The role of JAK2 inhibitors in MPNs 7 years after approval

Myeloproliferative neoplasms (MPNs) are clonal hematopoietic disorders including 3 main diseases: essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (MF).¹  Blast phase²  and post-PV and post-ET MF³  can occur. The same driver mutations of JAK2, CALR, and MPL genes are variably present and are mostly mutually exclusive in MPNs.¹  In PV, JAK2 mutations cover almost the whole mutational profile, with JAK2V617F being present in 95% to 97% of patients⁴  and exon 12 mutations in the remaining.⁵  In ∼5% of ET and 5% to 10% of primary MF, MPL mutations have been identified.⁶  Almost 70% of JAK2/MPL-negative ET and primary MF patients carry CALR mutations.⁷ 

JAK2, MPL, and CALR mutations have been functionally validated and are sufficient to engender an MPN phenotype in mice.^8,9  All these mutations have a gain-of-function effect on JAK2/signal transducer and activator of transcription (STAT) signaling, and studies of gene expression profile unequivocally showed that activated JAK2 signaling is seen in all MPN patients, irrespective of the driver mutation.¹⁰  The JAK/STAT pathway is the key signaling cascade in MPNs.

After the discovery of these mutations, pharmaceutical companies and investigators developed small-molecule inhibitors of JAK2 (JAKis) for the treatment of MPNs.¹¹  JAKis inhibit JAK2 and STAT phosphorylation, resulting in reduced cellular proliferation and induction of apoptosis in cell lines.¹¹  On 16 November, 2011, the US Food and Drug Administration (FDA) approved ruxolitinib (RUX) for the treatment of intermediate and high-risk MF, including primary MF and post-PV/post-ET MF. Subsequently, on 4 December, 2014, the FDA approved RUX to treat patients with PV having an inadequate response to or being intolerant of hydroxyurea (HU).

In clinical trials, responses obtained with JAKis are independent of the underlying driver mutation, as expected on the basis of the mutations’ effect on the JAK/STAT pathway and of JAKis’ mechanism of action per se.^12,13  However, the efficacy of JAKis in MPNs is partial, partly because the targeted pathway is required for normal hematopoiesis and because of the absence of specificity of JAKis for the mutated counterpart. A correlation between different genotypes and clinical phenotype has been shown in MPNs,^3,14-17  with a potential effect on JAKi efficacy. Furthermore, the JAK2V617F mutation has been described in cases of clonal hematopoiesis of indeterminate potential,^18,19  bearing clinical relevance. Healthy individuals (ie, not MPN diagnosed) carrying JAK2V617F have an increased risk of thrombosis.¹⁸  This potentially has implications on prognosis and therapy of MPNs and needs to be assessed in the future. Additional mutations acting outside the JAK-STAT pathway and, as a consequence, outside the direct JAKi effect, can affect gene expression, thus contributing to an aberrant transcriptional output. These can occur at diagnosis or be acquired during the disease course as a sign of genetic instability.^20,21  Evidence indicates that having >1 high-molecular-risk mutation (ie, mutations in ASXL1, EZH2, IDH1/2, or SRSF2) confers a detrimental prognosis in MPNs.^22,23  RUX has proven effective in MF patients carrying a high-molecular-risk profile^23,24  but, despite this, does not seem to fully overcome its negative impact on survival. Finally, host effect²⁵  and multiple pathways, including phosphatidylinositol 3-kinase (PI3K)/AKT/mTOR and RAF/MEK/extracellular signal-regulated kinase, cooperate and intersect with JAK-STAT, carrying weight in MPN pathogenesis and constituting potential therapeutic targets.

JAKis developed in MF include RUX,^26,27  fedratinib (FED),^28,29  pacritinib (PAC),^30,31  and momelotinib (MMB),^32,33  whose activity in early clinical trials is reported in Table 1 for comparison. Among these, only RUX entered clinical practice in MF.³⁴  All randomized MF studies with JAKis included patients with intermediate and high-risk primary MF and post-PV and post-ET MF. The primary end points were spleen volume response (SVR) (35% reduction by MRI) and symptom response (50% reduction using different symptom scores).

