Polycythemia vera (PV) is a JAK2-mutated myeloproliferative neoplasm (MPN) typified by clonal erythrocytosis. Along with primary myelofibrosis (PMF) and essential thrombocythemia (ET), PV belongs to the classic Philadelphia-negative MPN. PV can lead to significant morbidity and increased mortality, with patients experiencing a reduced quality of life, a high rate of vascular events, and the potential intrinsic risk of disease evolution. Mutations in JAK2 Exon 12-15 encompass the vast majority (97%–98%) of PV mutational landscape (1). In addition, over 50% of patients harbor non-driver mutations in specific myeloid genes with the most frequent being TET2 (18%), ASXL1 (15%), and SH2B3 (3%) (2).

Approximately 15% of patients with PV experience a fibrotic progression of their disease over time, termed post-PV myelofibrosis (post-PV-MF) (2) and about 3% evolve, over a period of 10 years, into accelerated (MPN-AP) and blast phase (MPN-BP) disease (3), operationally defined by 10% to 19% or ≥ 20% myeloid blasts in the peripheral blood or bone marrow, respectively (4). However, despite this formal classification, MPN patients with 5% or more circulating/marrow blasts can be on a spectrum that leads to MPN-BP, with similar poor survival rates (5).

Clinical and molecular risk factors for leukemic evolution have been identified by the Mayo-Florence study conducted on 410 patients with MPN-BP, with 97 patients whose condition progressed from an antecedent PV, either directly (46 patients) or through a post-PV-MF stage (51 patients). Hematological risk factors are leukocytosis ≥11 × 10⁹/uL (Florence cohort) or ≥15 × 10⁹/uL (Mayo cohort), older age (> 67 years, both cohorts), abnormal karyotype (Mayo cohort), presence of IDH2 and RUNX1 mutations (Mayo cohort), and SRSF2 mutations (Florence cohort) (3).

On these bases, investigators generated a mutation-enhanced international prognostic system for chronic phase disease (MIPSS-PV), which included presence of adverse mutations (SRSF2) (two points); abnormal karyotype (one point); age >67 years (three points); and leukocyte count ≥15 × 10⁹/uL (two points) (6). Additionally, it is well established that previous treatment with radiophosphorous, chlorambucil, or pipobroman increases the risk of leukemic transformation (3).

The evolution to AP and BP is an almost invariably fatal event in the course of PV (and MPN in general), with a median overall survival (OS) of 13 months for AP and 3–5 months for BP, due to disease aggressiveness and limited therapeutic options, especially for transplant ineligible patients (4). The lack of standardized treatment(s) mirrors our still fragmented understanding of the molecular mechanisms underlying the progression of an MPN to AP/BP. It is currently acknowledged that the acquisition of additional somatic mutations and epigenetic alterations in hematopoietic progenitor and stem cells play a pivotal role in leukemic transformation, accounting for the selection and expansion of the malignant clone on one side, and on the other side, the generation of a tumor supportive proinflammatory microenvironment. Two main models of leukemic transformation have been proposed: (i) the acquisition of additional mutations by the JAK2/CALR/MPL-positive clone (7) and (ii) the emergence and expansion of a leukemic clone distinct from the clone bearing an activating driver mutation, eventually overcoming the MPN JAK2/CALR/MPL-mutated clone (7). In the first setting, the most common genetic alteration is represented by TP53 mutations, detected in approximately 15% of patients with JAK2V617F (showing cooperative activity) and 25% of patients with CALR mutations. In the second scenario, a decrease in variant allele frequency (VAF) or loss of pre-existing driver mutation is usually observed (4). In addition to these two scenarios, another possible mechanism has been suggested, focused on the timing of occurrence of mutations involving chromatin modifiers, spliceosome complex components, DNA methylation modifiers, tumor suppressors, and transcriptional regulators. These mutations are described as coexisting mutations since their presence within a clone can precede the acquisition of the driver mutation or can occur subsequently. Grabek et al. (8) suggested that specific coexisting mutations in the same driver-mutated clone may favor direct leukemic transformation while others may drive progression to myelofibrosis which in turn may evolve into AP/BP.

