JAK2^V617F allele burden in polycythemia vera: burden of proof

Polycythemia vera (PV) is a hematopoietic stem cell (HSC) neoplasm defined by activating somatic mutations in the JAK2 gene, characterized clinically by overproduction of red blood cells, platelets, and neutrophils. PV results in varying burdens of disease-specific symptoms, high rates of vascular events, and evolution to a myelofibrosis phase or acute leukemia. The pivotal discovery in 2005 that JAK2^V617F underlies nearly all PV transformed our understanding of the disease pathobiology and formed the basis for disease classification criteria and targeted therapeutics.^1-3 Despite this, contemporary management of PV largely remains focused on normalization of peripheral blood cell counts with little regard to the clonal expansion of JAK2 mutant cells over time. Recent data from prospective clinical trials for PV have demonstrated the ability of some therapies to significantly decrease not only peripheral blood cell counts but also JAK2^V617F mutant allele burden. This review focuses on the role of JAK2^V617F clonal burden in PV outcomes, highlighting the prospective clinical trials where JAK2^V617F burden is measured.

JAK2 functions as a signal transducer of growth and differentiation in the HSC compartment for cytokine receptors that do not have intrinsic kinase activity, including the erythropoietin receptor, the thrombopoietin receptor, and the granulocyte colony–stimulating factor receptor. The activating JAK2 mutation, JAK2^V617F, results in increased signal transduction, which translates to panmyelosis and increased numbers of red cells, neutrophils, and platelets. (Figure 1A). Mutations in JAK2 are highly associated with mitotic recombination events, which render cells homozygous for JAK2^V617F, such that unmutated heterozygous and homozygous JAK2^V617F stem cells and progenitors may populate the hematopoietic compartment in varying and dynamic combinations.^4-7,JAK2^V617F heterozygous vs homozygous cells differ in their biological behavior as gene dosage influences the degree of enhanced JAK2^V617F signal transduction (Figure 1B). The variability in JAK2^V617F genotypes and their clonal expansion results in quantitative measures of JAK2^V617F that range from as low as a fraction of a percent to as high as 100%, expressed clinically as variant allele fraction (VAF) (Figure 1C).⁸ As quantitative JAK2^V617F testing has become sensitive to VAFs of <1%, virtually all patients with a clinical phenotype of PV harbor JAK2^V617F (97%) or mutations in exon 12 of JAK2.^8-12 The effect of JAK2^V617F gene dosage to a PV phenotype is recapitulated in murine models of JAK2^V617F PV, where stages of PV phenotype depended both upon the quantity of JAK2^V617F expressed, and whether JAK2^V617F was expressed at the HSC or progenitor level.^13-15 Genome-wide association studies in JAK2^V617F-positive clonal hematopoiesis (CH) or myeloproliferative neoplasm (MPN) show highly significant associations with loci in JAK2, TERT, SH2B3, TET2, ATM, CHEK2, PINT, and GFIB, with nearly 18% of heritable risk attributed to these and other loci.^16-18 Inflammatory states and/or genetic predisposition to inflammation are risk factors for both the acquisition and expansion of JAK2^V617F clones, as is male sex.^10,19-24 

JAK2^V617F increases signal transduction and drives quantitative and qualitative alterations in HSC biology. JAK2^V617F-mutant HSCs elaborate excess cytokines and reactive oxygen species, which alter the bone marrow niche, reinforce the growth advantage of the clonal cells, and drive genomic instability, myelofibrosis, and osteosclerosis.^14,25-31 Mechanistically, excessive JAK2 signaling in the HSC setting results in a pseudohypoxia state owing to hypoxia-inducible factor (HIF) stabilization by reactive oxygen species. Increased HIF leads to a switch from aerobic metabolism to glycolysis and enhances HSC self-renewal via the downregulation of LKB1/ STK11.^29,32 The disordered niche in addition allows for increased egress of CD34⁺ cells, which lodge in spleen and other tissue that begets extramedullary hematopoiesis (Figure 1B).^33,34 Indeed, JAK2^V617F VAF in PV positively correlates with circulating CD34⁺ cell levels, bone marrow cellularity, and splenomegaly (a haven for extramedullary hematopoiesis) (Table 1).^11,12,38 

