Introduction

Essential thrombocythemia (ET) is a chronic myeloproliferative neoplasm (MPN) characterized by megakaryocyte hyperplasia, thrombocytosis, as well as thrombotic and hemorrhagic complications1,2,3. Although the median survival of ET patients is approximately 18 years, ET has the potential to transform into acute myeloid leukemia (AML)4,5. Since patients after transformation to AML have a dismal prognosis, it is necessary to clarify the biological mechanism for applying suitable therapeutic options.

Approximately 80% of ET patients have a mutation in one of the JAK2, CALR, or MPL gene, while these mutations are not associated with the risk of AML transformation1,2,6. Furthermore, up to 20% of ET patients are negative for all three mutations, referred to as triple-negative ET. It has been demonstrated that several additional genetic alterations, which are engaged in epigenetic regulation, cell-growth signaling, and RNA splicing machinery, cooperate with JAK2, CALR, or MPL mutation for the transformation to AML from ET7,8,9. The order of mutation acquisition also influences clonal evolution in MPNs, while it is not fully understood how clonal evolution occurs during disease progression in ET patients7,9,10.

On the other hand, both JAK2 mutated- and unmutated-AML clones have been identified in transformed AML cells from JAK2-mutated MPNs, suggesting the presence of an initiating clone without JAK2 mutation for ET and transformed AML7,9,11. Although several initiating mutations, such as TET2 and DNMT3A mutations, have been identified in triple-negative ET patients, initiating clones without JAK2, CALR, and MPL mutations are not fully known. Moreover, the clonal evolution step from the initiating clone without JAK2, CALR, and MPL mutations is poorly understood.

The latent period of transformation to AML from ET is so long that it is difficult to compare the genetic status between transformed AML and ET at the initial diagnosis. In this study, we analyzed paired samples at the initial diagnosis and transformation to AML in eight ET patients, and identified three patients in whom JAK2-unmutated AML clones developed from JAK2-mutated ET in the chronic phase. We further investigated to identify initiating clones without JAK2 mutation, and clarified the clonal evolution process during disease progression.

Methods

Patients and samples

Clinical characteristics of eight ET patients are shown in Table1 and Supplementary Table 3. The median duration from the diagnosis of ET until transformation to AML was 10.0 years (range: 1.5 – 20.0 years). Bone marrow (BM) or peripheral blood (PB) samples were obtained from each patient both at the ET phase and AML phase, and they were cryopreserved before use except for UPN2. At the initial diagnosis of UPN2, only DNA was available. From UPN8, BM cells at complete remission (CR) after chemotherapy for transformed AML and at relapse after achieving CR as well as a buccal swab were obtained. BM mononuclear cells at CR after chemotherapy for transformed AML were incubated with following antibodies before being sorted into hematopoietic stem cell (HSC) and hematopoietic progenitor cell (HPCs) fractions by a flow cytometer (FACSAria, BD Biosciences, San Jose, CA, USA): Human Lineage Cocktail 4 (CD2, CD3, CD4, CD7, CD8, CD10, CD11b, CD14, CD19, CD20, CD56, CD235a), anti-human CD34-APC (8G12), anti-human CD38-PE/Cy7 (HB7) (BD Biosciences), anti-human CD45RA-PerCP/Cy5.5 (HI100), anti-human CD123-PE (6H6) (BioLegend, San Diego, CA, USA) antibodies. HSCs were defined as lineage-marker (Lin)-CD34+CD38-fraction, HPCs were defined as Lin-CD34+CD38+CD123+CD45RA- (common myeloid progenitor, CMP), Lin-CD34+CD38+CD123+CD45RA+ (granulocyte–macrophage progenitor, GMP), and Lin-CD34+CD38+CD123-CD45RA- (megakaryocytic/erythroid progenitor, MEP) fractions.

Table 1 Patients’ characteristics.
Full size table

In addition, we collected BM or peripheral blood mononuclear cells from 34 patients with ET in the chronic phase.

High-molecular-weight DNA and total RNA were extracted from each sample using QIAamp DNA Blood Mini Kit (QIAGEN, Hilden, Germany), QIAamp DNA Investigator Kit or QIAamp RNA Blood Mini Kit, and subjected to further analysis.

