Myelofibrosis: Genetic Characteristics and the Emerging Therapeutic Landscape Open Access

William Dameshek (1900–1969) is credited for coining the concept of myeloproliferative disorders (1951; ref. 1), which are now referred to as myeloproliferative neoplasms (MPN; ref. 2). The original MPN members included primary myelofibrosis (PMF), polycythemia vera (PV), essential thrombocythemia (ET), chronic myeloid leukemia, and Di Guglielmo's syndrome (erythroleukemia; ref. 1). Pre-Dameshek PMF luminaries include Gustav Heuck (1854–1940) who first described PMF (1879; ref. 3) and recognized its association with bone marrow fibrosis (BMF), extramedullary hematopoiesis (EMH), and osteosclerosis; Max Askanazy (1865–1940; ref. 4); and Herbert Assmann (1882–1950; ref. 5). Pseudonyms for MF used in the past include agnogenic myeloid metaplasia (6), chronic idiopathic myelofibrosis (7), and myelofibrosis with myeloid metaplasia. In 2006, the International Working Group for Myelofibrosis Research and Treatment (IWG-MRT) selected “PMF” as the preferred term (8). In 1939 (9), Vaughan and Harrison underscored the relationship between PMF, PV, and ET, in terms of their origination from a common mesenchymal reticulum cell, a view subsequently shared by others (10, 11) and ultimately led to the formal description of the MPN concept by Dameshek (1).

Dameshek's aforementioned concept of MPN was genetically ratified in 2005 by the seminal discovery of a Janus kinase 2 (JAK2 located on chromosome 9p24) gain of-function mutation (JAK2V617F; a G to T somatic mutation at nucleotide 1849, in exon 14, resulting in the substitution of valine to phenylalanine at codon 617) in PV, ET, and PMF (12–15). Additional MPN driver mutations have since been described and include JAK2 exon 12 described in JAK2V617F-negative PV (16); CALR (calreticulin; located on chromosome 19p13.2; refs. 17, 18) and MPL (myeloproliferative leukemia virus oncogene; located on chromosome 1p34; ref. 19). Among these mutations, JAK2 is the most frequent with frequencies of approximately 98% in PV (95% JAK2V617F and 3% JAK2 exon 12), 50% to 60% in ET, and 55% to 65% in PMF (20, 21). CALR and MPL mutations are often absent in PV, except rare exceptions (22, 23), while their frequencies in ET are estimated at 20% to 25% and 3% to 4%, respectively, and in PMF at 20% to 25% and 6% to 7% (21). Approximately 10% to 15% of patients with PMF or ET do not express any one of the three MPN driver mutations and are operationally referred to as being triple-negative, although that might not be the case during higher sensitivity testing (21, 24). MPN driver mutations have also been reported in other myeloid malignancies, including myelodysplastic syndromes with ring sideroblasts associated with marked thrombocytosis (MDS-RS-T; 50% frequency; refs. 25, 26).

JAK2 and MPLW515K/L/A/R and S505N mutations are believed to directly activate JAK2-STAT, resulting in clonal myeloproliferation that is cytokine independent or hypersensitive (19, 27). Frameshift CALR mutations mostly include type 1 (52-bp deletion in exon 9) or type 2 (5-bp insertion in exon 9), and less frequently a myriad of type 1-like or type 2-like variants (28). The precise mechanism of mutant CALR-induced myeloproliferation is less clear but one possibility includes mutant CALR binding to the extracellular domain of MPL in the endoplasmic reticulum, leading to dimerization and transfer to cell surface and activation of JAK-STAT (29). Mutant CALR-induced mouse models have suggested a primary effect on platelet production that is followed by development of MF (30). Although the central role of JAK-STAT activation in MPN has been highlighted (29, 31), the particular concept is confounded by the coexistence of an inflammatory state in MF with aberrant cytokine expression and the fact that activated JAK-STAT is a nonspecific phenomenon in cancer (32). Furthermore, “targeted therapy” with JAK inhibitors has so far failed to induce selective suppression of the disease clone in MF (33).

