Platelet Heterogeneity in Myeloproliferative Neoplasms

Abstract

Myeloproliferative neoplasms (MPNs) are a group of malignant disorders of the bone marrow where a dysregulated balance between proliferation and differentiation gives rise to abnormal numbers of mature blood cells. MPNs encompass a spectrum of disease entities with progressively more severe clinical features, including complications with thrombosis and haemostasis and an increased propensity for transformation to AML. There is an unmet clinical need for markers of disease progression. Our understanding of the precise mechanisms that influence pathogenesis and disease progression has been limited by access to disease-specific cells as bio-sources. Here, we review the landscape of MPN pathology and present blood platelets as potential candidates for disease-specific understanding. We conclude with our recent work discovering progressive platelet heterogeneity by subtype in a large clinical cohort of MPN patients.

Myeloproliferative neoplasms (MPNs) are a group of chronic haematological malignancies characterised by clonal over-production of mature cells of myeloid lineages¹. They are defined by a combination of clinical, laboratory, morphological and molecular genetic features². Three phenotypic subtypes, essential thrombocythemia (ET), polycythaemia vera (PV), and primary myelofibrosis (MF) constitute the ‘classical MPNs’. All have shared molecular genetic pathogenesis-somatic mutations in one of three driver genes (JAK2, CALR, MPL) that lead to constitutive JAK/STAT signalling in haematopoietic stem cells (HSC)^3–7. Clinically they have distinct features, but abnormalities of haemostasis and thrombosis, and risk of progression to acute myeloid leukaemia (AML) are common to all ¹. The chronic nature of MPNs and the accessibility of peripheral blood cells implies that these disorders could potentially serve as a generalizable model for events in cancer progression⁸. Furthermore, recent discovery that the mutational event precedes diagnosis by many years^{9, 10} highlights the opportunity MPNs present to investigate early detection and intervention strategies in cancer.

Our early work¹¹ studying platelets in patients with MPNs established a methodological foundation for exploring subtype-specific signatures, not only of relevance to bleeding and thrombosis outcomes in these patients, but also of underlying disease pathobiology. More recently¹², extended additional data in over 100 MPN patient samples across all three MPN subtypes identify distinct platelet transcriptomic signatures associated with disease processes. Megakaryocytes, which produce platelets, are derived from HSCs, reside in the bone marrow and contribute to MPN pathogenesis^13–16. Platelets therefore provide both a snapshot of the status of the megakaryocyte at the time of platelet release, and a window into the bone marrow microenvironment.

MYELOPROLIFERATIVE NEOPLASMS

Clinical features

The clinical phenotype of the three classical MPNs differ, although there are overlapping features¹. ET is characterised by an elevated platelet count and PV by an elevated haematocrit. Myelofibrosis, the most severe disease in the group, is typified by cytopaenias and presence of reticulin fibrosis in the bone marrow². All patients experience systemic symptoms such as fatigue that impair quality of life ¹⁷. In ET, life expectancy may be near normal, whereas in PV, median life expectancy is around 13 years. Median life expectancy in MF is around 5 years although this varies according to risk score, highlighting the heterogeneity within this disease ¹⁸. There is a risk of transformation to AML, which has a worse prognosis than de-novo AML^{19, 20}. Significant problems in haemostasis and thrombosis occur in all three diseases.

There are challenges in gathering data about long term outcomes for patients with MPNs, and in designing trials in this patient group²¹. Firstly, the diseases are relatively rare, affecting the cohort size. Secondly, adverse outcomes in these patients occur over decades of chronic progressive disease^{3, 4, 22}.

Genetic landscape

In around 90% of patients with MPNs, an acquired mutation that promotes JAK/STAT signalling is identified^{4, 23}. The JAK/STAT pathway transduces signals from cytokines including erythropoietin, thrombopoietin and granulocyte colony stimulating factor ²⁴. A point mutation that activates JAK2, JAK2V617F, is present in around 95% of patients with PV and 40–60% of patients with ET and MF ^25–28. Many of the remaining patients with ET and MF have frameshift mutations in the endoplasmic reticulum chaperone protein calreticulin^{29, 30} which cause ligand-independent activation of the TPO receptor ^{31, 32}.

In addition to driver mutations, recurrent mutations are seen in TET2, DNMT3A, IDH1 and 2, ASXL1 and EZH2^{4, 5}, which contribute to disease initiation and progression by affecting DNA methylation and histone modification³³. Profiling these mutations is used to refine prognosis amongst patients with MF¹⁸. Genes which acquire mutations in progression to AML include TP53 and transcription factors such as RUNX1 ^{19, 20, 34}.