Cytogenetic and molecular analyses

Cytogenetic G-banding analysis was performed using standard methods. Chimeric gene transcripts (Major BCR::ABL1, Minor BCR::ABL1, PML::RARA, RUNX1::RUNX1T1, CBFB::MYH11, DEK::NUP214, NUP98::HOXA9, MLLT1::KMT2A, MLLT2::KMT2A, MLLT3::KMT2A, and MLLT4::KMT2A) were examined at transformation of AML in all patients, as previously reported12.

Target sequencing of 54 genes, which are frequently identified in the presence of myeloid malignancies, was performed using TruSight Myeloid Sequencing Panel and Illumina MiSeq sequencer according to the manufacturer’s instructions (Illumina, San Diego, CA, USA) (Supplementary Table 1)13,14,15 in paired ET and AML samples of UPN1-8, CR and relapsed samples of UPN8. Sequence variation annotation was performed using known polymorphism databases, followed by mutation characterization, as previously reported15. Whole-exome sequencing (WES) was also performed on ET and transformed AML cells as well as a buccal swab in UPN8, as previously reported16. The variants detected in ET and/or AML samples but not in buccal swab were selected for further analysis. Each predicted variant sequence was confirmed by Sanger sequencing. The pathogenicity of variants were predicted with 5 pathogenicity tools: FATHMM17, LRT18, MutationTaster19, PolyPhen-220 and SIFT21.

Copy-number abnormalities were identified using CNACS pipeline (https://github.com/OgawaLabTumPath/CNACS); the data of allele frequencies and sequenced depth of SNPs were used as input data22.

Mutations in whole coding regions of ZNF143, UBR4, and SMARCC2 genes were analyzed in 40 patients with ET including UPN1-5 and UPN7. Whole exon regions of ZNF143, UBR4, and SMARCC2 genes were amplified by the primer pairs indicated in Supplementary Table 2. Amplified products were subjected to mutation analysis using Nextera XT DNA Library Prep Kit and MiSeq sequencer according to the manufacturer’s instructions (Illumina).

Single-cell mutation analysis

Lineage-/CD34+/CD38- BM cells at CR after chemotherapy for transformed AML in UPN8 were sorted as single cells in a well of a 96-well plate. All processed 96 wells contained a single cell as verified by visual inspection under microscope.

Genomic DNA was extracted from each cell, and subsequently amplified by REPLI-g mini Kit (QIAGEN, Hilden, Germany). The DNA was analyzed for six mutations; JAK2V617F in exon 14, TP53R248W, TP53V173L, SMARCC2D381E, UBR4R450H and ZNF143S286R, and also a SNP within the JAK2 c.2490G > A in exon 19, by direct Sanger sequencing of a PCR amplified region surrounding the target site with primer pairs indicated in Supplementary Table 2.

Patient-derived xenograft model

Mononuclear cells isolated from fresh BM samples of UPN1 and UPN8 at transformation to AML were intravenously injected into 6-week-old NOD/Shi-scid, IL-2Rγnull (NOG) mice (purchased from the Central Institute for Experimental Animals, Tokyo, Japan) at 1 × 107 viable cells per mouse, as previously reported15. T cells from patient BM samples were depleted by intraperitoneally injecting an anti-human CD3 (OKT3) antibody (Exbio Antibodies, Prague, Czech Republic). NOG mice were not pre-conditioned with irradiation. The engraftment of primary AML cells was monitored every 3 weeks in PB from the tail vein followed by flow cytometric analyses using FACSAria2 with anti-mouse CD45-PerCP (30-F11) (BioLegend), anti-human CD3-APC (UCHT1), and anti-human CD45-PE (HI30) antibodies (BD Biosciences). Mice were sacrificed when PB human CD45+ reached > 0.5% at 2 time-points, followed by flow cytometric assessment of BM for engrafted human cells with the same antibodies. The human CD45+ fraction was sorted from PDX BM by magnetic cell separation using MACS MicroBeads (human CD45 MicroBeads; Miltenyi Biotec, Bergisch Gladbach, Germany). Genomic DNA was extracted from unfractionated BM sample and subjected to targeted deep sequencing for 54 genes related to myeloid malignancies and SMARCC2, UBR4, and ZNF143 as described in a previous section.