The pathogenetic role of MPN driver mutations is highlighted by their origin at the stem cell level and the demonstration of persistent JAK-STAT activation and induction of mutant JAK2/CALR/MPL–driven MPN phenotype in mice (13, 34). Recent studies have highlighted the central pathogenetic role of persistent MPL-driven JAK-STAT activation in MPN; this is presumably accomplished, in addition to activating MPL mutations, by direct activation and altered trafficking of MPL by JAK2 V617F or binding of extracellular domain of MPL by mutant CALR that begins in the endoplasmic reticulum and leads to dimerization of immature receptors and their translocation to the cell surface for activation (29, 35, 36). It is currently assumed that the phenotypic differences between MPN variants are in part contributed by differences in the conformations of specific cytokine receptors that lead to distinct signaling outcomes for EPOR and MPL and further modified by interactions with other cooccurring mutations, including those of epigenetic regulators, and their order of acquisition (29, 34). Additional mechanisms of phenotype diversity associated with MPN-driving mutations include variations in signal intensity (often related to JAK2V617F allele burden) and specificity of downstream signals, including STAT5, STAT1, and STAT3 activation (37). Phenotype-wise, JAK2V617F, versus mutant CALR, is associated with older age, higher hemoglobin level, leukocytosis, lower platelet count and increased risk of thrombosis, while the latter was associated with younger age, male sex, higher platelet count, lower hemoglobin level, lower leukocyte count and lower incidence of thrombotic events (17, 18, 38, 39); type 2 versus type 1 CALR mutations were associated with higher platelet count (40). Patients with CALR-mutated PMF are also younger and present with higher platelet count and lower frequencies of anemia, leukocytosis, and spliceosome mutations (41). Recently published mouse studies appear to recapitulate the phenotypic differences seen among patients with different driver mutations (42, 43). Regardless, the underlying mechanisms that enable single mutations to result in different MPN phenotypes remain not fully understood (44–47).

Other mutations in PMF with frequencies of 10% or higher include ASXL1 (additional sex combs-like 1), TET2 (TET oncogene family member 2), SRSF2 (serine/arginine-rich splicing factor 2), and U2AF1 (U2 small nuclear RNA auxiliary factor 1; refs. 48–50); the latter has been associated with severe anemia and thrombocytopenia in PMF (50) and treatment response to imetelstat (51) while all three mutations (ASXL1, SRSF2, U2AF1) have been associated with inferior survival (52–54). Other mutations who are less frequent in chronic phase disease but with significantly higher frequency in blast phase MPN include IDH1/IDH2 (isocitrate dehydrogenase 1 and 2; refs. 55, 56), TP53 (tumor protein p53; refs. 56, 57), DNMT3A (DNA cytosine methyltransferase 3a; ref. 58), and LNK mutation (59, 60). Other infrequent mutations reported in MPN include SF3B1 (splicing factor 3B subunit 1), which are characteristic and more frequent in MDS-RS and MDS-RS-T (25), EZH2 (enhancer of zeste homolog 2), which were shown to be prognostically relevant (61), Casitas B-lineage lymphoma proto-oncogene (CBL), which are more frequent in juvenile and chronic myelomonocytic leukemia (62) and SETBP1 (SET binding protein 1), which are more frequent in chronic neutrophilic leukemia and atypical chronic myeloid leukemia (63, 64).