Thrombosis and bleeding in MPNs

Patients with MPNs are paradoxically at risk of both thrombotic and bleeding complications. The risk of arterial and venous thrombosis is increased in patients with MPNs, with the time shortly after diagnosis being the period of highest risk. In the first three months following diagnosis, arterial thrombotic events are around three times commoner in patients with MPNs than they are in matched controls, and venous events are around ten times commoner. Rates are similar across PV, ET and MF ³⁵ . Arterial events are more common than venous events in MPN patients, and in PV and ET cardiovascular disease is the most common cause of death³⁶. Table 1 provides an overview of recent literature documenting clinical outcomes of patients with ET and PV.

Table 1:

Representative overview of published literature on clinical outcomes of patients with ET and PV

MPN Subtype	Study population (n)	Follow up (time)	Survival (time)	Thrombosis (all)	Thrombosis (arterial)	Thrombosis (venous)	Bleeding	Transformation
ET118	891	0-27 years (median 6.2)	Median overall survival 14.7 years (similar to sex and age standardized European population)	22% at 15 years				Fibrotic 9%, Leukemic 2% (at 15 years)
ET47 (high vascular risk)	809 (404 HU plus aspirin, 405 anagrelide plus aspirin)	12-72 months (median 39)	Death 27/404 (hydroxyurea plus aspirin) 31/405 (anagrelide plus aspirin)		17/404 (hydroxyurea plus aspirin) 37/405 (anagrelide plus aspirin)	14/404 (hydroxyurea plus aspirin) 3/405 (anagrelide plus aspirin)	Serious bleeding 8/404 (hydroxyurea plus aspirin) 22/405 (anagrelide plus aspirin)	Myelofibrosis 5/404 (hydroxyurea plus aspirin), 16/405 anagrelide plus aspirin) AML or myelodysplasia 6/404 (hydroxyurea plus aspirin), 4/405 (anagrelide plus aspirin)
ET1, 119 (lacking high risk features for vascular events)	382 (final analysis included 176 aspirin alone and 192 aspirin plus hydroxycarbamide	0-187 months (median 73 months)	Death 7/176 (aspirin alone), 10/182 (aspirin plus hydroxycarbamide)		7/176 (aspirin alone) 5/182 (aspirin plus hydroxycarbamide)	3/176 (aspirin alone) 4/182 (aspirin plus hydroxycarbamide)	Serious bleeding 2/176 (aspirin alone)3/182 (aspirin plus hydroxycarbamide)	Myelofibrosis 5/176 aspirin alone, 1/182 hydroxycarbamide plus aspirin AML 2/176 aspirin alone, 3/182 hydroxycarbamide plus aspirin.
PV120	1638	Mean 2.7 years, STDEV 1.3 years (4393 person years)	3.7 deaths/100 patients/year (45% cardiovascular mortality, 13% haematological transformation)	5.5/100 patients/year			Major bleeding 0.8/100 patients/year, any bleeding 2.9/100 patients/year	Haematologic transformation 1.3/100patients/year
PV62	518 (253 aspirin, 265 placebo)	1-3 years (1478 person-years)	Death 9/253 aspirin, 18/265 placebo	8 out of 253 aspirin, 21/265 placebo	5/253 aspirin, 13/265 placebo	17/253 aspirin, 41/265 aspirin	Major bleeding 3/253 aspirin, 2/265 placebo. Any bleeding 23/253 aspirin 14/265 placebo
PV43	365 (182 low target haematocrit <0.45, 183 high target haematocrit 0.45 to 0.5)	1.5-48.1 mos (median 31)	Death 3/182 low haematocrit, 6/183 high haematocrit	8/182 low haematocrit, 20/183 high haematocrit	4/182 low haematocrit, 14/183 high haematocrit	1/182 low haematocrit, 6/183 high haematocrit	2/182 low haematocrit, 5/183 high haematocrit	Myelofibrosis 6/182 low haematocrit, 2/183 high haematocrit. AML or myelodysplasia 2/182 low haematocrit, 1/183 high haematocrit

The pathogenesis of thrombosis in MPN patients is complex. Contributory factors include cell counts, clonal haematopoiesis, specific effects of driver mutations, and effects of systemic inflammation on platelets, granulocytes and endothelial cells^37–41. Evidence for the contribution of elevated haematocrit and platelet counts is mixed. Amongst patients treated for PV cardiovascular events are more frequent in those with less stringent haematocrit control ^{42, 43}. Thrombosis occurs even in individuals with a normal haematocrit however ⁴⁴. Studies in ET have not shown an association between platelet count and thrombosis^{45, 46}, and differences in rates of thrombosis were observed in a trial where platelet count did not differ between arms⁴⁷.