Ethics approval

Informed consent for banking and further studies including genetic analysis of samples was obtained from all patients, and approval was obtained from the ethics committees of all participating institutions according to the Declaration of Helsinki. All methods were performed in accordance with relevant regulations and guidelines. All animal procedures were approved by the Institutional Animal Care and Use Committee of Nagoya University and carried out in accordance with the Regulations on Animal Experiments of Nagoya University and the ARRIVE guidelines.

Results

Genetic analysis in paired samples

Cytogenetic and RT-PCR analyses confirmed that no patients’ samples had t(9;22)(q34;q11)/BCR::ABL1 abnormality. Mutation statuses of the patients at the initial diagnosis of ET and AML transformation are shown in Table 2. At ET, two patients (UPN1 and UPN2) showed CALR mutations; MPLS505N mutations were also identified in one patient (UPN2), while both wild-type and mutant MPL mRNA were not expressed as previously reported23 Another patient (UPN3) had MPLW515L and other five patients (UPN4-8) had JAK2V617F mutation at ET. These driver mutations were detected at VAFs 18.4% to 75.1%. Additional mutations were identified at ET in seven out of eight patients: TET2 or TP53 mutations, three patients; ASXL1 mutations, two patients; EZH2 or PHF6 mutation, one patient. These additional mutations were detected at a variety of VAFs (3.0—43.4%), most which were lower than VAFs of driver mutations. A part of ET clones with CALR, JAK2 or MPL mutation or other clones without these driver mutations seemed to have additional mutations.

Table 2 Mutation status in ET and transformed AML cells.
Full size table

At AML transformation, five patients (UPN1-5) had the same CALR, MPL, or JAK2 mutation as that of ET along with new additional mutations: TP53, NRAS, U2AF1, RUNX1, or CEBPA mutation. These results indicated that ET and transformed AML cells developed from the same CALR, MPL, or JAK2-mutated initiating clone acquiring additional mutations in UPN1-5 (Fig. 1).

Figure 1
figure 1

Serial mutational spectrum from ET to transformed AML. The serial mutational spectrum of each patient in the clinical course from ET to transformed AML is shown in fish plot format. Although five patients (UPN1-5) had the same CALR, MPL, or JAK2 mutation as that of ET at AML transformation, three patients (UPN6-8) in whom the JAK2V617F-mutated clone was dominant at ET showed that the dominant clones at AML transformation did not have JAK2V617F mutation.

Full size image

Notably, three patients (UPN6-8) in whom the JAK2V617F-mutated clone was dominant at ET showed that the dominant clones at AML transformation did not have the JAK2V617F mutation. In two patients with JAK2V617F and TET2 mutation(s) at ET (UPN6 and UPN7), JAK2V617F mutation was not detected or detected at low VAF (9.9%) in transformed AML cells; instead VAFs of TET2 mutation were detected at almost same or increased level compared with ET. Furthermore, TP53, IDH2, NRAS, SRSF2 and CBL mutations were additionally identified in transformed AML cells. These results indicated that TET2-mutated, but not JAK2-mutated, clones were the common initiating clones in UPN6 and UPN7 (Fig. 1). However, in transformed AML of UPN8, VAF of TP53R248W mutation increased to 87.5 from 4.7%, while JAK2V617F mutation decreased to 2.7 from 40.5% and TP53V173L mutation was stable at VAF below 5%; no other mutations common to both ET and AML were detected. UPN8 at AML transformation had acquired a complex karyotype including numerical or structural abnormalities of chromosome 5 and 7, which is supposed to be involved in the transformation (Table 1).

TP53 mutations were detected seven of eight patients at AML with a VAF 37.4% in one patient whose blast rate was 10% in the sample and over 80% in other patients. The deletion of chromosome 17 resulting in the loss of heterozygosity (LOH) at TP53 locus were detected in six of these patients (Table 1). TP53 mutations were already detected in ET samples in three patients using target sequencing with 1.0% as cutoff VAFs for mutations; whereas the possibility that the minor clones with TP53 mutations at VAF lower than 1.0% had existed in other four patients at ET has not been ruled out.

We showed a model of clonal changes from ET to AML in each patient assumed by VAFs of analyzed genes (Fig. 1); other models can be also drafted. Therefore, a single cell analysis was performed to identify detailed clonal dynamics in UPN8 as described in a later section.

