Managing myelofibrosis (MF) that “blasts” through: advancements in the treatment of relapsed/refractory and blast-phase MF

MF is a type of myeloproliferative neoplasm (MPN) that originates from dysregulated clonal hematopoiesis. MF is recognized due to its distinct molecular profile (ie, presence of Janus kinase 2^V617F [JAK2^V617F], CALR, and myeloproliferative leukemia virus oncogene mutations), hematologic hypo- or hyperproliferation, and capacity to transform into acute myeloid leukemia (AML). Initial treatment in MF is tailored to prognostic risk, splenomegaly, mitigating burdensome symptoms, and maintaining blood counts. The symptom burden reduction, quality of life improvement, reduction in splenomegaly, stabilization of bone marrow fibrosis, and increased overall survival with ruxolitinib have solidified its role as a first-line agent in symptomatic or intermediate- to high-risk MF. However, the clinical challenges accompanying patients who are initially refractory, relapsed, or intolerant to ruxolitinib or those who progress to blast-phase disease are becoming increasingly relevant as MF patients live longer. In this article, we detail our treatment paradigm for the challenging clinical scenarios of ruxolitinib-relapsed/refractory MF and blast-phase myeloproliferative neoplasm (BP-MPN).

MF can arise as a secondary process from antecedent essential thrombocythemia or polycythemia (termed secondary MF), or it can arise de novo as primary MF. Over the disease course, individuals with MF may experience worsening of cytopenias, infections, transfusion dependence, progressive splenomegaly, or transformation to AML. Mortality for MF patients is most often due to progression to secondary AML (17%), progressive bone marrow failure/transfusion dependence (10%), thrombosis (7%), infections (6%), bleeding (3%), or other secondary malignancies (2%).¹ 

MF exhibits many common overlapping genetic mutations with AML (Figure 1).^2-4  This is particularly true for progressive and BP-MPN disease, where mutational profiles have been found to have high concordance with those seen in de novo AML, post-MPN AML, and myelodysplastic syndrome (MDS)^5-7 ; complex of karyotype or multiple genetic mutations are risk factors for poor prognosis and disease progression.⁸  Genomic alterations in BP-MPN are 3-fold more abundant compared with in chronic-phase MPN (P < .001).⁹ 

Available for treatment outside of the clinical setting in the United States since 2011, ruxolitinib remains the only Food and Drug Administration (FDA)–approved therapy for MF. In the recently developed National Comprehensive Cancer Network (NCCN) guidelines, ruxolitinib is suggested for use in low-risk symptomatic patients as well as intermediate- and high-risk patients, with consideration of bone marrow transplantation and clinical trial if applicable.⁹  Although originally approved for use due to 2 large phase 3 clinical trials showing symptom and spleen response, additional benefits may include overall survival benefit¹⁰  and stabilization or reversal of bone marrow fibrosis.¹¹  In an independent, blinded review by 3 hematopathologists for consensus-based World Health Organization bone marrow histology grading, MF patients treated with ruxolitinib showed a greater chance of reversal or stabilization of bone marrow fibrosis compared with patients treated with the best available therapy.¹¹  However, this bone marrow effect was not observed in all patients and is still of uncertain clinical significance. The survival advantage for upfront ruxolitinib has been found independent of a clinical trial setting, with analysis of 2 commercial claims databases in the United States finding a median overall survival rate of 30 months in ruxolitinib-treated MF patients compared with 22 months for other treatments (N = 6982; hazard ratio, 0.7; 95% confidence interval, 0.6-0.8).¹²  Overall, MF patients are likely living longer in the era of first-line ruxolitinib treatment. Despite these significant advancements, ruxolitinib is not curative for MF, and ∼50% of individuals discontinue ruxolitinib after 3 years, most commonly due to loss or diminished response; also, ruxolitinib has not been found to significantly alter the rate of transformation to AML.¹³  Rather, 35% of patients on ruxolitinib will experience clonal evolution associated with decreased overall survival (6 vs 16 months).¹⁴ 

Inadequate response to ruxolitinib represents a spectrum of intolerance and refractory/relapsed disease features related to the spleen and/or symptoms leading to inability to continue ruxolitinib for its clinical indication.