There are several indicators that platelet function is altered in MPNs. There is upregulation of genes associated with thrombosis in platelets of patients with PV and ET compared to controls⁴⁸. Clinical platelet function tests show increased aggregation in response to several ligands in patients with ET compared to healthy individuals⁴⁹. Examination of the specific effect of JAK/STAT activation by JAK2V617F in platelets has produced conflicting results in different models, that nonetheless corresponds to the thrombotic and bleeding risks in patients. There is evidence both for a pro-thrombotic tendency⁵⁰ and for impaired haemostasis with reduced granule formation and impaired aggregation⁵¹.

Increased rates of cardiovascular disease are seen in individuals with clonal haematopoiesis of undetermined significance ⁵². Evidence for a specific role of the driver mutation influencing thrombotic risk comes from ET, where patients with CALR mutations have significantly lower rates of thrombosis than patients with JAK2 or MPL mutations, despite those with CALR mutations having higher platelet counts^53,54. Controversy over whether allele burden influences thrombotic risk ^{45, 55} also arises because peripheral blood measurements only partially reflect the cell populations⁹ in the bone marrow.

Inflammation is a risk factor for thrombosis, and in PV and ET an association is seen between C-reactive protein levels and occurrence of thrombotic events⁵⁶. In addition to endothelial activation caused by systemic inflammation, there are reports of JAK2V617F in endothelial cells and a suggestion this may promote thrombosis⁵⁷. Increased platelet activation and platelet-leucocyte complexes are seen in patients with MPNs and are thought to contribute to thrombosis⁵⁸.

Bleeding problems can arise in patients with MPNs with thrombocytosis or thrombocytopaenia, for example in ET the relationship between major haemorrhage and platelet count is a U-shaped curve⁴⁶. Around 6% of patients will experience bleeding complications at the time of diagnosis⁵⁹. Thrombocytosis can cause acquired von Willebrand disease, with selective depletion of large von Willebrand factor multimers⁶⁰ although the bleeding risk associated with thrombocytsis may differ between MPN sub-types⁶¹. Bleeding related to anti-platelet drugs is an important consideration in this patient population.

Therapy

Controlling thrombotic risk is the main goal of therapy in PV and ET. In PV there is evidence that thrombotic risk reduction can be achieved with control of the haematocrit⁴³, and aspirin⁶². In ET, treatment decisions are based on individualized thrombotic risk. The link between platelet count and thrombotic risk is less clear^{46, 47}, although there is evidence that hydroxycarbamide reduces risk in high-risk patients⁶³. Aspirin is recommended to reduce thrombotic risk⁶⁴, although in ET there is a lack of prospective evidence for benefit. There is an unmet need for tools to personalise aspirin treatment in MPNs⁶⁵ since some patients experience excessive bleeding related to aspirin, whereas others suffer thrombotic events despite taking standard doses of aspirin. In ET, higher platelet turnover is associated with increased rates of thrombosis⁶⁶. There is evidence that rapid platelet turnover reduces the effectiveness of aspirin, and that this can be overcome by adjusting dose regimens⁶⁷. The identification that platelet RNA is a biomarker for response to antiplatelet therapy is highly relevant⁶⁸. Clopidogrel may be used as an alternative in patients who are intolerant of aspirin although evidence for this is lacking⁴⁴. For patients with venous thrombotic events, anticoagulation treatment followed by long-term prophylaxis is recommended due to a high recurrence risk. Traditionally this is with heparins and vitamin K antagonists, although there is emerging evidence that direct oral anticoagulants may be effective and safe^{69, 70}.

Drugs inhibiting JAK2, such as Ruxolitinib, have been designed to target the molecular cause of the disease and have been partially effective in MF⁷¹ and PV⁷², but less so in ET⁷³. Interferons are also used, and there is evidence that they alter the balance between normal and malignant haematopoiesis⁷⁴.