The pathogenetic role of mutations other than JAK2/CALR/MPL in MPN is much less understood but believed to involve cooperation with driver mutations, leading to downstream disruption of epigenetic (e.g., ASXL1, TET2, EZH2, IDH1, IDH2, DNMT3A), RNA splicing (e.g., SRSF2, U2AF1, SF3B1), or transcriptional (TP53, IKZF1, NF-E2, CUX1) regulation, which might facilitate disease progression and leukemic transformation (56, 65). The recent demonstration of TET2, ASXL1, and DNMT3A mutations in “normal” elderly individuals has added to the complexity regarding their precise pathogenetic contribution (66, 67). More recent observations in MF include identification of STK11 as a tumor suppressor and that loss of LKB1/STK11 leads to stabilization of HIF1α and thus might promote disease progression (68), impaired megakaryocyte maturation associated with reduced GATA1 expression that might stem from ribosomal deficiency (69), and identification of aurora kinase A (70, 71), peroxisome proliferator-activated receptor gamma (PPARγ; ref. 72) and cyclin-dependent kinase 6 (CDK6; ref. 73), as therapeutic targets in MF.

Among the JAK2 mutation–related MPNs (MF, PV, ET), MF is uniquely characterized by more intense bone marrow and splenic tissue stromal changes, including reticulin/collagen fibrosis and neoangiogenesis (74). The mechanisms of such intense stromal reaction are not fully understood but might involve abnormal proliferation of fibroblasts, endothelial, and other mesenchymal cells mediated by inflammatory cytokines derived from clonal megakaryocytes or mononuclear cells, as well as in situ cell interactions involving microvascular endothelial cells and mesenchymal cells (75–81). Whether or not some of the latter stromal cells are part of the underlying disease clone is currently debated although the presence of their increased proliferation has been demonstrated and in certain cases phenotypically characterized (82–85). The potential role of fibroblasts in bone marrow fibrosis is supported by their morphologic and phenotypic resemblance to hematopoietic stem cells and monocytes and their participation in the production of extracellular matrix components (86). Several studies in MF have suggested increased expression of fibroblast growth factors or their receptors and pathogenetic interaction between megakaryocytes and fibroblasts (75–77). The reactive nature of bone marrow fibrosis in MF is further supported by clinical observations that document its complete resolution under effective targeted therapy of the disease-initiating clone (51, 87).

Over the last several years, we have been involved in the development of both the 2008 (88) and 2016 (89) World Health Organization (WHO) classification system for MPNs. According to these consensus guidelines, PMF is operationally subclassified into “prefibrotic” and “overtly fibrotic” variant (90–92). Diagnosis of overt PMF requires the presence of three major criteria and at least one minor criterion: major criteria include (i) presence of grade ≥2 reticulin/collagen bone marrow fibrosis, associated with megakaryocyte proliferation and dysmorphia, (ii) presence of JAK2/CALR/MPL or other clonal markers, or absence of evidence for reactive bone marrow fibrosis, and (iii) absence of evidence for another myeloid neoplasm; minor criteria include the presence of (i) anemia, (ii) leukocytosis, (iii) palpable splenomegaly, (iv) increased serum level of lactate dehydrogenase (LDH), and (v) leukoerythroblastosis; the latter indicates presence of nucleated red cells, immature granulocytes and/or dacryocytes (teardrop-shaped erythrocytes) in the peripheral blood. The diagnostic criteria for prefibrotic PMF are almost identical to those of overt PMF, with two exceptions: (i) the first major criterion does not require the presence of grade ≥2 reticulin/collagen bone marrow fibrosis, and (ii) leukoerythroblastosis was not listed as a minor criterion. Because more than 80% of patients with PMF harbor mutations other than JAK2/CALR/MPL, additional mutation screening by next-generation DNA sequencing (NGS) might help confirm clonality in triple-negative cases (93). About 15% of patients with ET or PV develop a PMF-like phenotype over time, referred to as post-ET or post-PV MF. The diagnosis of post-PV or post-ET MF should adhere to criteria published by the IWG-MRT; these criteria require documentation of (i) an antecedent WHO-compliant diagnosis of PV or ET, and (ii) presence of grade ≥2 reticulin/collagen bone marrow fibrosis, as well as presence of at least 2 minor criteria, including (i) anemia, (ii) leukoerythroblastosis, (iii) increasing splenomegaly, (iv) development of constitutional symptoms, and, in case of post-ET MF, (v) increased LDH.