Generally, being ruxolitinib refractory suggests that one has lost a spleen or symptom response to single-agent ruxolitinib. Features associated with the greatest risk of loss of response to ruxolitinib are (1) a mean titrated ruxolitinib dose of ≤10 mg twice daily, (2) delay in ruxolitinib treatment start ≥2 years from diagnosis, (3) transfusion dependence, (4) platelet counts <200 × 10⁹/L, (5) grade 3 bone marrow fibrosis, (6) large splenomegaly (spleen palpable ≥10 cm below the left costal margin), and (7) International Prognostic Scoring System Intermediate-2 or high risk.¹² 

For patients who can tolerate continued ruxolitinib dosing but have loss of symptom or spleen response, optimizing their ruxolitinib dose can often lead to reinduction of ruxolitinib benefit. For those with dose-limiting anemia leading to suboptimal ruxolitinib dosing, we typically combine ruxolitinib with erythropoietin or danazol. In a multicenter, phase 2 trial of ruxolitinib and danazol, this combination resulted in stabilization of anemia in some patients (55.5% of patients on prior ruxolitinib had hemoglobin stabilization or improvement, with 88.9% having stable or increased platelets).¹⁵  The combination of erythropoietin analogs and ruxolitinib seemed more effective, with an overall response rate of 87.9% and a median response duration of 31 months.¹⁶  Generally, patients will undergo an improvement in erythropoietin levels after ruxolitinib initiation,¹⁷  and therefore, we will typically assess erythropoietin level after a patient is on a stable dose of ruxolitinib and use a threshold of <500 mU/mL (as per current NCCN guidelines¹⁸ ) to help guide our selection of either danazol or erythropoietin analog. Outside the use of ruxolitinib, the threshold of an erythropoietin level <125 mU/mL has been recommended¹⁹  due to its association with an anemia response.²⁰ 

We also consider symptom-directed therapies that can be used safely in combination with ruxolitinib for individuals who have symptom needs despite the best tolerable dose of ruxolitinib therapy. Although this topic is beyond the scope of this review, we generally consider specific treatment of erythromelalgia symptoms (ie, aspirin), pruritus (ie, H1 and H2 blockers, aspirin, UV light, cromolyn sodium, hydroxyzine, and topical steroids), and fatigue (ie, exercise, yoga, psychostimulants, selective and serotonin reuptake inhibitors).

Although clearly defined response criteria have been established for ruxolitinib,²¹  inadequate ruxolitinib response remains a spectrum of inconsistently defined features that lack consensus in clinical trials (Table 1).^22-25  In a clinical setting, generally, we use a patient-centric definition of inadequate ruxolitinib response consisting of ruxolitinib administration for at least 2 months (if achievable) with at least 1 of the following: (1) loss or lack of spleen or symptom response that negatively impacts quality of life, (2) sustained red blood cell transfusion requirement or dose-limiting anemia despite anemia-directed combinational therapies (as mentioned above), or (3) need to adjust dose of ruxolitinib to subtherapeutic dosing for thrombocytopenia, sustained or severe bleeding, or hematoma. Although ruxolitinib dosing typically correlates to ruxolitinib response,¹²  patients may experience symptomatic benefit at doses as low as 5 mg twice daily.

For all patients with higher-risk disease features, we begin the discussion of allogeneic stem cell transplantation as soon as possible in the disease course (Figure 2), a conversation that typically occurs before a patient is ruxolitinib refractory or relapsed.²⁶  Given worsened outcomes after ruxolitinib discontinuation, patients who are candidates would undergo transplant at the time of best response in their disease course and before ruxolitinib-relapsed or -refractory disease.²⁷  Clinical trial options should be reviewed at the first signs of ruxolitinib-relapsed disease.

Alone or in combination with ruxolitinib, alternative commercially available treatments are available for MF with variable degrees of response. For individuals with early MF experiencing elevations in blood counts, interferon can be considered, although interferon can worsen symptom burden and risk of depression. Long-acting and more tolerable interferon agents are under investigation that have shown promising clinical activity in MF, including pegylated-interferon-α2a.²⁸  Hypomethylating agents (HMAs) can be considered for alternative therapy, although their response rate seems to be highest in accelerated and blast-phase disease and carries risk of high-grade myelosuppression. In a general population of MPN patients, azacitidine (AZA) showed a response rate of 25% at 5 months, although 29% experienced grades 3 to 4 neutropenia.²⁹  The combination of HMAs with ruxolitinib seems promising compared with single-agent ruxolitinib, with clinical trials of combinational ruxolitinib and AZA (starting after cycle 4) showing high response rates (72%) and spleen response (79%).³⁰  Due to this excellent response rate, the combination of ruxolitinib and HMA is our preferred treatment out of the commercially available treatment options. The histone deacetylase (HDAC) inhibitor panobinostat, which is currently FDA approved for use in multiple myeloma, can also be considered for use in MF populations on an off-label basis. In a phase 2 trial of panobinostat, the response rate was 36%, with spleen reduction in 34%.³¹  Trials of the combination of panobinostat and ruxolitinib are ongoing (NCT01693601). Lenalidomide as a single agent has a 23% response rate with an improvement in anemia (19%) and/or spleen size (10%), although dose- and treatment-limiting effects can include myelosuppression (88% with at least grade 3 hematologic toxicity).³²  We generally avoid the combination of ruxolitinib with lenalidomide due to excessive toxicities, because almost one-half of patients undergoing this combination in a clinical trial setting experienced dose interruptions and discontinuation within 3 months.³³ 

