;14(10):R115.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13:86.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Alimam S, Villiers W, Dillon R, Simpson M, Runglall M, Smith A, et al. Patients with triple-negative, JAK2V617F- and CALR-mutated essential thrombocythemia share a unique gene expression signature. Blood Adv. 2021;5(4):1059–68.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Marin-Bejar O, Romero-Moya D, Rodriguez-Ubreva J, Distefano M, Lessi F, Aretini P, et al. Epigenome profiling reveals aberrant DNA methylation signature in GATA2 deficiency. Haematologica. 2023;108(9):2551–7.

    CAS  PubMed  PubMed Central  Google Scholar 

  30. Cabezon M, Malinverni R, Bargay J, Xicoy B, Marce S, Garrido A, et al. Different methylation signatures at diagnosis in patients with high-risk myelodysplastic syndromes and secondary acute myeloid leukemia predict azacitidine response and longer survival. Clin Epigenetics. 2021;13(1):9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. O’Brien G, Cecotka A, Manola KN, Pagoni MN, Polanska J, Badie C. Epigenetic signature of ionizing radiation in therapy-related AML patients. Heliyon. 2024;10(1): e23244.

    Article  CAS  PubMed  Google Scholar 

  32. Ferreira HJ, Heyn H, Vizoso M, Moutinho C, Vidal E, Gomez A, et al. DNMT3A mutations mediate the epigenetic reactivation of the leukemogenic factor MEIS1 in acute myeloid leukemia. Oncogene. 2016;35(23):3079–82.

    Article  CAS  PubMed  Google Scholar 

  33. Giacopelli B, Wang M, Cleary A, Wu YZ, Schultz AR, Schmutz M, et al. DNA methylation epitypes highlight underlying developmental and disease pathways in acute myeloid leukemia. Genome Res. 2021;31(5):747–61.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Yamato G, Kawai T, Shiba N, Ikeda J, Hara Y, Ohki K, et al. Genome-wide DNA methylation analysis in pediatric acute myeloid leukemia. Blood Adv. 2022;6(11):3207–19.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Singmann P, Shem-Tov D, Wahl S, Grallert H, Fiorito G, Shin SY, et al. Characterization of whole-genome autosomal differences of DNA methylation between men and women. Epigenetics Chromatin. 2015;8:43.

    Article  PubMed  PubMed Central  Google Scholar 

  36. Yousefi P, Huen K, Dave V, Barcellos L, Eskenazi B, Holland N. Sex differences in DNA methylation assessed by 450 K BeadChip in newborns. BMC Genomics. 2015;16:911.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Holl K, Chatain N, Krapp S, Baumeister J, Maie T, Schmitz S, et al. Calreticulin and JAK2V617F driver mutations induce distinct mitotic defects in myeloproliferative neoplasms. Sci Rep. 2024;14(1):2810.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  38. Plo I, Bellanné-Chantelot C, Mosca M, Mazzi S, Marty C, Vainchenker W. Genetic alterations of the thrombopoietin/MPL/JAK2 axis impacting megakaryopoiesis. Front Endocrinology. 2017. https://doiorg.publicaciones.saludcastillayleon.es/10.3389/fendo.2017.00234.

    Article  Google Scholar 

  39. Vainchenker W, Kralovics R. Genetic basis and molecular pathophysiology of classical myeloproliferative neoplasms. Blood. 2017;129(6):667–79.

    Article  CAS  PubMed  Google Scholar 

  40. Desterke C, Martinaud C, Ruzehaji N, Le Bousse-Kerdiles MC. Inflammation as a keystone of bone marrow stroma alterations in primary myelofibrosis. Mediators Inflamm. 2015;2015: 415024.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Nischal S, Bhattacharyya S, Christopeit M, Yu Y, Zhou L, Bhagat TD, et al. Methylome profiling reveals distinct alterations in phenotypic and mutational subgroups of myeloproliferative neoplasms. Cancer Res. 2013;73(3):1076–85.

    Article  CAS  PubMed  Google Scholar 

  42. Nielsen HM, Andersen CL, Westman M, Kristensen LS, Asmar F, Kruse TA, et al. Epigenetic changes in myelofibrosis: Distinct methylation changes in the myeloid compartments and in cases with ASXL1 mutations. Sci Rep. 2017;7(1):6774.

    Article  PubMed  PubMed Central  Google Scholar 

  43. McPherson S, McMullin MF, Mills K. Epigenetics in myeloproliferative neoplasms. J Cell Mol Med. 2017;21(9):1660–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Jaffe AE, Irizarry RA. Accounting for cellular heterogeneity is critical in epigenome-wide association studies. Genome Biol. 2014;15(2):R31.

    Article  PubMed  PubMed Central  Google Scholar 

  45. Sontag S, Bocova L, Hubens WHG, Nüchtern S, Schnitker M, Look T, et al. Toward clinical application of leukocyte counts based on targeted DNA methylation analysis. Clin Chem. 2022;68(5):646–56.