HAEMATOPOIETIC STEM CELLS AND THE BONE MARROW NICHE IN MPNS

The HSC population in patients with MPNs is heterogeneous, consisting of wild-type HSCs and sub-clones of HSCs with one or more mutations ⁸. Haematopoietic stem and progenitor cells reside within niches⁷⁵, specialised microenvironments in the bone marrow that sustain the equilibrium between proliferation and differentiation⁷⁶. Maintenance and progression of haematological malignancies is facilitated by reciprocal interactions between the malignant cells and surrounding niche cells ⁷⁵. Inflammation in the microenvironment ⁷⁷ and JAK/STAT activation in wild-type cells ⁷⁸ are important contributors to the niche changes in MPNs⁷⁹.

PLATELETS IN MPNs

Platelet origin and function

Blood platelets are critical in multiple processes and diseases, from their traditional role in haemostasis and wound healing to inflammation, immunity, cancer metastasis and angiogenesis^80–82. Platelets originate from bone marrow precursor megakaryocytes, which themselves are differentiated from HSCs and the two cell types together play an important role in our understanding of MPN pathology. Megakaryocyte expansion is a diagnostic criterion in all three MPNs². Populations of HSCs primed for megakaryopoiesis can be identified at an early stage of differentiation^{15, 83, 84}. Megakaryocytes from patients with MF proliferate more than those from healthy controls and aberrantly overexpress myeloid transcription factors⁸⁵. Megakaryocytes are a cellular component of the HSC niche⁷⁵ and modulate HSC quiescence^{86, 87}. In mice, expression of Jak2V617F in megakaryocyte lineage-committed cells triggers cell non-autonomous increased erythropoiesis ¹⁴. Megakaryocytes produce cytokines that promote fibrosis including IL6, CXCL4 and TGF-β^{14, 88, 89}. Treatment with alisertib, an Aurora Kinase A inhibitor, that results in decreased megakaryocyte numbers and increased maturation led to a reduction in fibrosis in a mouse MF model¹⁶ and in five out of seven patients in a phase 1 clinical trial⁹⁰.

The platelet transcriptome

Platelets contain a complex transcriptional landscape of messenger RNAs, unspliced pre-mRNAs, ribosomal RNAs, transfer RNAs and microRNAs^{13, 91–93}. Most platelet RNA expression results from the transcription of nuclear DNA in the megakaryocyte, and thus reflects the status of the megakaryocyte at the time of platelet release into the circulation^{13, 91}. This is overlaid with further complexities arising from splicing events triggered by receptor activation at the platelet surface^{94, 95} and inter-cellular transfer of RNA into megakaryocytes⁹⁶ or circulating platelets⁹⁷. The molecular signature of platelets is therefore changed in disease conditions^{80, 97, 98}.

It is important to examine the transcriptome of megakaryocytes as well, given their centrality in the pathogenesis of MPNs; and identify the differential contributions to the platelet transcriptome from the parent megakaryocytes versus the peripheral disease environment. However, there are challenges to studying megakaryocytes from patient cohorts. Bone marrow biopsy is a required procedure and presents significantly more discomfort and risk to patients than a peripheral blood draw. Megakaryocytes are a relatively rare cell in healthy bone marrow though their transcriptomes have been examined^{99, 100} by enriching with density gradient centrifugation followed by positive selection for CD61 and identification in single cell analysis based on high expression of megakaryocyte-specific genes. For MF there is the added complication that fibrosis often prevents liquid bone marrow from being extracted by aspiration. The large size and fragility of megakaryocytes are also a concern. Single cell technologies can be used to identify and study megakaryocyte progenitors amongst haematopoietic stem and progenitor cells¹⁵, but it is unclear to what extent these reflect the transcriptomic profiles of mature megakaryocytes. In-vitro differentiation of megakaryocytes from stem cells or patient-derived HSCs, which can now be performed in large enough numbers to perform transcriptomic profiling^{101, 102}, may be complicated by signatures reflecting the in-vitro differentiation process, and lack features conferred by the bone marrow microenvironment. Taken together, these challenges to studying megakaryocyte transcriptomes highlight the potential for exploring the platelet transcriptome as a feasible and viable alternative.

Transcriptomic heterogeneity in MPNs

In the context of MPNs therefore, the platelet transcriptome represents a biomarker of megakaryocyte activity, thus capturing information on the HSC niche and providing a snapshot of the underlying thrombotic, haemostatic and inflammatory derangements. Figure 1 illustrates the different potential influences on the platelet transcriptome in MPNs.