Independent from the molecular pathway leading to MPN-AP/BP, it is well established that the mutational profile and overall prognosis of acute myeloid leukemia (AML) resulting from an antecedent MPN remains distinct from that in de novo AML (2, 4, 9). Indeed, mutations in FLT3 and NPM1 are rarely seen in MPN-BP, while, in addition to driver mutations, frequently mutated genes are ASXL1 (30%), TET2 (25%), SRSF2 (22%), RUNX1 (20%), and TP53 (17%). Moreover, 25% and 46% of patients, harbor mutations in 3 and ≥4 genes, respectively. TP53 is the only individual mutation to correlate with shorter OS (9).

In PV, the evolution to MPN-AP/BP generally occurs through a fibrotic progression, represented by the secondary (post-PV) MF stage, whose diagnosis is codified by the IWG-MRT criteria (10). However, direct leukemic transformation from a florid disease (PV) to AP/BP may also occur. Data from the Mayo-AGIMM study indicate that cases of post-PV AP/BP without a documented MF phase range from 37% (AGIMM cohort) to 53% (Mayo cohort). Similar rates were found for post-ET BP (58% and 55% in the two cohorts, respectively) (3).

So far, studies have mainly focused on the comparison of the clinical and biological characteristics of MPN-AP/BP vs. de novo AML. Here, we analyzed the morphological, clinical, and molecular features of patients with post-PV AP/BP arising from a direct progression from PV to leukemic phase (overall, termed post-PV-BP), and compared them with those transitioning through a documented overt fibrotic stage (termed post-PV-MF-BP), documenting that post-PV-BP is characterized by unique features, distinct from post-PV-MF-BP.

The current study is a single-center descriptive retrospective review of 10 consecutive cases of AP/BP evolving from an antecedent PV, either directly, without a documented overt fibrotic progression (termed post-PV-BP, five patients, with two AP and three BP), or through a documented MF stage, i.e., diagnosis of post-PV-MF (termed post-PV-MF-BP, five patients, all BP), followed at the Hematology and BMT Unit of Parma University Hospital.

Demographic, clinical, and molecular data at the time of PV diagnosis and leukemic transformation were collected in a dedicated database after patient de-identification by univocal alphanumeric codes (Ethical Committee approval: protocol 6923, 17/02/2020). Diagnosis of PV, post-PV-MF, and AP/BP was determined according to the WHO/IWG-MRT criteria (10, 11).

Cytogenetic analysis was performed according to the International System for Human Cytogenetic Nomenclature criteria. High-risk karyotype included monosomal karyotype and single or multiple cytogenetic abnormalities of −7, inv(3)(q21.3q26.2)/t(3,3)(q21.3;q26.2), i(17)(q10)/-17/abn(17p) (3). NGS analysis for target myeloid genes was performed by Myeloid Solutions Panel (SOPHiA Genetics, Saint Sulpice, Switzerland). Alignment, base calling, and variant annotation were performed with SOPHiA DDM software and investigating available databases (such as NCBI, COSMIC, GNOMAD, CLINVAR, and ExAC). Pathogenic relevant mutations (category A and B) were characterized by a functional analysis through Hidden Markov Models (FATHMM) score >0.90, while genetic variants that have been flagged as an SNP in the Catalogue of Somatic Mutations In Cancer (COSMIC) database and/or reported as of undetermined significance/benign/likely benign in ClinVar database belonged to category C or D and had a low FATHMM score (<0.5).

Bone marrow biopsies were fixed in neutral buffered formalin, decalcified with ethylenediamine tetra-acetic acid and paraffin embedded for routine diagnostic purposes. Paraffin sections were used for hematoxylin-eosin staining and for reticulin staining. BM age-adjusted cellularity was evaluated according to the European Myelofibrosis Network (EUMNET) consensus (12), and the grading of BM fibrosis was assessed semi-quantitatively following the European consensus guidelines (12). BM sections were also immunostained by using antibodies against CD34 (clone QBend/10), CD61 (clone2f2), CD71 (clone MRQ-48), and myeloperoxidase (polyclonal antibody) (Ventana-Roche, Basilea, Switzerland). Immunostaining was carried out with polymeric system Ultraview 3,3′-diaminobenzidine DAB IHC Detection Kit (Ventana-Roche, Basilea, Switzerland), on a BenchMark ULTRA IHC/ISH instrument (Roche, Basilea, Switzerland), in accordance with the manufacturer’s protocols. Images were captured and analyzed by an Olympus DP22 digital camera (Ventana-Roche, Basilea, Switzerland).