PV and ET studies with RUX (Tables 2 and 3) included patients with HU intolerance/resistance with the aim to define the activity as second-line treatment.^35-39  No other JAKis were studied on a large scale in the PV/ET setting.

While acknowledging the impossibility of performing direct comparisons of trial results, we believe that confronting patient characteristics across trials can be of interest. Median age is similar between studies (ranging from 63 to 69 years). The rate of post-PV and post-ET MF (vs primary MF) is variable, ranging from 35% in JAKARTA-1 (FED vs PBO)²⁸  to 55% in COMFORT-1 (RUX vs PBO).²⁶  This must be taken into account when comparing results, because survival of post-PV and post-ET MF is different from that of primary MF.⁴⁰  Splenomegaly is a key feature in MF (and its reduction is the primary end point of all studies); median baseline spleen size varied from 12 (PAC, PERSIST-1)³⁰  to 16 cm (RUX, COMFORT-1)²⁶  from the left costal margin. The median baseline value of hemoglobin is ∼10 to 11 g/dL across studies. Some studies, such as the COMFORT trials, included only patients with platelet counts >100 × 10⁹/L; thrombocytopenia below this threshold is a known negative prognostic factor in primary MF. Differently, PAC was studied also in patients with a platelet count <100 × 10⁹/L (PERSIST-2).

Overall, patients entering these trials were in advanced phases of MF, and most received HU before enrollment. Looking at trial results (Table 1), RUX, FED, and MMB seem very active on splenomegaly, all JAKis tackle symptomatology, PAC appears to be particularly attractive for cytopenic patients, MMB alleviates anemia, and FED is extremely active as a second line after RUX failure.

On the basis of a 60% hematocrit control rate and a 40% SVR rate, RUX is an effective second-line therapy in PV.^35,36,39,41  The RESPONSE trials included mainly HU-failing patients; however, RUX efficacy/safety data have also been captured in interferon-failing patients.⁴²  Despite these very positive results, some words of caution are needed regarding PV.

First, hematocrit is a surrogate end point of thrombosis.⁴³  Nonetheless, it is of interest that in the 80-week follow-up analysis of the RESPONSE study, the rate of all thrombotic events was 1.8 × 100 patient-years of exposure to RUX and 8.2 × 100 patient-years of exposure to standard care.³⁹  At the last 4-year follow-up update, the incidence of thrombosis was reduced to 1.2 × 100 patient-years, reverting the natural trend in PV.⁴⁴ 

Second, transferring the definition of HU resistance/intolerance⁴⁵  into clinical practice is problematic. For instance, in a patient receiving few phlebotomies along with HU to maintain hematocrit levels <45%, is RUX necessary? A significantly higher rate of thrombosis was found in patients treated with HU plus ≥3 phlebotomies per year compared with HU plus 0 to 2 phlebotomies per year (20.5% vs 5.3% at 3 years) in a retrospective analysis.⁴⁶  Conversely, phlebotomy intensity did not affect thrombosis in a prospective clinical trial that was, however, designed with a different end point.⁴⁷  Hence, there is no definitive evidence that the number of phlebotomies is an issue in PV. On the other hand, patients with progressive spleen enlargement, symptomatology, or signs of myeloproliferation despite HU, or those who are intolerant of HU, are candidates for RUX, as these aspects impact events and quality of life.^48,49 

Third, the design of available trials cannot address the possible role of RUX in delaying or preventing post-PV MF, although progression to MF has been described under RUX without an exceedingly long follow-up.^44,50 

In HU-resistant/intolerant ET, RUX achieved clinically meaningful and durable reductions in platelet and leukocyte counts and improvements in ET-related symptoms, as reported in a phase 2 trial with a long follow-up.³⁷  However, when compared with current second-line treatments, RUX was equivalent to standard therapy (MAJIC-ET trial).³⁸  It has thus become questionable to use RUX in this setting, as ET (especially if diagnosed according to the WHO 2017 criteria)¹  is a benign disease, the risk of MF evolution being quite low, and other opportunities such as interferons or anagrelide being an option for these patients.^51,52 