Figure 1. The transcriptomic signature of circulating platelets provides a window into disease biology in MPNs.

Megakaryocytes are expanded in MPNs and comprise a mixture of wild-type cells and cells with driver mutations and sub-clones with additional mutations, reflecting the heterogeneity within the HSC pool from which they are derived. Megakaryocytes influence surrounding cells in the bone marrow niche through non-cell-autonomous effects and are themselves influenced by environmental factors including inflammation and fibrosis. As well as reflecting the status of the megakaryocyte at the time of platelet release into the circulation, the platelet transcriptome is further influenced by inter-cellular transfer of RNA and by splicing events triggered by receptor activation. The transcriptome therefore gives insights into disease biology and carries signatures that distinguish between disease types and stages of disease progression.

Our recent work^{103, 104} using RNA sequencing of purified platelets from two cohorts of patients with ET, PV, and MF (contrasted with healthy donors) confirms the intra-disease heterogeneity and identifies novel therapeutic targets. The platelet transcriptome is significantly reprogrammed in the MPN setting, with a wealth of transcript associations that may be missed in using conventional tissue sources such as serum, plasma, whole blood, or bulk bone marrow (Table 2 summarizes MPN studies across these bio-sources and suggests potential for future investigations on assessing intersecting signatures/pathways). Our data also closely overlaps with two other recent platelet transcriptomic studies on select MPN subtypes: thrombo-inflammatory signatures in PV from Gangaraju and Prchal et al¹⁰⁵ and fibrosis-associated signatures in MF from Guo and Erber et al¹⁰⁶.

Table 2.

Summary of cell types for transcriptomic study in MPNs, study findings, advantages and disadvantages of each source.

Cell type	MPN subtypes	Findings
Whole blood121	Healthy donors, ET, PV, MF	• Dysregulation of genes involved in inflammation • Detects signatures that might be missed by profiling isolated cell types
Granulocytes122	Healthy donors, ET, PV, MF	• JAK-STAT signature common across disease sub-types and driver mutations • TET2 mutation is associated with a distinct signature • Extent of clonal predominance may influence signatures
Circulating stem and progenitor cells123	JAK2V617F mutated PV	• Reveals heterogeneity, separates patients into distinct groups with different rates of disease complications • Rare cell type, may not be sufficiently abundant in all MPN subtypes
Circulating stem and progenitor cells15	MF, mobilised cells in healthy donors	• Heterogeneity of Mk progenitors in MF cell type, cost if performing single cell sequencing
Platelets124	ET, reactive thrombocytosis	• Transcript profiles distinguish ET from reactive thrombocytosis
Platelets106	Healthy donors, MF	• Distinguishes MF from controls, separates patients into groups with and without fibrosis
Platelets12, 104	Healthy donors, ET, PV, MF	• Two mutually validating MPN patient RNA-seq cohorts discriminate each clinical phenotype; and identify progressive transcriptomic markers that also enable predictive signatures for MF.

Distinctive signatures of over and under-expressed genes are seen for each disease entity, with the greatest differences from normal seen in MF. Of note, these are independent of driver mutation, highlighting the significance of other factors in disease pathogenesis and progression. The data corroborates the role of inflammation in MPN pathogenesis and identifies gene expression signatures reflecting activation of inflammatory signalling pathways. Characteristic changes are noted in MF patients treated with the JAK1/2 inhibitor ruxolitinib, both confirming known mechanisms of action as well as identifying potential new or combinatorial targets for MPN therapy.

Molecular pathways differentially activated between MPN subtypes identified expected immune modulatory responses (e.g. a consistent interferon alpha/gamma, and IL2 STAT5) in addition to robust (FDR<0.05) signaling in oxidative phosphorylation (OXPHOS), mTORC1, and reactive oxygen species (ROS) production pathways. Coagulation- and complement-associated gene sets were also expectedly enriched across ET, PV, and MF. Particularly in MF, cycle progression and proliferation pathways around c-MYC and E2F targets, and G2M checkpoint pathways emerged as highly significant (FDR < 0.001) and altogether pointing to a strong unfolded protein response as a key factor, likely attributed to a chronic integrated stress response (ISR)¹⁰⁷. Together, these data demonstrate that in addition to immune factors such as type I/II interferons and dysregulation of interleukin-dependent inflammatory responses, which have been linked to MPNs, platelet transcriptional signatures of proliferation, metabolic, and proteostasis signaling are a feature of MPN pathogenesis.