Prominent clinical features in MF include anemia (attributed primarily to ineffective erythropoiesis), hepatosplenomegaly (attributed to EMH and cytokine-mediated organ congestion), constitutional symptoms including fatigue, night sweats, and low-grade fever, progressive cachexia with loss of muscle mass, bone pain, splenic infarct, pruritus, nonhepatosplenic EMH, thrombosis and bleeding (94). Consequences of hepatosplenic EMH include portal hypertension that might lead to variceal bleeding or ascites and those of nonhepatosplenic EMH include spinal cord compression, ascites, pleural effusion, pulmonary hypertension or extremity pain. It is currently assumed that aberrant cytokine production by clonal cells and host immune reaction contribute to PMF-associated bone marrow stromal changes, ineffective erythropoiesis, EMH, cachexia, and constitutional symptoms (95). Causes of death in PMF include leukemic progression that occurs in approximately 20% of patients (96).

Beginning in 2009, a number of prognostic models in PMF have been described and summarized in Table 1 (52–54, 97–99). The most recently developed prognostic models in PMF incorporate mutations and/or cytogenetic information: MIPSS70, MIPSS70+ version 2.0 (MIPSSv2) and GIPSS (52–54). MIPSS70 (mutation-enhanced international prognostic scoring system for transplant-age patients) utilizes mutations and clinical variables (54); MIPSSv2 (the karyotype-enhanced MIPSS70) utilizes mutations, karyotype and clinical variables (52); GIPSS (the genetically-inspired prognostic scoring system) is based exclusively on mutations and karyotype (53).

MIPSSv2 incorporates 5 genetic and 4 clinical risk factors (52); the five genetic risk factors include very high risk (VHR) karyotype (4 adverse points), unfavorable karyotype (3 points), ≥2 high molecular risk (HMR) mutations (3 points), presence of one HMR mutation (2 points), absence of type 1/like CALR mutation (2 points); the four clinical variables in MIPSSv2 include constitutional symptoms (2 points), severe anemia, defined by hemoglobin levels of <8 g/dL in women and <9 g/dL in men (2 points), moderate anemia, defined by hemoglobin levels of 8–9.9 g/dL in women and 9–10.9 g/dL in men (one point) and circulating blasts ≥2% (one point; Fig. 1). MIPSSv2 includes five risk categories: very high risk (≥9 points); high risk (5–8 points); intermediate risk (3–4 points); low risk (1–2 points); and very low risk (zero points); in patients age 70 years or younger, the corresponding median survivals (10-year survival rates) were 1.8 years (<5%), 4.1 years (13%), 7.7 years (37%), 16.4 years (56%) and “median not reached” (92%; Fig. 1). In instances where cytogenetic information is not available, MIPSS70 (54) performs as well as MIPSSv2, whereas GIPSS offers a prognostic model that is exclusively dependent on mutations and karyotype (53). Most recently, RAS/CBL mutations in PMF were associated with poor response to ruxolitinib therapy, poor prognostic features and inferior survival; in the latter regard, however, incorporation of such information to MIPSSv2 did not show additional value (100, 101).

Allogeneic hematopoietic stem cell transplant (AHSCT) currently remains the only treatment modality in MF that can prolong life and, in some cases, cure the disease (102). However, transplant-related complications are not trivial and might result in death or substantial morbidity in about half of the cases (103). Therefore, for the individual patient, the risk of AHSCT must be balanced against expected survival without AHSCT. On the other hand, current drug therapy in MF, including use of JAK2 inhibitors, is mostly palliative and has not been shown to modify disease natural history or prolong survival (Table 2; refs. 33, 104–108). Figure 1 outlines our current treatment algorithm that is based on aforementioned risk stratification model, MIPSSv2.