Immunophenotyping was performed using BD FACSLyric™ Flow Cytometer (BD™ Biosciences, Franklin Lakes, NJ, USA). Peripheral blood samples (3–5 mL in EDTA tubes) were processed with lyse and wash protocol. After labeling with antiCD45-APC-H7, antiCD34-PerCP-Cy5.5, antiCD33-PE, antiCD71-APC, and CD117-PE-Cy7 (all from BD™ Biosciences), the different cell populations were identified according to a sequential and cumulative gating strategy, adapted from the ISHAGE guidelines for stem and progenitor cell analysis by flow cytometry (13), combining specific immunostaining with the analysis of lateral dispersion, or side scatter (SSC), and frontal dispersion, or forward scatter (FSC).

For statistical analysis, numerical variables were summarized by their median and range, and categorical variables by count and relative frequency (percentage). Differences in the distribution of continuous variables were calculated by Mann–Whitney/Kruskal–Wallis tests, while categorical variable comparisons were established by χ2/Fisher’s exact test. A P value <0.05 was considered statistically significant. Analysis was performed with dedicated software (Prism 5, GraphPad Software Inc., San Diego, CA, USA).

The natural history of PV encompasses the evolution into post-PV-MF, usually occurring as a late event with reported 10-, 15-, and 20-year incidences of 27.4%, 39.9%, and 61.1%, respectively (19), and the transformation into AP/BP, with incidence estimates of 3%–7% (20). Despite the progress in our understanding of the genetic basis of leukemogenesis (as acknowledged by the 2022 WHO classification), post-PV-AP/BP still lacks a comprehensive characterization and falls into the AML category defined by differentiation (21). Thus, disease-specific clinical, laboratory, and molecular markers are eagerly awaited, to improve diagnosis and define personalized therapeutic approaches.

Depending on the study cohorts, one-third up to approximately one-half of leukemic evolutions occur directly from the PV stage, i.e., without a documented overt fibrotic phase. In this study, we targeted the differential morphological, clinical, and molecular features of AP/BP with a direct onset from the PV stage and those of AP/BP arising after the diagnosis of post-PV-MF. To this end, we performed a single-center retrospective descriptive study on 10 post-PV acute myeloid leukemias: 5 post-PV-BP (with two AP), and 5 post-PV-MF-BP.

The two cohorts displayed similar hematologic parameters at the time of PV diagnosis. By contrast, at the time of leukemic evolution, AP/BP directly arising from PV displayed significantly lower leukocyte counts and spleen diameters as compared to those passing through an overt fibrotic phase, likely due to the compensatory extramedullary hematopoiesis that typifies overt fibrotic stages. The most striking differences emerged from BM morphological analysis: all post-PV-BP were characterized by significantly higher cellularity, lower degree of fibrosis, and dysplastic features involving all three lineages, most prominently the erythroid and megakaryocytic compartments. Of note, these aspects were attributable to the intrinsic characteristics of the disease and not to the iatrogenic effects of previous treatment(s), since all 10 patients underwent hydroxyurea for comparable periods (post-PV-BP: median 76 months, range: 16–144 vs. post-PV-MF-BP: 67 months, range: 7–102, P=0.463).

Recent studies have provided valuable insights into the genomic and molecular alterations driving the transition to blast phase in MPNs (7, 8, 17). Mutations in genes such as TP53, ASXL1, IDH1/2, and RUNX1 have been identified as key drivers of BP-MPN, contributing to increased genomic instability, resistance to therapy, and clonal evolution (8). Understanding these molecular events is crucial for identifying novel therapeutic targets and designing personalized treatment approaches. NGS analysis of our two cohorts shows that post-PV-BP is significantly enriched in DNMT3A mutations, and overall, in mutations of genes that segregate along the DNA methylation pathway (i.e., TET2 and IDH1/2).

DNMT3A and TET2 are the most frequently identified variants among patients with clonal hematopoiesis of indetermined potential (CHIP) and provide selective advantage that foster clonal expansion and myeloid skewing. Both genes are more frequently mutated in MDS and MDS/MPN as compared to PV (22) and both increase in frequency in secondary AML (either post-MDS or post-MPN) (23). However, while DNMT3A mutations are generally present in early-stage disease, TET2 mutation can occur either before or after the acquisition of JAK2V617F mutation, leading to a biclonal disease or to a competitive advantage driving clonal dominance (24). In our cohort, we were able to retrieve NGS data at the time of PV diagnosis for one patient (post-PV-BP3), demonstrating that DNMT3A and JAK2 mutations were already present in the chronic stage. Based on the accumulating evidence on the molecular hits during the history of these disorders, we can speculate that the acquisition of DNMT3A mutation might occur as an early genetic event that might play a role in the dysplastic phenotype that typifies post-PV-BP. Indeed, clonality studies are necessary to answer this question and to assess whether the pre-leukemic clones in the two examined patient groups are independent or not from the clones driving chronic phase PV or post-PV-MF.

