Subclonal emergence of polycythemia vera, chronic myelomonocytic leukemia, and chronic myeloid leukemia

Abstract

The present longitudinal study reports a unique patient followed over almost three decades who sequentially developed polycythemia vera, chronic myelomonocytic leukemia, and chronic myeloid leukemia. The patient received successive hydroxyurea, ruxolitinib, and a combination of ruxolitinib and nilotinib. The clonal architecture dynamic was reconstructed using targeted high throughput asymmetric capture sequencing, allowing detection and quantification of mutations in 43 myeloid genes and BCR::ABL1 fusion in multiple bone marrow or peripheral blood samples and in single cell-derived colonies obtained from bone marrow colony-forming cell assays. This analysis has uncovered an unexpected subclonal link between three myeloid malignancies, all stemming from a DNMT3A/TET2 double mutant clone. Over a period of more than 30 years, this clone underwent major telomere shortening. However, a striking sustained major molecular response of the terminal dominant clone carrying all driver mutations was achieved by combination therapy with nilotinib and ruxolitinib. The remaining clone driving both polycythemia and chronic myelomonocytic leukemia remained unaffected and evolved to myelofibrosis and proliferative CMML.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00277-025-06417-8.

Introduction

Myeloproliferative (MPNs) and myeloproliferative/myelodysplastic (MPN/MDS) neoplasms are rare blood malignancies, mostly affecting adults and the elderly, that can progress to acute myeloid leukemia. These cancers are characterized by excessive production of mature myeloid cells that can be associated with bone marrow precursor dysplasia for myeloproliferative/myelodysplastic syndromes. MPNs are driven by JAK2, CALR, MPL mutations, or BCR::ABL1 fusion. All these mutations lead to the unregulated activation of proliferative signaling in the myeloid progenitors while persevering their differentiation. These mutations rarely coexist in the same patient. In most co-occurrence cases, these genetic events are present in two competing clones [1]. Their coexistence is exceptional, likely due to their redundant effects on cellular proliferation and survival. MDS/MPN may associate these driver mutations with mutations in epigenetic regulatory genes (mostly TET2, ASXL1, and DNMT3A) and in splicing genes (mostly SRSF2 and SF3B1). Diagnosing these different malignancies, which share numerous cytological and molecular features, may be challenging, especially when several driver mutations are involved. We present here the case of a patient who was initially diagnosed with polycythemia vera (PV) and then chronic myelomonocytic leukemia (CMML), the most frequent MDS/MPN, and finally chronic myeloid leukemia (CML) over two decades.

Case report presentation

A 51-year-old woman (UPN1, Supplementary Table 1) was admitted to Henri Mondor University Hospital in October 1997 due to abdominal pain and vomiting. Her initial blood tests showed hyperleukocytosis (20.3 × 10⁹/L) with neutrophilia (13.3 × 10⁹/L) and monocytosis (5.4 × 10⁹/L), moderate polycythemia (hemoglobin 15.8 g/dL, hematocrit 44%) with microcytosis (mean corpuscular volume 70fL), and thrombocytosis (515 × 10⁹/L) along with hepatic cytolysis (aspartate aminotransferase 942 U/L, alanine aminotransferase 680 U/L). Bone marrow aspiration was hypercellular with erythroid hyperplasia. The karyotype was normal. Ultrasound confirmed hepatomegaly and an accumulation of fluid in the peritoneal cavity. Computed tomography revealed hepatic vein occlusion, and the patient was diagnosed with Budd-Chiari syndrome. She had no previous medical history or known medical conditions. Microcytosis and thrombocytosis were consistent with severe iron deficiency (ferritin = 8.7 µg/L– normal range: 50–120 µg/L). However, hemoglobin levels remained high, which suggested an underlying myeloproliferative neoplasm. This hypothesis was confirmed by the spontaneous erythroid colony formation without erythropoietin, leading to a polycythemia vera (PV) diagnosis. BCR::ABL1 quantification was negative in 1997, and in 2005, a JAK2^V617F mutation was identified (Fig. 1A).

Fig. 1.