The growth trajectories of mutant HSCs are varied and reflect the stem cell fitness of mutation-bearing clones that are both mutation-specific and age-specific.^39-41 In MPN, high peripheral blood VAF may overestimate the concomitant HSC burden, given the advantage of JAK2-mutated progenitors to terminally differentiate.^4,5 The development of clonal dominance at the HSC and progenitor level occurs variably in the MPN and is a manifestation of stem cell fitness. Stem cell fitness in the MPN, and particularly PV, is correlated with JAK2^V617F VAF in HSC and progenitor compartments, with higher VAF in these compartments associating with higher fitness levels.^5-7 The highest fitness levels associated with worse event-free survival and poorer clinical response and predicted those outcomes better than peripheral blood VAF alone. Thus, peripheral VAF is an inadequate surrogate for evaluating stem cell fitness in the MPN in general.⁷ However, clinically in PV, a high peripheral VAF is a marker for developing clonal expansion of HSC and is an established risk factor for progression to post-PV myelofibrosis (MF).^4,5,8,11,38 

JAK2^V617F VAF is also a critical determinant of both quantitative and qualitative aspects of terminally differentiated cells in PV (Figure 1A-B). Peripheral blood JAK2^V617F VAFs greater than 50% correlate with a higher leukocyte count, a higher absolute neutrophil count, a higher hematocrit (HCT), and a lower platelet count than VAFs <50% (Table 1).^11,12,38 When considering JAK2^V617F VAF in PV as a continuous variable, WBC and HCT tightly positively correlate with VAF, whereas mean corpuscular volume and platelet count negatively correlate with VAF.⁸ Further, higher JAK2^V617F VAF is associated with increased red cell mass, decreased serum iron, decreased transferrin saturation, and lower serum erythropoietin (Table 1).^8,37,38,42 Excess HIF may be the basis of the reduced hepcidin, increased erythroferrone, and altered iron metabolism observed in patients with PV, although additional mechanisms may be at play.⁴³ PV neutrophils have increased HIF-mediated gene expression as well as increased expression of tissue factor, F3, the primary initiator of the extrinsic pathway of coagulation, which again correlate with JAK2^V617F VAF (Figure 1B)^32,43-46 Further, clonal neutrophils in PV exhibited enhanced ability to release extracellular traps, activate integrin signaling, and enhance adhesion.^47-49 

JAK2^V617F and its VAF influence both quantity and quality of blood cells, critical factors associated with the high prevalence of thrombotic events before, at, and after the diagnosis of PV.^50,51 Several studies have identified an independent signal of JAK2^V617F VAF in thrombosis risk, showing that in PV, VAF ≥50% exerts risk for venous thrombosis (VTE), even after adjusting for prior events, white blood cell count, and age.^35,38,52 Focusing on 576 patients with PV, Guglielmelli confirmed that JAK2^V617F VAF >50% independently associated with a higher VTE risk and noted that this was evident in both conventionally low-risk and high-risk patients with PV.³⁸ Moreover, JAK2 signaling interacts with traditional vascular risk factors, and in PV, JAK2^V617F VAF associated with a higher neutrophil to lymphocyte ratio and carotid plaque burden, and higher elevations of C-reactive protein (Table 1).^36,53 

The JAK2^V617F VAF dosage effect on both cytoses and thrombotic risk observed in PV is recapitulated in clonal hematopoiesis (CH) studies in non-MPN populations. Individuals with JAK2^V617F CH had significantly higher white blood cell counts, platelet counts, and hemoglobin concentrations in addition to higher rates of arterial and venous thrombotic events compared with those without CH.^54-57 In 2019, Cordua et al explored the association between JAK2^V617F-positive CH and VAF in 19 958 participants in the Danish General Suburban Population, using an assay sensitive to 0.009% VAF. JAK2^V617F-positive CH was detected in 3.1% with a median VAF of 2.1%. Individuals with JAK2^V617F CH and higher VAFs had higher blood cell counts and a greater risk for developing venous thrombosis, pulmonary embolism, or a cerebrovascular event, demonstrating the increase in thrombotic events associated with JAK2^V617F positivity and with increasing clonal burden (Table 1).²³ 