A 50% palpatory SVR rate with RUX was documented in ∼60% of patients with advanced MF and was fairly long-lasting (median duration of response 3.2 years).⁵³  A debate on the role of spleen size in MF was consequently opened. The prospective COMFORT trials revealed a 1.14-fold higher risk of death for each additional 5 dL of spleen volume at baseline and a possible role for SVR under RUX as surrogate end point for survival.⁵⁴  In fact, any SVR upon RUX was associated with a better prognosis compared with a <10% reduction. Patients who achieved reductions of ≥25% had a prolonged survival compared with those who had no change or an increase in spleen volume.⁵⁴  To achieve this end point, patients should receive the most appropriate dose of RUX, taking into account that dose modifications are routinely performed especially in the first 4 to 6 months of treatment. In advanced MF, the magnitude of SVRs achieved with RUX is what we currently expect, meaning that new drugs or combinations must overcome this result. An international consensus will help shape what will be required of upcoming therapies in terms of spleen response, such as the abrogation of splenomegaly in those with a huge spleen and longer duration of spleen response.

The second achievement of RUX is symptom control. Beyond the obvious benefit in terms of patients’ quality of life (well documented with ad hoc questionnaires in clinical trials), reducing symptoms means eliminating variables impacting on outcomes.^55,56  In addition, symptom burden is guided by cytokine activation.⁵⁷  One of the most relevant JAKi-derived insights into MPN pathobiology is the understanding of the role of inflammation. First, a huge variety of cytokines are activated by driver mutations and induce MF complications, eventually affecting survival.⁵⁸  Second, RUX reverses proinflammatory cytokines.⁵⁹  Third, clinical improvement with RUX correlates with a reduction in plasma levels of several proinflammatory cytokines.⁶⁰  Although this anticytokine effect occurs mainly through JAK1 inhibition, itacitinib, a selective JAK1 inhibitor, achieved symptom reduction in 28% to 35% of advanced MF patients (60% intermediate-2 and high risk Dynamic International Prognostic Scoring System [DIPSS]),⁶⁰  similarly to RUX, thus not producing an additional advantage at that stage. The future challenge will be to switch off cytokines, using JAKis or itacitinib, or to minimize their pathogenetic relevance, using PRM-151 (a recombinant form of pentraxin-2 acting as anti-fibrotic agent),⁶¹  earlier in disease development when the overt effect of cytokines is not yet dominant.

Beyond the effect on symptoms, cytokine activation induces progressive bone marrow fibrosis, which is relevant for outcomes in MF and PV.^62,63  RUX induces stabilization and reduction of bone marrow fibrosis in 32% and 16% of MF, respectively.⁶⁴  Abrogation of bone marrow fibrosis has been proposed as a potential end point for new agents or combinations. Although this is clearly intriguing, such an end point will require the evaluation of many biopsy samples, and we argue that the sole reduction of fibrosis, without the achievement of a normalization of blood counts and splenomegaly, is not a critical end point for advanced MF patients.

Concerning the effect of RUX on JAK2V617F mutational load, the majority of MF patients obtained a mild effect, with one-third having a 20% reduction.⁶⁵  In PV, the mean changes from baseline ranged from −12% to −40%, with very few obtaining a complete molecular response and 32% a partial one.⁶⁶  The net advantage of reducing the JAK2-mutated clone in MPNs is a challenging question for future research. To date, in MF, there is no clear evidence that reducing the clone implies a control of clinical parameters. In PV, on the contrary, 80% of patients with significant allele burden reductions had a spleen response without, however, a significant correlation with cell count improvements. This limits the potential usefulness of monitoring molecular dynamics and suggests a slow or absent correlation between genotype and clinical phenotype improvements. Another similar discrepancy has been described with imetelstat (a telomerase inhibitor) in ET.⁶⁷  This molecule obtained 100% hematological responses and 88% partial molecular responses without affecting vascular events (4 severe episodes in 18 ET patients treated), which still remain the most relevant problem in PV and ET.

The effect of RUX on survival generated some debate. These are the available data on this topic: the 5-year update of the COMFORT trials showed a median survival of 5.3 years in RUX-treated patients, which was significantly improved compared with control data.⁵³  Furthermore, standard therapies in historical matched cohorts allowed lower median survival rates compared with RUX in advanced primary MF.^68,69 

The widespread use of RUX in MF and, to a lesser extent, HU-resistant/intolerant PV has highlighted some clinically noteworthy aspects complementing the data emerging from clinical trials.