There is no evidence to support the value of specific drug therapy in asymptomatic patients with MIPSSv2 “low” or “very low” risk disease, whose expected 10-year survival rates were reported at 50% and 86%, respectively; expected survival in such patients was retrospectively estimated to be superior in the absence of AHSCT as first-line therapy (109, 110). Therefore, observation alone is a reasonable treatment strategy in such patients, in the absence of treatment-requiring symptoms. AHSCT is the preferred treatment of choice for MIPSSv2 “high” or “very high” risk disease, where 10-year expected survival rates without transplant might be as low as 10% and <3%, respectively (Fig. 1; refs. 109, 110). In transplant-ineligible patients with high/very high-risk disease, as well as in those with symptomatic low/intermediate risk disease, clinical trial participation might be the most appropriate treatment choice, considering the current dearth of disease-modifying drugs.

Choice of drugs for combatting MF symptoms is based on the primary clinical indication and is outlined in Table 2 and Fig. 1. For anemia, we consider androgen preparations, prednisone, immunomodulatory drugs (IMiDs; thalidomide, lenalidomide, pomalidomide), or danazol, depending on tolerability of potential side effects for the individual patient (111). For symptomatic splenomegaly, our first-line drug of choice is hydroxyurea (112) and secondarily JAK2 inhibitors (e.g., ruxolitinib, fedratinib). Ruxolitinib is our first-line drug of choice for constitutional symptoms while involved-field radiotherapy is most effective for symptomatic non-hepatosplenic EMH or localized bone pain. Splenectomy might be necessary for drug-refractory spleen-associated symptoms. Anemia response rates to each one of the aforementioned drugs are in the vicinity of 15 to 25% and response durations average about one to two years; lenalidomide works best in the presence of del(5q31) (87). We are less enthusiastic about the use of radiotherapy for hepatosplenic EMH (danger of treatment-induced protracted thrombocytopenia; ref. 113) and the use of erythropoiesis stimulating agents (ESA; refs. 114, 115) or luspatercept (116), because of inadequate efficacy (112).

Ruxolitinib and fedratinib are now FDA approved for use in MF and are considered effective treatment options for countering splenomegaly and constitutional symptoms in hydroxyurea-refractory MF, with response rates estimated at about 30% to 50% (117–120). However, treatment with these two drugs is complicated by treatment-emergent anemia and thrombocytopenia (120). Longer-term experience with ruxolitinib therapy in MF has disclosed a high rate of treatment discontinuation rate (92% after a median time of 9.2 months) and the occurrence of severe withdrawal symptoms during treatment discontinuation (“ruxolitinib withdrawal syndrome”), characterized by acute relapse of disease symptoms, accelerated splenomegaly, worsening of cytopenias and occasional hemodynamic decompensation, including a septic shock-like syndrome (105, 121, 122). In addition, several reports have now associated ruxolitinib with sometimes fatal serious opportunistic infections (123–125), and as of late, poor immune antibody response to COVID-19 vaccination (126). It is important to underscore the fact that neither ruxolitinib nor other JAK2 inhibitors in MF possess antitumor activity and none of them has been shown to reverse bone marrow fibrosis or induce cytogenetic or molecular remissions. Instead, their mechanism of action is based on their broad activity in suppressing inflammatory cytokines.