Our analysis also detected a non-canonical mutation of JAK2, namely, JAK2R1063H. A growing body of literature is highlighting the potential clinical significance of this germline variant in the MPN setting. Indeed, JAK2R1063H weakly hyperactivates JAK2/STAT5 signaling as compared to the acquired somatic V617F, but likely cooperates with another germline variant, JAK2E846D, in hereditary erythrocytosis (25) and with JAK2V617F itself in MPNs, favoring a more aggressive disease phenotype with significantly higher neutrophil counts and enrichment in thrombotic events (26). A potential pro-thrombotic role of this variant is further supported by a whole-exome sequencing study conducted in 22 young ischemic stroke patients with familial clustering of stroke, in which JAK2R1063H was detected in a proband with embolic stroke of undetermined source and prothrombotic status (27). Finally, in a subsequent next-generation sequencing study on 2154 MPN, MPN-BP, and AML patients, JAK2R1063H was predicted to have a functional consequence on JAK2 by three distinct predictive algorithms and was associated with an increased risk of leukemic transformation when combined with canonical JAK2 mutation (16).

NGS analysis also allowed information retrieval on genetic variations that have been flagged as an SNP and/or reported as of undetermined significance/benign/likely benign. Although no differences were found among post-PV-BP and post-PV-MF-BP, our analysis shed light onto germline variants of potential interest in MPN-BP, such as JAK2 (rs2230722 and rs2230724), HRAS (rs12628), ASXL1 (rs6058694), and TP53 (rs2454206).

HRAS rs12628, originally described by Taparowsky et al. (28), was shown to be associated with increased risk of oral carcinoma, colon, gastric and bladder cancer, and cutaneous melanoma (18, 29–31). This genetic variant has also been associated with syndromic congenital disorders, such as Noonan syndrome, Noonan-related syndrome, and Costello Syndrome (32). To the best of our knowledge, no data are currently available on the contribution of this SNP to hematologic malignancies. JAK2 rs2230722, located in exon 6, and JAK2 rs2230724, present in exon 19 have already been described in MPNs (17), with the latter being associated with the progression to BP, especially in individuals older than 45 years old (33) and therefore suggested as a potential genetic marker of leukemic progression in MPNs. ASXL1 rs6058694 has been reported in AML (34) and in a case report of triple-negative ET unresponsive to therapy (35). Additionally, TP53 rs2454206 has been extensively investigated as a germline predisposition factor for cancer. Despite conflicting data, this SNP may increase the risk of colorectal cancer (36), chronic lymphocytic leukemia (37), and acute myeloid leukemia with adverse molecular and cytogenetic risk (38).

With all the limits of the small number of patients studied and the descriptive intent, this is the first report formally describing recurrent morphological, clinical, and hematological stigmata that differentiate BPs that arise directly from PV from those occurring after post-PV-MF. This peculiar phenotype is consistent with the molecular signature of the disease, typified by mutations of genes involved in the DNA methylation pathways, occurring with a high frequency in MDS and MDS/MPN, leading us to speculate that BP occurring directly from the “florid” phase might be characterized by a distinct physiopathological mechanism.

Further studies in larger cohorts are warranted to translate these observations into robust evidence that may guide diagnosis and advise targeted therapeutic choices, like, for instance, hypomethylating agents.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by Area Vasta Emilia Nord Ethical Commitee (prot. 6923, 17/02/2020). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

LPe: Investigation, Writing – original draft, Data curation. GP: Data curation, Investigation, Methodology, Writing – original draft. SC: Methodology, Investigation, Writing – original draft. CM: Methodology, Writing – review & editing. EuM: Methodology, Writing – review & editing. LPa: Writing – review & editing, Investigation. MG: Writing – review & editing, Investigation. GR: Writing – review & editing, Supervision. MV: Conceptualization, Supervision, Writing – review & editing. CC: Conceptualization, Data curation, Writing – original draft, Supervision, Writing – review & editing. ElM: Conceptualization, Data curation, Funding acquisition, Writing – original draft, Supervision, Writing – review & editing.