Search for initiating mutations by Whole-exome sequence

The UPN8 patient received chemotherapy after AML transformation, and achieved complete remission (CR); however, the patient subsequently showed relapsed AML. We also analyzed the mutation status at CR and relapse using BM samples. Mutation statuses of JAK2V617F, TP53R248W, and TP53V173L in BM samples at CR and relapse were almost the same as those at the initial diagnosis of ET and AML transformation, respectively (Table 3). These results suggest that ET clones with JAK2V617F were dominant at CR and AML clones with TP53R248W re-increased at relapse, and moreover that the initiating clone with genetic mutations other than these mutations could exist. Therefore, we performed WES analysis to search for initiating mutations in samples at ET or AML phase in UPN8.

Table 3 Mutation status during the disease course in UPN8.
Full size table

WES analysis revealed that ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations were commonly identified in both ET and transformed AML cells (Fig. 2a). These mutations were not identified in the buccal swab sample indicating that these were somatic mutations (Supplementary Fig. 1a). These mutations were classified as pathogenic in most of the in-silico prediction tools (Supplementary Table 4). We also interrogate The Cancer Genome Atlas (TCGA) database, which comprises 10,967 samples across 32 different cancer types24 for exploring the possibility that these mutations are single nucleotide polymorphisms. UBR4R450H mutation had been detected in one patient with endothelial cancer as single nucleotide variation with unknown significance and neither of ZNF143S286R or SMARCC2D381E mutation had been detected. WES showed VAFs of ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations at ET and AML were 47.9 and 60.6%, 47.7 and 48.5%, and 50.7 and 47.9%, respectively. Furthermore, VAFs of ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations at CR and relapse were similar to those at ET or AML phase (Table 3). These results suggest that cells harboring these three somatic mutations can be common ancestors of both ET and AML.

Figure 2
figure 2

Clonal evolution in UPN8. (A) Comparison of VAFs of mutated genes detected by whole-exome sequence analysis between ET and transformed AML cells. ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations were identified at the same VAF levels in ET and transformed AML cells. (B) Mutation analysis in HSC and HPCs. Mutation statuses of JAK2V617F, TP53R248W, ZNF143S286R, UBR4R450H, and SMARCC2D381E in HSC, CMP, GMP, and MEP fractions were analyzed. (C) Single-cell mutation analysis of BM samples at CR after chemotherapy for transformed AML. CD34+/CD38- cells were sorted as single cells, and we analyzed JAK2V617F, TP53R248W, TP53V173L, ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations. The closed square and triangle indicate homozygous and heterozygous mutations, respectively.

Full size image

Mutation analysis in HSC and HPC fractions

We then analyzed the mutation status in HSC and HPCs (CMP, GMP, and MEP) fractions to clarify the pattern of mutational acquisition throughout the ET- and AML-phase in UPN8. We used the BM sample at CR after chemotherapy for transformed AML, which was thought to be composed of ET-, AML- and normal cells. JAK2V617F, ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations were identified in HSC and HPCs fractions (Fig. 2b) at similar VAFs to whole mononuclear cells (MNC) (Table 4), suggesting that ET clones had already acquired these mutations at hematopoietic stem cell stage and differentiated to progenitor cells. In contrast, TP53R248W mutation was identified in the HSC fraction at a higher VAF compared to MNC and HPCs fraction. Given that TP53R248W mutation was detected in the AML sample at a much higher VAF compared to ET sample, these results suggest that AML clones, which were concentrated into the HSC fraction, did have acquired TP53R248W. The failure to detect TP53R248W mutation in the CMP fraction may be attributed to an inability of the current methods rather than to an actual absence of this mutation considering that the mutation was detected in both GMP and MEP fractions.

Table 4 Mutational status in HSC, HPC fractions at complete remission in UPN8.
Full size table