Serial development of three hematological malignancies (A) Peripheral blood parameters: hematocrit (%), hemoglobin level (g/dL), white blood cell count (WBC, 10⁹/L), and absolute monocyte count (AMC, 10⁹/L). Normal ranges are indicated with dotted lines (B). UPN1 monocyte subsets were analyzed with Navios cytometer (Beckman Coulter^®) and Kaluza Analysis software (v2.2.1, Beckman Coulter^®). (C) BCR::ABL1 transcripts and JAK2V617F quantifications were done using GeneXpert (Cepheid^®) and QX200 ddPCR (assay dHsaMDV27944642, Biorad^®) or MutaQuant (Qiagen^®) qPCR, respectively. cMo: CD14 + + CD16- classical monocytes; PV: Polycythemia vera

Cytoreductive therapy using hydroxyurea was started in November 1997, followed by intermittent discontinuations due to thrombocytopenia. In 2010, hydroxyurea was stopped following basal cell carcinoma diagnosis and replaced by phlebotomies. While disease progression with spleen enlargement was evident in November 2014, treatment with the anti-JAK1/2 inhibitor ruxolitinib was initiated.

In 2019, hyperleukocytosis increased sharply to 44.2 × 10⁹/L, including immature granulocytes (18%) and persistent monocytosis (4.9 × 10⁹/L) (Fig. 1A). Bone marrow (BM) aspiration analysis and circulating monocyte subset immunophenotyping (Fig. 1B) supported chronic myelomonocytic leukemia (CMML) diagnosis.

In parallel, molecular and cytogenetic analyses were performed due to increased basophilia (4 × 10⁹/L - normal range 0-0.2 × 10⁹/L). A BCR::ABL1 fusion transcript (e14a2) (Fig. 1C) and a Philadelphia chromosome (karyotype: 46,XX, t(9;22)(q34;q11) [22]/46,XX [1]) were detected, resulting in the diagnosis of chronic myeloid leukemia (CML). Nilotinib, a BCR::ABL1 tyrosine kinase inhibitor (TKI), was introduced in addition to ruxolitinib.

Single-cell-derived colonies obtained from BM progenitor cultures at CML diagnosis and six months after nilotinib start were performed. Clonal evolution was deduced from BM (n = 3) or PB samples (n = 4) collected in 1997, 2005, 2018, 2019, and 2020. These analyses identified ten different clones and showed an unexpected Russian doll-like clonal architecture leading to the linear subclonal development of the three hematological malignancies (Fig. 2B and C).

Fig. 2.

A complex clonal architecture (A) Ten clones were identified by NGS sequencing on single-cell derived colonies, peripheral blood, and bone marrow. (B) Clonal hierarchy leading to the emergence of three hematological malignancies. Biallelic inactivation of TET2 in clone Y suggested chronic myelomonocytic leukemia was secondary to polycythemia vera and developed before chronic myeloid leukemia. (C) NGS analyses were performed on single cell-derived colonies (n = 32) at CML diagnosis and six months after nilotinib initiation. Libraries were prepared using an in-house protocol with a pre-pooling capture-based enrichment method (SureSelect XT-HS1 library prep kit, Agilent) and aCAP-seq [2] to quantify BCR::ABL1 fusion, 43 genes, and mutational hotspots. Paired-end sequences were obtained on MiSeq (cartridge V3, 2 × 300 bp, Illumina). NGS sequencing was also performed on total PB (n = 4) and BM samples (n = 3) collected from 1997 to 2020 (dot lines) using the XT-HS1 capture protocol as previously described6. JAK2 polymorphisms (rs2230724, rs10974955, rs3837256, rs10815163 from dbSNP build 156) were used to quantify the JAK2V617F homozygous clone burden. A matrix of the clonal hierarchy evolution was then constructed and plotted using the fishplot R package (v0.5), completed with manual annotations. Due to their low VAF, NRAS clones were randomly plotted on the fishplot. PV: Polycythemia vera. CMML: Chronic myelomonocytic leukemia. CML: Chronic myeloid leukemia

In 1997, the founder clone carried DNMT3A^R771X (39%) and TET2^R1465X (40%) mutations in the peripheral blood (PB), and it was not possible to determine which of these two mutations arose first (Table 1).

Table 1.

NGS monitoring of clonal evolution

	29/10/1997 PV diagnosis	21/10/2005	3/9/2018	28/8/2019 CML diagnosis	6/2/2020	7/5/2020	21/10/2020
DNMT3A Arg771*	39	45	46	47	46	46	47
TET2 Arg1465*	40	46	47	50	46	46	46
JAK2 Val617Phe	27	35	89	95	63	60	56
TET2 4182 + 2T > A	2	33	37	47	44	42	44
PPM1D Leu484*	0	33	41	48	43	44	43
BCR::ABL1	0	0	13	43	0	0	0
NRAS Gly13Asp	0	0	2	0	4	2	1
NRAS Gly12Asp	0	0	0	0	0	1	0
TET2 Gly1370Glu	7	1	6	0	3	3	0