There is also evidence that suggests that a ruxolitinib holiday may be effective in reacquisition of drug sensitivity in some ruxolitinib-refractory patients.³⁴  Heterodimerization with other JAKs occurring during persistent JAK activation has been shown to be reversible after a drug holiday in vitro models.³⁴  In a case series of 13 MF patients who experienced an inadequate response or loss of response to ruxolitinib, the majority of patients experienced improvement in constitutional symptoms and spleen size with ruxolitinib rechallenge (92% and 69%, respectively). The mean duration of time off of ruxolitinib therapy was 8 weeks (range, 3-176 weeks).³⁵  Although the optimal duration of holiday is unknown and this strategy has not yet been tested in a clinical trial setting, a ruxolitinib holiday could be considered for some patients who can tolerate a lapse in active treatment. However, such strategies should be considered cautiously due to the risk of symptom rebound after abrupt cessation of JAK inhibitor therapy.

Splenectomy can be considered in instances of prominent anemia, where the suspected cause is massive splenomegaly or when a patient is severely symptomatic due to splenomegaly with no alternative treatment options (ie, likely have failed at least 2 JAK inhibitors, including ruxolitinib). Given the increased morbidity and mortality associated with this procedure, we typically are very selective for a population that has limited comorbidities and lacks alternative treatment options.³⁶  Splenic radiation can be helpful in select cases, although it is generally avoided due to risk of prolonged myelosuppression.³⁶ 

In addition to JAK-targeted novel therapeutic agents, numerous ongoing clinical trials are investigating other molecular targets either as single agents or in combination with ruxolitinib (Table 2). We typically consider a clinical trial option first before using commercially available drug alternatives or combinations.

Since the approval of ruxolitinib in 2011, no new JAK inhibitors have made it through the FDA approval process for MPN, although there remains promising ongoing investigation. Fedratinib, a selective JAK2 inhibitor, elicited responses in 47% (400 mg) and 50% (500 mg) of patients with Intermediate-2 or high-risk MF and showed superiority to placebo for control of splenomegaly and symptoms in the JAKARTA I Trial.³⁷  In the JAKARTA II Trial, fedratinib showed activity in the second-line setting for individuals who had previously been on ruxolitinib.³⁸  In November 2013, fedratinib was placed on an FDA clinical hold due to concerns of Wernicke’s encephalopathy. However, in a recent analysis of the 8 of 877 treated patients with suspected cases of Wernicke’s encephalopathy, only 1 case was found to have Wernicke’s encephalopathy, and it predated the patient’s fedratinib use.³⁹  Given this, the FDA lifted the clinical hold for fedratinib, and it is now undergoing clinical development.⁴⁰  Pacritinib was evaluated in the PERSIST 1⁴¹  and PERSIST 2²⁴  studies and is currently undergoing additional evaluation in a randomized, phase 3 dose optimization study (PAC203; NCT03165734). In the fall of 2017, the SIMPLIFY 1⁴²  and SIMPLIFY 2²⁵  trials of momelotinib were placed on hold due to inability to meet primary end points. Clinical trials of these alternative JAK2 inhibitors have shown reinduction of eliciting symptom benefit, and these agents remain a good treatment option in the ruxolitinib-refractory setting.

With the success of therapeutic targeting of the JAK-STAT axis, focus has turned to drugging downstream signaling pathways that are dysregulated in MF. These treatments, used either as single agents or in combination with ruxolitinib, target ligands of pathway activation, including the hedgehog, phosphatidyl 3-kinase (PI3K), proto-oncogene serine/threonine-protein kinase, cyclin-dependent kinase 4/6, polyadenosine 5′-diphosphate-ribose polymerase (PARP), aurora kinase, SMAC, and heat shock protein pathways. Other specific targets under investigation include antifibrosing agents (PRM151), epigenetic modification (HDAC inhibitors), telomerase inhibition (imetelstat), and sympathicomimetic signaling (mirabegron).