Single-cell mutation analysis

We further explored the common initiating clone for ET and AML in UPN8 using single-cell mutation analysis. The CD34+/CD38- cells in BM samples at CR after chemotherapy for transformed AML were sorted as single cells and analyzed for JAK2V617F, TP53R248W, TP53V173L, ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations. We analyzed 96 single-cells and obtained results in 26 single-cells (Fig. 2c). Contrary to expectations from the VAF in the target sequencing and WES, JAK2V617F mutation was identified only in 13 of the 26 (50.0%) cells, and 11 of them were homozygous. We investigated whether the cells with JAK2V617F in exon 14 had LOH on JAK2 locus even though copy number change or imbalance was not observed on JAK2 locus (Supplementary Fig. 1b). Analysis of the SNP within the JAK2 exon 19, c.2490G > A, revealed that most of cells with homozygous JAK2V617F had only c.2490G allele at the SNP site (2–6 and 11–13 in Fig. 1c) , most of cells with wild type JAK2 had c.2490A allele (14–15, 17–21 and 25 in Fig. 1c), and all two cells with heterozygous JAK2V617F (1 and 10 in Fig. 1c) and some cells with homozygous JAK2V617F or JAK2WT had heterozygous SNP (7–9 or 22–25 in Fig. 1c). These data support the view that homozygous JAK2V617F was caused by mitotic recombination leading to acquired uniparental disomy (UPD) on chromosome 9p without copy number change on JAK2 locus, which had been reported in MPNs25. The patterns of the JAK2 SNP in some cells, e.g. clone 16 in Fig. 1c, were not compatible with the model. None of the cells with TP53R248W mutation harbored JAK2V617F mutation, which suggests AML clones with TP53R248W mutation are derived from cells without JAK2V617F mutation as expected from the VAFs in the bulk sequencing analyses. ZNF143S286R, UBR4R450H and SMARCC2D381E mutations were identified both in JAK2V617F-mutated and -nonmutated cells and also in cells with TP53R248W mutation but not in the cell with TP53V173L. These results suggest that ET clone with JAK2V617F and AML clone with TP53R248W could be derived from the common initial clone with ZNF143S286R, UBR4R450H, and SMARCC2D381E mutations, and clones with TP53V173L could be derived from another clone without these three mutations. However, the mutational pattern of ZNF143S286R, UBR4R450H, and SMARCC2D381E was diverse both in JAK2V617F-mutated and – unmutated cells; only a part of the JAK2V617F-mutated cells (nine of the 13) and un-mutated cells (11 of 13) were co-mutated with ZNF143S286R UBR4R450H, or SMARCC2D381E mutation, and the numbers of co-mutations also varied between cells (Fig. 2c).

Engrafted clone in patient-derived xenograft model

Our previous study had shown that AML-PDX models are useful for analyzing the clonal dynamics and that chemotherapy-resistant clones dominantly engraft in AML-PDX models even when they are minor in primary AML15. AML transformed from MPNs are often refractory to chemotherapy, and serial mutational analysis in this study has shown that some transformed AML samples contain multiple clones. We then tried to identify the clones which effectively graft and propagate in the PDX model. We inoculated NOG mice with transformed AML cells of three patients (UPN1, -2 and -8) and established two AML-PDX models from UPN1 and -8. (Table 5). In UPN1, engrafted cells harbored CALRK385fs47, TP53C238S and U2AF1Q157R mutations, but not ASXL1G643fs mutation. Furthermore, VAFs of CALRK385fs47 (40.2%) and U2AF1Q157R (49.5%) mutations in the engrafted cells were the same as primary AML cells, while that of TP53 mutation increased to 99.9 from 37.4% in primary AML cells. These results suggest that a major clone in AML with CALRK385fs47, TP53C238S and U2AF1Q157R with TP53 LOH which is attributed to deletion of chromosome 17 engrafted and propagated in PDX. In UPN8, the engrafted cells harbored TP53R248W (VAF, 98.7%), ZNF143S286R (VAF, 52.2%), UBR4R450H (VAF, 50.4%), and SMARCC2D381E (VAF, 44.8%) mutations at the same VAFs as the primary AML cells, but not JAK2V617F or TP53V173L mutations. These results suggest that a major clone in AML with TP53R248W, ZNF143S286R, UBR4R450H, and SMARCC2D381E engrafted in PDX.

Table 5 Comparison of mutation status between AML and PDX cells.
Full size table

Screening of ZNF143, UBR4, and SMARCC2 mutations in ET or other myeloid neoplasms

Lastly, we screened whole coding regions of ZNF143, UBR4, and SMARCC2 genes in 40 patients with ET including UPN1-5 and UPN7. However, we did not find mutations in any of the patients. An additional interrogation of Beat AML database(n = 672)26 and data from other 10 studies targeting myeloid neoplasms (n = 9,889)27,28,29,30,31,32 using cBioPortal33,34 revealed that ZNF143S286R mutation had been detected in one patient with ET as a somatic mutation32. Neither UBR4R450H nor SMARCC2D381E had been detected in the database.