Anemia and thrombocytopenia are among the expected dose-dependent on-target effects of RUX. Both COMFORT studies in advanced MF have shown a drop in mean hemoglobin levels in virtually all patients during the first weeks/months of treatment, with 51% receiving at least 1 transfusion of packed red blood cells and ∼5% requiring RUX discontinuation for this reason.^53,64  Among 163 IPSS intermediate-1 MF patients treated with RUX, anemia (any grade) was documented in 54% of patients (grade 3/4 in 24.5%), suggesting a somewhat milder hematologic toxicity profile in patients with early disease.⁷⁰  Although disease-related anemia negatively impacts survival and is a risk factor included in most prognostic scoring systems, a post hoc analysis showed that new or worsening postbaseline RUX-induced anemia did not affect survival.⁷¹  Other JAKis such as PAC, MMB, or NS-018 (a JAK2/Src inhibitor currently under investigation showing 56% SVR)³⁴  seem to be less toxic for hematopoiesis, with a lower rate of anemia and thrombocytopenia.

How to improve anemia in MF (disease or RUX induced) is still not clear. Transfusions and appropriate iron chelation (envisaged for allogeneic stem cell transplant [SCT] candidates) might be a good option, and erythropoiesis-stimulating agents have produced some results.⁷²  Ongoing phase 2 trials with sotatercept (NCT01712308)⁷³  and luspatercept (NCT03194542), acting on the late-stage maturation of erythroblasts, are designed to evaluate anemia responses in MF patients with and without concomitant RUX and could prove instrumental in the cure of anemia in MF patients, regardless of JAKi treatment. An alternative approach under investigation combines RUX and itacitinib in MF patients in whom cytopenias preclude dose optimization (NCT03144687). Preliminary information on these different approaches will become available in 2018/2019 and will help in the design of appropriate randomized phase 3 trials.

Evidence before JAKi development indicated that infections accounted for ∼10% of deaths in MF patients,⁵⁵  with higher IPSS risk category and splenomegaly being the most impactful variables for infections.⁷⁴  In COMFORT-2, infections often involved the urinary tract (25%) or lung (13%). Frequent cases of herpes zoster infection (11%) and septic shock (8%) and few cases of tuberculosis (1%) were reported. A list of case reports or series concerning infections under RUX has recently been published.⁷⁵  In PV, the RESPONSE trials reported similar and relatively low rates of bacterial infections in the RUX and in the control arm.^35,36  Recently, SIE-ELN guidelines on RUX therapy have been published,⁷⁶  without any restriction to the use of RUX but recommending caution, specific monitoring, or prophylactic measures in patients with at least 1 risk factor.

Prospective trials in PV and MF have disclosed a significant rate of nonmelanoma skin cancers (NMSCs) in RUX-treated patients.^35,36,65,77  Importantly, both COMFORT and RESPONSE trials excluded enrollment of patients with active malignancy within the preceding 5 years, with the exception of specific skin cancers. The incidence rates for skin basal and squamous cell carcinoma were 2.7 vs 3.9 (COMFORT-1) and 6.1 vs 3.0 (COMFORT-2) per 100 patient-years of exposure in the RUX-randomized group and control arm, respectively. In PV, the exposure-adjusted rate of NMSCs in RUX-randomized patients was 5.1 (week 208) vs 4.4 (week 80) per 100 patient-years and 2.6 (week 208) vs 2.0 (week 80) per 100 patient-years in the crossover group. Rates were higher among patients with a history of NMSCs.^39,44  In a small series of RUX-treated patients, B-cell lymphomas developed in 4 out of 69 patients (5.8%) with RUX compared with only 2 of 557 (0.4%) in the control group.⁷⁸ 