    Article  PubMed  Google Scholar 

  46. Hubens WHG, Maie T, Schnitker M, Bocova L, Puri D, Wessiepe M, et al. Targeted DNA methylation analysis facilitates leukocyte counts in dried blood samples. Clin Chem. 2023;69(11):1283–94.

    Article  PubMed  Google Scholar 

  47. Cabezón M, Malinverni R, Bargay J, Xicoy B, Marcé S, Garrido A, et al. Different methylation signatures at diagnosis in patients with high-risk myelodysplastic syndromes and secondary acute myeloid leukemia predict azacitidine response and longer survival. Clin Epigenetics. 2021. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13148-021-01002-y.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Kalinkova L, Sevcikova A, Stevurkova V, Fridrichova I, Ciernikova S. Targeting DNA methylation in leukemia, myelodysplastic syndrome, and lymphoma: a potential diagnostic, prognostic, and therapeutic tool. Int J Mol Sci. 2022;24(1):633.

    Article  PubMed  PubMed Central  Google Scholar 

  49. Figueroa ME, Skrabanek L, Li YS, Jiemjit A, Fandy TE, Paietta E, et al. MDS and secondary AML display unique patterns and abundance of aberrant DNA methylation. Blood. 2009;114(16):3448–58.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  50. Wagner W. How to translate DNA methylation biomarkers Into clinical practice. Front Cell Dev Biol. 2022;10: 854797.

    Article  PubMed  PubMed Central  Google Scholar 

  51. Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13:86.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

Open Access funding enabled and organized by Projekt DEAL. This research was supported by funds from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) within CRU344/417911533 “Untangling and Targeting Mechanisms of Myelofibrosis in Myeloproliferative Neoplasms” (W.W., S.K.), 363055819/GRK2415 (W.W.); WA1706/14-1 (W.W.); SFB 1506/1 (W.W.); ZE 432/10-1 (M.Z.); KO2155/7-1 (S.K.), KO2155/7-2 (S.K.), KO2155/9-2 (S.K) and KO2155/6-1 (S.K); and the ForTra gGmbH für Forschungstransfer der Else Kröner-Fresenius-Stiftung (W.W.).

Author information

Authors and Affiliations

  1. Institute for Stem Cell Biology, RWTH Aachen University Medical School, 52074, Aachen, Germany

    Esra Dursun Torlak, Vithurithra Tharmapalan & Wolfgang Wagner

  2. Helmholtz-Institute for Biomedical Engineering, RWTH Aachen University Medical School, 52074, Aachen, Germany

    Esra Dursun Torlak, Vithurithra Tharmapalan, Martin Zenke & Wolfgang Wagner

  3. Department of Hematology, Oncology, Hemostaseology and Stem Cell Transplantation, Medical Faculty of RWTH Aachen University, University Hospital Aachen, 52074, Aachen, Germany

    Kim Kricheldorf, Joelle Schifflers, Madeline Caduc, Martin Zenke & Steffen Koschmieder

  4. Center for Integrated Oncology, Aachen Bonn Cologne Düsseldorf (CIO ABCD), Aachen, Germany

    Kim Kricheldorf, Joelle Schifflers, Madeline Caduc, Martin Zenke, Steffen Koschmieder & Wolfgang Wagner

Authors
  1. Esra Dursun Torlak
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Vithurithra Tharmapalan
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Kim Kricheldorf
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Joelle Schifflers
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Madeline Caduc
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Martin Zenke
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Steffen Koschmieder
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Wolfgang Wagner
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

E.D.T. analyzed DNAm and gene expression data. V.T. performed iPSC experiments and supported analysis of DNAm data. K.K. and J.S. performed clinical data curation, data extraction, and data analysis. M.C. and S.K. conducted treatment and consenting of patients, logistics of bio samples, clinical interpretation, and data analysis. M.Z. provided the JAK2 V617F iPSC lines and support in differentiation. W.W. initiated the research, designed, and supervised the study. V.T., E.D.T., and W.W. wrote the manuscript, and all authors approved the final version.

Corresponding author

Correspondence to Wolfgang Wagner.

Ethics declarations

Ethics approval and consent to participate

All samples were obtained after informed and written consent in accordance with the Declaration of Helsinki and the research was specifically approved by the local ethics committee of RWTH Aachen University (EK 041/15, EK 206/09 and EK 127/12).

Consent for publication

Not applicable.

Competing interests

W.W. and V.T. are involved in the company Cygenia GmbH (www.cygenia.com) that can provide service for epigenetic analysis to other scientists. Apart from this the authors have no competing interests to declare.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Dursun Torlak, E., Tharmapalan, V., Kricheldorf, K. et al. DNA methylation in primary myelofibrosis is partly associated with driver mutations and distinct from other myeloid malignancies. Clin Epigenet 17, 72 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13148-025-01877-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13148-025-01877-1

Keywords

Clinical Epigenetics

ISSN: 1868-7083

Contact us