Investigations of immune modulation in MF carry great potential, including trials of nivolumab⁴³  (NCT02421354) and durvalumab⁴⁴  (NCT02871323) in primary and secondary MF and pembrolizumab⁴⁵  (NCT03065400) in chronic, accelerated, and blast-phase MF.

Some individuals will be unable to tolerate ruxolitinib for alternative reasons (ie, other side effects, severe allergic reactions, and personal preference). We typically consider patients who are unable to tolerate a dose of ruxolitinib above 5 mg twice daily due to nonhematologic complications intolerant to ruxolitinib as well. In these individuals, our treatment algorithm is the same as in patients who relapsed on ruxolitinib as described above.

With outcomes worse than standard AML, salvage therapies in accelerated and BP-MPN rely on the tolerability and availability of cytoreductive therapies and allogeneic stem cell transplantation. In a recent review of BP-MPN, median overall survival was 3.6 months, a statistic that has not altered over the last 15 years.⁴⁶  The formal International Working Group for MF Research and Treatment definition of blast phase is defined as persistent elevation in peripheral blood or bone marrow blasts of 20% or more.⁴⁷  A second but closely related entity with shortened overall survival is accelerated-phase MPN, defined as individuals with an elevation of peripheral or bone marrow blasts between 10% and 20%.⁴⁸ 

Clinical clues that the patient may be progressing to accelerated or blast phase include the rapid onset of severe cytopenias, transfusion dependence, and constitutional symptoms. Our initial workup in this setting includes a repeat bone marrow biopsy with karyotype to help understand the extent of blast cell involvement. The reevaluation of next generation genetic sequencing, if not done recently, is also critical in this setting to determine whether targetable mutations exist.

Despite the worsened survival of blast-phase patients compared with AML patients as mentioned above, current treatment options for blast-phase MF mirror closely the current treatment paradigm in AML (Figure 3).

Although the antiquated “7 plus 3” induction with salvage allogeneic stem cell transplant is still the most widely used initial treatment option for fit individuals, the rapid growth of available therapies in AML has created the opportunity for more targeted and tolerable agents in BP-MPN outside of a clinical trial setting. Many of these therapies have been investigated in populations relevant to blast-phase MF, including therapy-related AML and AML with dysplastic changes.

Allogeneic stem cell transplantation in the blast-phase population remains the only curative treatment applicable to a select population of young and otherwise fit individuals. In general, fit patients are considered to have a biologic or physiologic age of <60 years, adequate left ventricular ejection fraction, near-normal pulmonary function testing, serum creatinine <1.5 mg/dL, serum bilirubin <2 mg/dL, liver function <2 times the upper limit of normal, and a body weight of 95% to 145% of ideal.⁴⁹  However, even in this setting, outcomes are still suboptimal. In a recent analysis, survival rates at 5 years were 10% for patients who had achieved complete remission or incomplete count recovery and underwent subsequent allogeneic stem cell transplant.⁴⁶  Despite this, transplant still remains the superior treatment option, with only 13% 5-year survival for similar patients who were not transplanted.⁴⁶ 

For individuals who are not eligible for upfront traditional induction chemotherapy with allogeneic stem cell transplantation or clinical trials, we recommend an individualized approach incorporating disease burden, the best matching of treatment options via genetic sequencing, and prioritization of patient goals

HMAs (ie, AZA and decitabine), either alone or in combination with ruxolitinib as well as in combination with low-dose cytarabine, have also been used successfully in the setting of BP-MPN. As mentioned above, the combination of HMA with ruxolitinib appears to have very good response rates even in the setting of progressive disease. In 1 small case series of patients with BP-MPN, individuals experienced reduction in symptoms and splenomegaly and were able to attain disease stability with tolerable adverse events.⁵⁰ 

There are 4 new approved agents that may have clinical utility in a BP-MPN population. The clinical agent CPX-351 (Vyxeos) is a new liposomal formulation that delivers a fixed ratio of daunorubicin and cytarabine, and it is approved for newly diagnosed treatment-related AML or AML with myelodysplasia-related changes. In patients 60 to 75 years of age with newly diagnosed treatment-related AML or AML with myelodysplasia-related changes, CPX-351 seemed to be more tolerable than standard cytarabine and daunorubicin induction.⁵¹  Given that BP-MPN seems genetically related to MDS, CPX-351 may represent a more tolerable agent and potentially, a therapeutically active agent in elderly BP-MPN patients.