NGS analyses were performed at seven time points from 1997 to 2020. Mutations are indicated as gene p. and VAF (%). BCR::ABL1 was quantified by aCAP-seq at the genomic level

PV: Polycythemia vera. CML: Chronic myeloid leukemia

The founder clone subsequently acquired the JAK2^V617F (27%) and two minor subclones (TET2^G1370E at 7% and TET2c.4182 + 2T > A at 2%), indicating that for this patient, PV was secondary to a TET2^mut/DNMT3A^mut clonal hematopoiesis. Between 1997 and 2005, the TET2c.4182 + 2T > A subclone outcompeted the TET2^G1370E subclone and their parental triple mutant clone (DNMT3A/TET2/JAK2). In silico prediction by SpliceAI lookup (500) indicated that this variant likely induces a donor splice site loss (DL score = 1.0), leading to intron retention. Together with TET2^R1465X, this second event may lead to a biallelic inactivation of TET2 and to a nonfunctional protein. It later acquired an additional PPM1D^L484X in the context of hydroxyurea treatment that may have contributed to its selection [3, 4]. Interestingly, biallelic inactivation of TET2 in the dominant clone was associated with a progressive rise in the absolute monocyte count (AMC) that later remained almost constantly above 1 × 10⁹ /L, indicating that CMML developed between 1997 and 2005 and was secondary to PV (Fig. 1B). Biallelic mutations in TET2 [5–7], chronic monocytosis above 1 × 10⁹/L for several years, and circulating cMo > 94% [8] altogether supported the diagnosis of CMML in 2019. Single cell-derived colony sequencing performed at CML diagnosis in 2019 showed that the CMML driving clone later became homozygous for JAK2^V617F and then acquired BCR::ABL1 fusion, leading to CML. While no RNA samples were available to backtrack the occurrence of BCR::ABL1, it was detected by aCAP-seq with a VAF at 13% almost one year before diagnosis (Table 1).

In this study, we didn’t analyze copy number variations that could have contributed to disease development by CGH array or SNP array. However, allelic frequency analysis of JAK2 polymorphisms (rs2230724, rs10974955, rs3837256, rs10815163 from dbSNP build 156) indicated that JAK2^V617F+/+ resulted from a copy-neutral loss of heterozygosity.

As expected, ruxolitinib efficiently reduced the patient’s spleen size between 2015 and 2019 until CML diagnosis, as previously described in PMF [9], PV [10], and, more recently, in CMML [11, 12]. However, ruxolitinib failed to eradicate the JAK2^V617F clone since the mutation burden kept increasing, ultimately evolving to homozygosity. Notably, two additional RAS pathway mutations (NRAS^G13D and NRAS^G12D) found at low frequency appeared upon ruxolitinib treatment, while this observation was also recently reported in larger MPN cohorts [13, 14].

CML arose upon ruxolitinib therapy, indicating that this drug was not efficient alone in preventing CML development. However, when nilotinib was introduced together with ruxolitinib, clonal analysis and molecular monitoring of BCR::ABL1 by qRT-PCR (Figs. 1C and 2C) demonstrated an efficient clearing of the highly mutated dominant CML clone carrying DNMT3A^R771X, TET2^R1465X, JAK2^V617F+/+, TET2c.4182 + 2T > A, PPM1D^L484X, and BCR::ABL1. %(BCR::ABL1/ABL1)^IS was quantified at 1.1% and 0.18% at three and seven months following nilotinib treatment start. Remarkably, major molecular response (MMR) was evident at 15 months (data not available at 12 months), and it has remained stable for three years (last available quantification: %(BCR::ABL1/ABL1)^IS=0.018% on August 2023). The nilotinib and ruxolitinib combination succeeded in establishing a durable major molecular response, while the founder clone drove post-PV myelofibrosis evolution together with proliferative CMML (WBC = 41.1 × 10⁹/L, AMC = 11.1 × 10⁹/L, on August 2023).

Discussion and conclusions

This observation contrasts with previous studies reporting that additional mutations at CML diagnosis were associated with an increased risk of relapse, a delayed MMR, and an increased risk of progression to blast crisis when treated with TKI alone [15–18].

BCR::ABL1 and Ph⁻ MPN driver mutation co-existence has been reported and reviewed [1]. In 49.2% of CML and Ph⁻ MPN co-occurrences reported, CML diagnosis was posterior to Ph⁻ MPN with either JAK2^V617F or CALR mutations. This meta-analysis showed that in most cases, MPN driver mutation allelic frequency and BCR::ABL1 ratio had opposite growth kinetics following TKI introduction, indicative of the coexistence of two independent clones. In these cases, most CML patients responded to TKI. However, in the patient, BCR::ABL1 and JAK2^V617F mutations are present in the same clone, and dual targeting was beneficial, as also demonstrated in vitro and in vivo on CML CD34⁺ cells when a TKI is combined with ruxolitinib [19] or pimozide [20], a STAT5 inhibitor [21].