Several studies performed before the introduction of JAKis found that MPN patients are at increased risk of developing second malignancies. A study on 1915 consecutive patients with MPNs found that the risk of developing a lymphoid neoplasm was 2.79-fold higher than that of the control population.⁷⁹  Similar evidence derives from a Danish population-based study (n = 7,229),⁸⁰  and from the Surveillance, Epidemiology, and End Results Program registry (n = 20 250), the latter reporting an incidence of second cancers (all sites) of 15.2% at 10 years.⁸¹  In the Surveillance, Epidemiology, and End Results Program series, second malignancies overrepresented with respect to the general population were both hematological (myeloid and lymphoid) and nonhematological, the latter including the following sites: oral cavity, brain, esophagus, lung and bronchus, skin (melanoma), uterus and ovary, prostate, kidney, and thyroid. Neoplasms found in patients treated with RUX seem to be mostly related to the skin. However, careful monitoring and data collection within international registries would be useful. In the RUX-treated patients we follow, a dermatological visit is prescribed at baseline and regularly during follow-up to detect early lesions. Furthermore, we educate our patients with respect to sun-protective behavior. In case of multiple skin cancer recurrence, we evaluate treatment discontinuation (especially in case of suboptimal response).

Primary resistance to JAKis is a clinical issue for few patients (<2% to 5%),^26,27  whereas losing a spleen response is frequent during follow-up.⁵³  The explanation for this is unclear. We offer some considerations: (1) The dose is of importance, and we recommend using the highest safe dose to control splenomegaly, since reducing RUX dose results in a loss of benefit in terms of SVR. (2) Some patients have progressive disease in terms of evolution into accelerated or blast phase, and in many of these cases, splenomegaly increases. (3) The biology of JAKi resistance is of great interest but remains without a clear translation into clinical decision-making to date. Some recently reviewed mechanisms of resistance have been identified in MPN cell lines and murine models.⁸²  These can be recapitulated as acquired JAK2 kinase mutations, persistence of JAK-STAT signaling due to JAK family heterodimer formation, or protective cytokine effects.

A big challenge for the health care system is treatment sustainability, especially considering the cost of new therapies. In the pre-JAKi era, HU, steroids, and red blood cell transfusions were relatively cheap, whereas splenectomy was a risky surgical intervention with a high economic impact. SCT has huge health care costs, and its indication is currently unchanged with respect to the pre-JAKi era. Some pharmacoeconomic analyses on RUX have been recently reviewed.⁸³  Overall, studies reported similar incremental efficacy of RUX with a plus of 1.04-2.51 quality-adjusted life years (QALYs). The incremental cost per QALY ranged from $40 000 to $54 000 per QALY saved.

The great interest within the scientific community, the widespread dissemination of information, and the availability of a drug rapidly and tangibly affecting quality of life has increased patient awareness with regards to MPNs, resulting in the creation of patient associations and awareness groups that may possibly affect health care policies with regards to MPNs in individual countries.

Today, there is evidence for the use of RUX in advanced phases of MF disease, ie, higher risk primary MF and post-PV and post ET MF according to the (D)IPSS^55,84  and the MF secondary to PV and ET prognostic model (MYSEC-PM),⁵⁶  respectively. Responses are mostly obtained within the first 6 months of therapy; afterwards, any patient permanently failing RUX should be considered for investigative clinical trials. With regards to this, prediction of RUX response would be of relevance, especially in a precision medicine era. Subgroup analysis of the COMFORT studies suggested that any MF patient can benefit from RUX.⁸⁵  However, harboring ≥3 mutations was inversely correlated with spleen response and time to treatment discontinuation.²⁴  A large retrospective study of 408 patients found that higher IPSS risk categories and delay in RUX initiation are predictors of lower RUX response rates.⁸⁶  This could suggest starting RUX earlier in the disease course, as conveyed by the National Comprehensive Cancer Network guidelines that advocate for the use of RUX in low and intermediate-1 risk MF patients with troublesome symptoms.⁸⁷  Prospective, single-arm trials evaluating the earlier use of RUX are available.^70,88  The JUMP study disclosed a high level of RUX efficacy (64% SVR at week 24) and a good safety profile in intermediate-1 risk MF.⁷⁰  In the management of intermediate-1 risk MF, we follow the recent European LeukemiaNet guidelines,⁵¹  using RUX as first line in the case of highly symptomatic splenomegaly.