Venetoclax (Venclexta) has received an FDA breakthrough therapy designation for use in combination with low-dose cytarabine in treatment-naïve elderly patients with AML who are ineligible for intensive chemotherapy. MF is a disease characterized by the overexpression of the antiapoptotic BCL-2 family of proteins (eg, BCL-XL and MCL-1).⁵²  However, previous BCL-2 inhibitors (ie, Obatoclax)⁵²  exhibited no significant single agent clinical activity in patients with MF at the dose and schedule evaluated. Despite this, venatoclax seems to be a promising combination treatment with potential efficacy in BP-MPN.

Enasidenib is a new agent approved for treatment of adults with relapsed or refractory AML with an isocitrate dehydrogenase-2 (IDH2) mutation.⁵³  In this population, enasidenib was found to have an overall response rate of 40.3%, with median response duration of 5.8 months. Given that the frequency of IDH2 mutations is higher in the post-MPN AML (31%) and BP-MPN (21.6%; up to 44% if JAK2^V617F is comutated) populations than in AML populations (12%)⁵⁴  and that IDH mutations are associated with leukemic transformation (P < .01) and worsened survival (P = .01),⁵⁵  this drug may have great clinical utility in the setting of blast-phase and post-MPN AML. The IDH1 inhibitor, ivosidenib, also may be a potential effective agent in post-MPN AML. Results from the phase 1 dose escalation and expansion single-arm trial of ivosidenib among a population of relapsed or refractory AML patients showed complete remission in 21.6% of treated patients with some complete molecular remissions.⁵⁶  This study led to the FDA approval of ivosidenib for relapsed and refractory AML patients with the IDH1 mutation in July 2018. Until clinical trials show efficacy upfront in the use of IDH mutant populations, we suggest continuing to wait for the relapsed or refractory setting before initiating use in BP-MPN.

Gemtuzumab ozogamicin is approved for use in CD33-positive adult AML patients. Gemtuzumab showed efficacy in both the phase 2 single-arm study of AML patients in first relapse (ie, MyloFrance-1⁵⁷ ) and the randomized phase 3 study of AML patients who were not candidates for intensive chemotherapy (ie, AML-19⁵⁸ ). Although MF patients are not typically CD33 positive, acute panmyelosis with MF, a specific subtype of MF, has been found to display this cellular marker.⁵⁹  Additionally, other studies on MF patients have found that spleens of MF patients can contain populations of CD33 malignant hematopoietic stem cells.⁶⁰  This suggests that, for those who are not candidates for intensive therapy or who do not have alternative treatment options and are CD33 positive, gemtuzumab should be considered.

For many patients, however, low-intensity clinical trials, supportive care, including transfusions for symptomatic support, or hospice may be appropriate. Emphasis should be given to the patient’s wishes regarding duration and intensity of care, with a frank discussion of end-of-life considerations, including caregiver burden. The value of palliative care should also be discussed with patients in efforts to assist with symptomatic support.

Over the 8 years since the initial approval of ruxolitinib, the first FDA-approved agent for the treatment of MF patients, our understanding and clinical expectations of JAK2 inhibition have deepened. The challenging clinical spectrum of ruxolitinib initial resistance, loss of response, and intolerance has become increasingly common. Alternative JAK inhibitors, novel compounds either alone or in combination with ruxolitinib (ie, inhibitors of the hedgehog, PARP, PI3K, and aurora kinase pathways), and immunotherapies are being investigated in the after single-agent ruxolitinib setting. MF progression to blast phase is associated with an aggressive course and very poor prognosis, with allogeneic stem cell transplant representing a risky but optimal treatment modality for those who are candidates. In the nontransplant setting, an individualized approach is optimal that takes into consideration disease features, targeted therapy where appropriate, and prioritization of patient goals.

The authors thank the many collaborators on the MPN Quality of Life Study Group for their continued efforts on behalf of progressing the care of patients with MPNs.

Ruben A. Mesa, Mays Cancer Center, University of Texas MD Anderson, University of Texas, 7979 Wurzbach Rd, San Antonio, TX 78249; e-mail: mesar@uthscsa.edu.