Discussion

We found that five of the eight patients showed the proliferation of JAK2, CALR, or MPL-mutated clones with additional mutations at AML transformation. Particularly, TP53 co-mutated clones dominantly proliferated in four of the five patients. Although the transformed AML cells did not harbor TP53 mutation in one patient (UPN5), they had additional mutations in TET2, ASXL1, RUNX1, and CEBPA genes. These results indicated that the additional mutations, particularly TP53 mutation, drove the JAK2, CALR, or MPL-mutated clone in the chronic phase of ET to AML transformation.

In contrast, the JAK2, CALR, or MPL-unmutated clone proliferated at AML transformation in three patients (UPN6-8). Notably, proliferated clones in transformed AML already existed at the initial diagnosis of ET in all patients. In UPN6 and 7, TET2-mutated clones were identified both at the initial diagnosis and AML transformation, while TP53, IDH2, and NRAS mutations in UPN6 and TP53, SRSF2, and CBL mutations in UPN7, which were additionally identified in transformed AML, were not detected at the initial diagnosis. These results indicated that TET2-mutated, but not JAK2-mutated, clones were common initiating clones for ET and transformed AML. Interestingly, VAF of NRASG12S mutation (6.4%) was much lower than the other mutations at AML transformation in UPN6. The NRAS co-mutated minor clone was also observed in transformed AML cells of UPN2. We reported that the Marimo cell line, which harbors CALRL367fs*43, MPLS505N, TP53C153Y, and NRASQ61K mutations, was established from the NRAS co-mutated clone in UPN223,35. These results support the suggestion that NRAS mutation provided further growth advantage to the transformed AML clone even with TP53 mutation.

In UPN8, SMARCC2, UBR4, and ZNF143 mutations as well as JAK2 and TP53 mutations were identified at the ET phase, while SMARCC2, UBR4, ZNF143, and TP53, but not JAK2, -mutated clones proliferated at transformation to AML. Since VAF of TP53 mutation increased to 87.5% in transformed AML cells, transition from heterozygosity to homozygosity in TP53 mutation might be associated with evolution to AML, as previously reported7,8,36 However, the effect of the pathophysiology of SMARCC2, UBR4, and ZNF143 mutations on the development and progression of ET is, to date, unclear. Each VAF of identified gene mutation was almost the same among HSC and HPC fractions in the CR state after chemotherapy for transformed AML. Furthermore, VAF of JAK2 mutation was the same as that of SMARCC2, UBR4, or ZNF143 mutation, and that of TP53 mutation was lower than the other mutations. Mutation analysis in the single cells sorted from the CD34+/CD38- fraction revealed that both JAK2V617F and JAK2 wild-type cells including TP53R248W-mutated cells had ZNF143, UBR4, and SMARCC2 mutation. These results suggest that ET clone and AML clone could be derived from the common initial clone harboring ZNF143, UBR4, and SMARCC2 mutations, although their biological significance is unclear. The further analysis of the SNP in JAK2 leaded us to surmise that most of cells in the fraction at CR had UPD on chromosome 9p resulting in homozygous JAK2 mutation with LOH. One model for the clonal change in UPN8 can be as follows. A part of clone with ZNF143, UBR4, and SMARCC2 gained JAK2V617F, underwent mitotic recombination on chromosome 9p and propagated as ET clones; and another clone harboring the three mutations gained TP53R248W in ET phase and evolved into AML clones. On the other hand, this model is not applicable to all cells. Some cells JAK2V617F had none of these three mutations and mutational pattern including zygosity of each variant varied between cells. This is a limitation of the single-cell mutation analysis by Sanger sequencing in this study, which warrant consideration.

PDX-model analysis of AML cells from UPN1 and UPN8 clarified that the transformation-associated clone had a growth advantage. In a NOG mouse inoculated with AML cells of UPN8, engrafted AML cells consisted of ZNF143, SMARCC2, UBR4, and TP53-mutated clones, but not JAK2-mutated clones. ZNF143, SMARCC2, and UBR4 mutations were not identified in 40 ET patients in this study; however, ZNF143S286R mutation has been reported in ET patients32,37. It is possible that these mutations are cooperatively involved in the mechanisms of disease initiation and evolution in this patient based on their known biological functions. Further study is required to clarify the biological mechanism of these mutations in the pathophysiology of ET.