Clonal expansion, as evidenced by increasing JAK2^V617F VAF over time, has consequences in both CH and PV. In the general population, JAK2^V617F CH progresses to overt MPN and is associated with increased JAK2^V617F VAF.^56,57 Cordua and colleagues enrolled 41 JAK2^V617F mutation-positive individuals from the Danish Suburban Population study to undergo re-evaluation on average 7 years after their initial exam. Individuals were classified as no MPN, possible MPN, or MPN based on bone marrow evaluation and peripheral blood counts. JAK2^V617F VAF baseline and increase from baseline were higher (5.9 and 9.4, respectively) in the MPN group than those measures in the no MPN group (.2 and .39, respectively).²⁴ Like in CH, in PV, JAK2^V617F VAF is a dynamic, time-dependent factor. Although the diagnosis of PV occurs most commonly in the sixth decade of life, acquisition of JAK2^V617F occurs in an individual years to decades before clinical presentation.^58,59 A 2010 study estimated an increase in neutrophil JAK2^V617F VAF over the first 10 years after PV diagnosis to be 1.4% per year, whereas a 2022 analysis of the hydroxyurea (HU)–treated arm of the PROUD-PV study showed an average VAF increase of 7% over 5 years (1.4% per year).^22,60 Increasing JAK2^V617F VAF in PV is associated with longer disease duration, higher risk of developing post-PV MF, higher lactate dehydrogenase (LDH), and more prevalent splenomegaly (Figure 1C; Table 1).^8,11,12,38 Indeed, the mean JAK2^V617F VAF in recently diagnosed patients with low-risk PV enrolled in the LOW-PV trial was 35%, whereas the mean VAF in patients with high-risk PV enrolled in the RESPONSE-1 trial was 75% (Table 2).⁶¹ 

The study of JAK2^V617F VAF has reinforced our understanding of the latency of clonal dynamics in PV and explains the decades-long courses of individuals with PV. The treatment of PV is therefore on decades-long terms, with the goals of preventing thromboses, improving PV-associated symptoms, and preventing the progression of the disease to MF or acute myeloid leukemia (AML) (Figure 2). The strategy to achieve these goals since the time of Dameshek had been to consider patients with PV as “generally normal” and to target the production of excess circulating blood cells.⁶⁴ However, we have learned that PV is fundamentally abnormal, with overactivation of the JAK2 signaling pathway resulting in a thrombo-inflammatory state and a disordered HSC niche that propagates disease progression to MF or AML (Figure 1B-C). Accordingly, clinical trials in PV must be carried out over at least a 5-year period to assess outcomes and their relationship to clonal dynamics.

The cornerstone of therapy in PV is controlling HCT (Figure 2; Table 3). The importance of HCT control was validated in CYTO-PV, where cardiovascular events and deaths were reduced by fourfold when hematocrit was targeted to below 45% vs 50%, with treatments to achieve these targets primarily phlebotomy and/or HU.⁷² Beyond HCT, control of white cell or platelet count and beneficial outcomes in PV have been more difficult to define, given the strong influence of HCT control on thrombosis outcomes. Problematically, real-world analyses in PV, where the primary therapies were phlebotomy and HU, failed to demonstrate the benefit of normalizing blood cell counts in reducing the risk of thrombosis, even when applying ELN response criteria to the analyses.^73,74 Control of HCT was a primary end point of the RESPONSE-1 and RESPONSE-2 trials, where ruxolitinib was tested against the best available therapy (BAT; majority HU) in individuals requiring phlebotomy and with an enlarged spleen (RESPONSE-1) or without an enlarged spleen (RESPONSE-2).^3,63 Ruxolitinib arms in both trials demonstrated superior HCT control, and a higher probability of normalization of blood counts, in both the original and 5 year extension trials.^68,69 The PROUD-PV and CONTINUATION-PV trials, where patients with PV were randomized to either ropeginterferon or HU, showed equivalent complete hematologic responses at 12 months but significantly higher complete hematologic responses in the ropeginterferon arms at 24, 36, and 60 months .^60,62 