Most importantly, platelet gene expression profiling in MPN offers directions for prediction of myelofibrosis. Applying machine-learning algorithms of LASSO penalized regression under two conditions of external validation ¹⁰⁸: temporal (using our two cohort design) and geographical (independently published datasets on healthy donors^{106, 109} and MF ¹⁰⁶, we uniquely discriminate MPN subtypes from each other, and healthy controls using three model types and predict MF at high accuracy. The highest performing model used a set of progressively differentiated MPN genes at an area under the (ROC) curve of 0.96 (temporal) and 0.97 (geographical); and rendered a core signature of <5 candidate markers as top predictors of disease progression.

Collectively, our work presents the platelet transcriptome as a proof-of-concept not only in understanding disease progression but also in deciphering mechanistic insights and developing predictive machine learning algorithms.

Single-cell technologies

Despite the wealth of insight gained from bulk RNA sequencing of purified platelets, this approach does not answer questions arising from heterogeneity of sub-populations of HSC and their progeny (Figure 1). Single-cell approaches applied to stem and progenitor cells are starting to offer this granular perspective.

Our understanding of normal and malignant haematopoiesis is being transformed by single-cell transcriptomics, where expression of thousands of genes measured from individual cells is used to infer differentiation trajectories^110–112. In MPNs, this has revealed heterogeneity in megakaryocyte progenitors in MF, where patients have a population resembling healthy controls, but also eight other distinct populations¹⁵. Methods that simultaneously determine the mutation status and transcript profile of single cells bring further resolution, including the potential to examine sub-clones with additional mutations^{113, 114}. Surprisingly, driver-mutated cells do not form novel clusters with unique gene expression profiles but are found across all the stem and progenitor cell populations^{114, 115}. Emerging technologies that combine single cell and spatial information¹¹⁶ have the potential to enrich this with information on influences of the marrow microenvironment.

Challenges remain in deciphering heterogeneity within platelet populations, for example to identify the contribution of wild-type and driver-mutated megakaryocytes to the circulating platelet population and establish whether there are transcriptomic differences between them. It will be of interest to see whether single-cell transcriptomic technologies can be adapted to study single platelets, although low RNA yields may be a potential limitation.

Early detection of MPNs

Novel work^{10, 117} using somatic mutations as molecular clocks to calculate the timing of driver mutation acquisition in patients with MPNs has revealed a very long latency between mutation acquisition and presentation with overt disease. Furthermore, it was inferred that the mutations would have been detectable many years before disease presentation^{10, 115}. Notably, rates of clonal expansion varied between individuals. Platelets are potential sources to investigate the multiple influences including inflammation and the bone marrow microenvironment which could be contributing to this trajectory. Future directions for personalised medicine in MPNs therefore encompasses not only risk prediction for patients already diagnosed with an MPN, but also early detection, risk stratification for progression to disease and potentially early intervention for individuals with low burden mutant clones¹⁰.

CONCLUSIONS

Platelets play a central role in the haemostatic and thrombotic complication of MPNs and provide a window into the stem and progenitor cell populations and bone marrow microenvironment that are responsible for disease pathogenesis and progression. The platelet transcriptome offers markers of disease phenotype, disease progression and biological insights into potentially targetable aspects of disease biology - one step closer to personalising medicine in MPNs.

Highlights.

• Myeloproliferative neoplasms (MPNs) arise when haematopoietic stem cells (HSCs) acquire driver mutations that cause abnormal blood counts.

• MPNs are a model for cancer progression broadly relevant to bone marrow biology.

• MPN disease phenotype is the consequence of a combination of factors including effects of driver mutations on HSCs and their progeny, non-cell-autonomous effects in the bone marrow microenvironment, and systemic inflammation.

• Thrombosis and bleeding cause significant morbidity and mortality in MPN patients. Megakaryocyte populations are expanded in patients with MPNs and are particularly central to disease pathogenesis in myelofibrosis.

• The platelet transcriptome integrates disease-specific information from the parent megakaryocytes as well as the bone marrow microenvironment and the peripheral circulation. There is an unmet need for markers of disease progression in MPNs and therefore, platelets are ideal peripherally accessible biosources that also reflect underlying disease.

Abbreviations

AML

acute myeloid leukaemia

CALR

calreticulin

ET

essential thrombocytosis

HSC

haematopoietic stem cells

JAK2

janus kinase 2

MF

myelofibrosis

MPN

myeloproliferative neoplasm

PV

polycythaemia vera