Altogether, the present report and previous results of in vitro experiments and early-phase clinical trials [19, 20, 22, 23] support future evaluation of the TKI plus ruxolitinib combination in CML patients harboring additional mutations at diagnosis, including JAK2^V617F, and also mutations in epigenetic regulators such as DNMT3A, TET2 or ASXL1.

Finally, the successive acquisition of three myeloid malignancies associated with high clonal turnover prompted us to evaluate the evolution of the telomere length (TL) in the patient’s hematopoietic cells over three decades. Samples from five other MPN patients with similar ages and high clonal burden in the peripheral blood were serially analyzed over a decade (median = 146 months [range: 124–201 months] together with healthy individuals at similar ages (Fig. 3). The patient’s telomere length was highly reduced by 77% (12.7 kb to 2.9 kb) over 23 years, corresponding to a slope of -0.40 kb/year, while the slope for healthy individuals was − 0.176 kb/year. The calculated slopes for the other MPN patients were heterogeneous (Supplementary Table 1). Two patients showed telomere shortening comparable to healthy individuals (-0.203 kb/year for UPN2 and 0.137 kb/year for UPN3). One PMF patient had low telomere length (2.76 kb) already at diagnosis and a TL decrease slope of -0.056 kb/year (UPN4). Finally, two MPN patients showed a higher rate of TL shortening than healthy individuals, with slopes respectively found at 0.314 kb/year (UPN5) and 0.396 kb/year (UPN6). The high rate of telomere shortening of the index case provides a rationale for telomere synthesis targeting using telomerase inhibitors such as Imetelstat evaluated in MPNs [24] and MDS [25].

Fig. 3.

A reduced telomere length Telomere length per chromatid measurements of healthy individuals (open circles and black line for linear regression fitting line), UPN1 (red), UPN2 (blue), UPN3 (green), UPN4 (brown), UPN5 (orange), and UPN6 (pink). Linear regression with 95% confidence interval (dot lines) and slopes were calculated with Prism v10.2.3 (GraphPad^®). Relative telomere lengths were measured by qPCR, and then absolute values were calculated using a reference human genomic DNA sample with known telomere length (Absolute Human Telomere Length Quantification qPCR Assay Kit, ScienCell^®). CML: chronic myeloid leukemia; CMML: Chronic myelomonocytic leukemia; cMo: CD14 + + CD16- classical monocytes; PV: Polycythemia vera

In conclusion, acquiring PV, CMML, and CML mutant drivers in the same clone can phenocopy bio-clinical features in a patient. However, identifying such clonal evolution may represent a significant clinical challenge, given the clinical and biological overlap between these entities. In the present case, CMML-initiating mutations could be detected years before CMML diagnosis and CML. Using a high throughput sequencing panel, including the detection of both driven mutations and fusion genes such as BCR::ABL1 by aCAP-Seq in patient follow-up, may thus be helpful to identify earlier these types of clonal evolution.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

CML

Chronic myeloid leukemia

CMML

Chronic myelomonocytic leukemia

MDS

Myelodysplastic syndrome

MPN

Myeloproliferative neoplasm

PMF

Primary myelofibrosis

TKI

Tyrosine kinase inhibitors

TL

Telomere length

Author contributions

VTQ performed NGS analysis with help from RL, BBJ, QB, and CJ. JC performed telomere length analysis. OWB performed flow cytometry analysis with help from ST. PE, CP, and LR provided clinical and biological data. IS and GG performed bioinformatics analysis. IS designed the research. IS and VTQ wrote the manuscript with input from LR, OWB, JC, and ST.

Funding

The authors received no grant for this case report.

Data availability

The data and information on materials related to this case report are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This case report complied with French regulations and was approved by the Henri Mondor. Institutional Review Board (No. 00011558). The study methodologies conformed to the standards set by the Declaration of Helsinki. All patient data were anonymized and de-identified before analysis, and informed consent was obtained from all patients.

Consent for publication

All authors have agreed with the content of the manuscript for publication.

Competing interests

IS is a speaker for Novartis and Incyte. IS is a co-inventor of a know-how licensed by AP-HP to Agilent Technologies. OWB has a patent issued relevant to this work. Other authors declare no conflict of interest.