Superior blood count control does translate to reduced thrombotic risk in prospective trials, although many years of follow-up are required to observe a significant signal. Thrombosis rates per 100 patient years in the 5-year follow-up studies of the RESPONSE-1 (1.2%) and −2 (1.5%) trials were lower in the ruxolitinib arms than those in BAT, although this did not achieve statistical significance. In contrast, MAJIC-PV, a randomized prospective trial testing upfront treatment with ruxolitinib vs BAT in high-risk PV, showed significantly improved thrombosis-free, progression-free, and overall survival in ruxolitinib-treated patients.⁷⁵ The PROUD-PV and CONTINUATION-PV trials showed low but very similar thrombotic event rates at 3- and 5-year follow-up in the ropeginterferon arm compared with HU; however, by 6 years of follow-up, event-free survival (disease progression, death, and thromboembolic events) was significantly higher among ropeginterferon alfa-2b–treated patients.⁷⁶ 

Prospective PV trials where VAF was collected show distinct dynamics in both the HU- and phlebotomy-only treatment arms. In the DAHLIA, PROUD-PV and Myeloproliferative Disorders Research Consortium 112, HU was shown to reduce JAK2^V617F VAF in the first year but then rebound and continue to rise after 2 years, such that the VAF increase over 5 years was roughly 1.2% per year (Table 3).^60,67,77 Similarly, the VAF rise in the BAT arm in RESPONSE-1 was 1.2% at week 32, and in RESPONSE-2, was 1.2% per year over the 5-year trial.^3,69 In the LOW-PV study, where phlebotomy alone was compared with phlebotomy plus low-dose ropeginterferon, the increase in JAK2^V617F VAF in the phlebotomy arm was 1% over 12 months (Table 3).⁶¹ 

In contrast, both ruxolitinib and ropeginterferon induced sustained reductions in JAK2^V617F VAF. In RESPONSE-1, the mean baseline VAF of 75% fell to 47% at 5 years in the ruxolitinib arm, whereas in RESPONSE-2, the median baseline VAF of 53% fell to 38% at 5 years.^68,69,78 Lower JAK2^V617F VAF nadirs were achieved by treatment with pegylated interferons. In a phase 2 trial of peginterferon alfa-2a in PV, VAF dropped from 45% to 3% by 3 years, with nearly identical dynamics in other phase 2 trials of either peginterferon alfa-2a or ropeginterferon, whether used alone or in combination with ruxolitinib (Table 4 ).^70,80,81 In the randomized trials of ropeginterferon vs phlebotomy alone (LOW-PV) or HU (PROUD-PV), VAF reductions in the ropeginterferon arms were nearly identical between the 2 studies at 12 and 24 months (Table 4).^60,61,79 Over the first 24 months of ropeginterferon treatment, JAK2^V617F VAF on average declined by 1% per month in the PROUD-PV study, and at a very similar rate in 4 additional trials of this biologic in PV.^60,79,82 In PROUD-PV/CONTINUATION-PV, 60 months of ropeginterferon reduced the mean baseline JAK2^V617F VAF from 37% to 8%, and 20% of individuals achieved a VAF below 1%.^60,76 (Table 4)

JAK2^V617F VAF reductions were more commonly associated with superior complete hematologic responses and spleen responses in both the RESPONSE-1 and PROUD-PV trials.^3,62 In analyses of >5-year follow-up of patients with PV treated with ropeginterferon, JAK2^V617F VAF was reduced compared with those treated with HU, and was associated with complete hematologic responses and reduced disease progression.⁶⁰ Although ruxolitinib has a more modest effect on VAF in PV when used in the high-risk setting, ruxolitinib was associated with deeper molecular responses compared with BAT, and those with deeper molecular responses (<2% VAF) had a greater likelihood of complete hematologic responses and improved progression-free survival.^75,83 