SCT indications in MPNs remain unchanged in the JAKi era; ie, patients with intermediate-2 and high risk MF and with intermediate-1 risk disease and high-risk features are to be considered for a SCT.^89,90  Because SCT and RUX indications tend to coincide and RUX can improve transplant-specific risk factors, such as splenomegaly and symptomatology, most patients are treated with RUX before undergoing SCT. Several studies on small numbers of patients have confirmed the feasibility of this approach.^91,92  Results of an ongoing prospective transplantation-based trial in MF are awaited. This recently initiated study (NCT03333187) sets out to assess the relative benefit of RUX without subsequent SCT vs RUX as a “bridge” to SCT, according to donor availability. The impact of RUX-induced responses at the time of SCT has been recently studied retrospectively.⁹³  A favorable outcome of MF patients experiencing clinical improvement with JAKi prior to SCT has been observed. This suggests a prognostic relevance of achieving a good “JAKi-inducible” response before SCT with RUX or alternative JAKis. Finally, in the case of RUX failure, outcome is dismal but can be improved by including SCT in the therapeutic program.⁹⁴ 

The “from now on” issue in MF is the treatment of patients after RUX failure. We have no definition of this condition, and we need a consensus for the future development of clinical trials. If we consider 35% reduction of spleen size by magnetic resonance imaging a satisfactory end point in clinical practice, then every patient who does not reach such a SVR or loses it during follow-up should be defined as RUX failure. On the basis of the already provided numbers, 60% to 70% of RUX-treated patients end up failing treatment at one point during follow-up. All patients in this condition are potentially candidates for other JAKis such as FED or PAC (if approved) or clinical trials (other JAKis, combinations, or other targeted therapies) or SCT. Among these patients, a subset discontinuing RUX undergoes clonal evolution (mostly acquiring ASXL1 mutations) with a very dismal outcome of 6 months.⁹⁵  Another condition to be considered as RUX failure is the progressive increase of blast cells in the peripheral blood or bone marrow. Having 5% to 9% blast cells confers a similar outcome to having accelerated phase disease (10% to 19% circulating blast cells).⁹⁶  However, RUX seems to improve survival in patients with 5% to 9% blast cells and not in those with 10% to 19% blast cells, a difference that needs to be substantiated in prospective trials.

How to cure, delay, or prevent RUX failure should ideally be answered by upcoming trials. The progressively improving understanding of the pathobiology of MF has prompted further exploration of alternative therapeutic targets (eg, those addressing epigenetics, bone marrow fibrosis, checkpoint inhibition, and cell cycle regulation/apoptosis) recently reviewed.⁹⁷  A number of RUX-based combination therapies have been successfully evaluated in preclinical models and are now being actively pursued in the clinic (ongoing JAKi-based combination trials listed in Table 4).

The introduction of JAKis in clinical practice has raised the bar of treatment goals in MF and PV. Spleen/symptom responses and impact on survival are the most relevant achievements in MF, similarly to hematocrit control in PV. Careful safety monitoring is useful in clinical practice with skin neoplasms and infectious disease screening protocols. Patients, physician-investigators, basic scientists, and pharmaceutical companies must collaborate to improve currently obtained results.

This work was supported by the Associazione Italiana per la Ricerca sul Cancro (Milan, Italy), Special Program Molecular Clinical Oncology 5x1000 (grant to Associazione Italiana per la Ricerca sul Cancro Gruppo Italiano Malattie Mieloproliferative, project 1005). The study was also funded by Fondazione Regionale Ricerca Biomedica (Milan, Italy; project 2015-0042; Genomic profiling of rare hematologic malignancies, development of personalized medicine strategies, and their implementation into the Rete Ematologica Lombarda clinical network), Fondazione Matarelli (Milan, Italy), Fondazione Rusconi (Varese, Italy), and AIL (Associazione Italiana contro le Leucemie-linfomi e mieloma) Varese ONLUS (Organizzazione Non Lucrativa di Utilità Sociale).

Contribution: F.P. and M.M. contributed to the conception, writing, and revision of the work.

Conflict-of-interest disclosure: F.P. serves on advisory boards for Sanofi, Gilead, Janssen Pharmaceuticals, Celgene, Novartis, Bristol-Myers Squibb, and Roche. M.M. declares no competing financial interests.

Correspondence: Francesco Passamonti, Dipartimento di Medicina e Chirurgia, University of Insubria, 21100 Varese, Italy; e-mail: francesco.passamonti@uninsubria.it.