Ruxolitinib and interferons have different impacts on the molecular response in PV, potentially explained by the differential ability of the agents to target the mutant long-term stem cell pool, but this interpretation is confounded by the fact that ruxolitinib trials treated higher-risk (and higher median VAF) patients with PV.^66,JAK2^V617F VAF sensitivity testing was <0.01% in many of the recent trials, and thus could detect 3 log reductions. The average depth of molecular response achieved by interferon therapies in the reported studies are 1 to 2-log reduction, although some patients may achieve more (Tables 2 and 4).⁸⁴ Moreover, in the interferon context, deeper molecular responses associate with longer disease remissions after discontinuation of therapy.^85,86 It remains to be seen whether deeper molecular responses in PV exert a more profound impact on disease progression, apropos of chronic myeloid leukemia treatment. This is an area of active study, with recent investigations reported on the clinical basis of variation in interferon response and pathways that potentiate the susceptibility of JAK2^V617F-mutant stem cells to interferon therapy, and trials that target PV earlier in the disease course. (Figure 2; Table 3).^66,71,87-89 Outstanding questions also center on the degree of VAF reduction to mitigate thrombosis risk, how closely peripheral VAF suppression reflects clonal suppression of JAK2-mutated stem cells, when to initiate therapy, and the durability of clonal suppression after achieving VAF reductions.

The JAK2^V617F discovery has elucidated the basis of the quantity of blood cell production, the unique qualities of JAK2^V617F cells, and the clonal expansion dynamics of JAK2^V617F HSCs in PV. Physicians and their patients can apply the knowledge of JAK2^V617F VAF to understand that risk is defined not by blood counts alone but on the intrinsic effects of JAK2^V617F VAF that impart risk to both thrombosis and disease progression (Table 5). The meaning of VAF reduction in PV is emerging through the analysis of prospective randomized trials where VAF was correlated with response and outcomes, and proof is mounting that reduction of VAF is tied to not only blood count control but also thrombosis risk reduction and disease progression reduction. Taken together, PV treatments that do not address clonal expansion of JAK2^V617F and therefore do not reduce JAK2^V617F VAF do not optimally address thrombosis or disease progression risk and represent missed opportunities for our patients to receive disease-modifying therapies (Table 5).

PV therapy, like chronic myeloid leukemia therapy, requires years of exposure to achieve molecular responses, and thus off-target effects, tolerability, immune suppression, and cancer development are additional major concerns and challenges (Table 5). Long-term exposure to agents that place JAK2^V617F HSCs under a selection stress without ultimate clonal suppression, translate to selection for mutational events that drive progression or AML (Tables 4 and 5).⁶⁵ The study of JAK2^V617F pathophysiology has yielded many new candidates to target clonal expansion in PV. Therapies that upregulate hepcidin,⁹⁰ suppress key inflammatory signaling pathways,^26,91,92 activate p53,^93,94 inhibit HIF,⁹⁵ target DNA repair pathways,^30,96 or facilitate JAK2^V617F stem cell extinction^97,98 are exciting developments for PV and will complement JAK2 inhibitor and interferon alfa–based therapies. Our clinical and research goals should be to incorporate these new targets in the treatment of PV, to use the knowledge of JAK2^V617F VAF to better prognosticate and monitor clonal dynamics, and to aspire to provide reliable, safe, and timely clonal suppression (Table 5).

The authors are honored to prepare this review and thank the global community of patients, clinicians, and scientists for participating in PV research.

A.R.M. receives research support through the MPN Research Foundation and National Institutes of Health (NIH), National Heart, Lung, and Blood Institute (NHLBI) grant 1R44HL162386-01A1. H.K. is supported by NIH, NHLBI grant T32HL007525 and the Nancy and Lou Grasmick Scholarship Fund. B.N.R. is supported by an HTRS Mentored Research Award (supported by an educational grant from CSL Behring) and NIH, National Center for Advancing Translational Sciences grant UL1TR002489.

The authors dedicate this work to Richard Silver, Jerry Spivak, and Josef Prchal, who, through their devotion to intellectual curiosity, mentorship, and science, have inspired generations of MPN physician-scientists.

Contribution: A.R.M. and B.N.R. designed the research and wrote and edited the manuscript; and H.K. designed the research, edited the manuscript, and created the figures.

Conflict-of-interest disclosure: A.R.M. received consulting fees from PharmaEssentia and Protagonist. B.N.R. has received consulting fees from PharmaEssentia, CTI BioPharma, and Incyte. H.K. declares no competing financial interests.

Correspondence: Alison R. Moliterno, Division of Hematology, Department of Medicine, Johns Hopkins University School of Medicine, Ross Research 1025, 720 Rutland Ave, Baltimore, MD 21205; e-mail: amoliter@jhmi.edu.