JAK2 inhibitors other than ruxolitinib that have undergone phase III clinical trials include fedratinib, momelotinib, and pacritinib (127). Fedratinib was recently approved for use in patients intolerant or resistant to ruxolitinib, with approximately a third of patients responding, regardless of the reason for ruxolitinib discontinuation (128). Compared with placebo, the spleen response rate with fedratinib 400 mg/day (36% vs. 1%) was similar to that seen with ruxolitinib (JAKARTA-1; ref. 129); spleen response rates for pacritinib were 19% versus 5% and 18% versus 3%, compared with BAT, which excluded (PERSIST-1) (130) or included (PERSIST-2; ref. 131) JAK inhibitor therapy, the latter involving patients with platelet count of <100 × 10⁶/L. Momelotinib was compared with ruxolitinib (SIMPLIFY-1) with similar spleen response rates (26.5% vs. 29%; ref. 132) and compared with BAT (SIMPLIFY-2) in patients failing treatment with ruxolitinib with meager spleen response rates (7% vs. 6%; ref. 133). The toxicity profiles were noticeably different for the aforementioned JAK inhibitors: fedratinib (Wernicke's encephalopathy, anemia, thrombocytopenia, gastrointestinal distress and elevations in serum liver function tests and pancreatic enzymes); pacritinib (cardiac events, severe diarrhea, nausea, thrombocytopenia, anemia and hemorrhage); momelotinib (peripheral neuropathy, thrombocytopenia, first-dose effect including dizziness, nausea, hypotension, headache, and flushing; refs. 107, 108). Among the currently available JAK inhibitors, momelotinib is uniquely identified as being potentially useful in alleviating anemia, in addition to the expected responses in spleen size and symptoms, and the particular activity has been ascribed to the drug's inhibitory activity against activin receptor type 1 (ACVR1), which upregulates hepcidin synthesis (134).

MIPSSv2-stratifed outcome analysis of AHSCT in MF was assessed in a retrospective study of 110 patients with a median follow-up of 64 months (110); approximately 89% of evaluable patients were MIPSSv2 high or very high risk and the remaining intermediate risk. The 5-year overall survival rate for the entire patient cohort was 65%, progression-free survival 60%, relapse rate 17%, nonrelapse mortality 24%, grade 2–4 acute graft versus host disease (GVHD) 45%, and chronic extensive GVHD 59%. In the particular study (110), somatic mutations did not affect outcome after AHSCT, whereas survival in patients with intermediate or high risk MIPSSv2 risk category (median not reached at last follow-up) was superior to that of patients with very high risk MIPSSv2 risk category (median 25 months; ref. 110). The better than 50% five-year survival was also apparent in our own institutional experience where neither very high risk nor unfavorable karyotype affected post-transplant survival (102). In a much larger study of 2,224 patients with MF who underwent AHSCT, 781 received myeloablative (MAC; median age 53 years) and 1,443 reduced intensity (RIC; median age 58 years) conditioning; outcome in the two groups were mostly similar in regards to engraftment period, rates of grade 2–4 acute GVHD (MAC 28% and RIC 31%), 5-year nonrelapse mortality (MAC 35% and RIC 34%), and 5-year survival (MAC 53% and RIC 51%; ref. 135). More recent studies have confirmed the value of AHSCT in older patients (136), the possibility of using family mismatched donors (137) and newer effective therapies for GVHD (e.g., ruxolitinib, ibrutinib; refs. 138–140). In a recent study of 556 transplanted patients with MF and age ≥65 years (median 67, range 65–76), followed for a median of 3.4 years, 5-year survival, nonrelapse mortality, and relapse rates were 40%, 37%, and 25%, respectively (136). Similarly favorable outcome data have also been reported in patients receiving alternative donor grafts (141). Unresolved issues with AHSCT in MF include the role of splenectomy (142–144), involved field radiotherapy (145, 146), and use of JAK2 inhibitors before and after AHSCT (147–151). A number of recent publications have addressed the issue of pretransplant management of splenomegaly in MF, including splenectomy, splenic irradiation, and use of JAK2 inhibitors (142, 144, 145, 147, 152). There is currently no consensus regarding either the need or specific treatment option in this regard and decisions are best made on a case-by-case basis (152).

In this section, we provide a synopsis of selected novel agents under active clinical investigation in patients with MF (Table 3).

CPI-0610, is an oral inhibitor of bromodomain and extraterminal domain (BET) proteins that inhibits cytokine production and promotes megakaryocytic and erythroid differentiation. In the phase II (MANIFEST) study, CPI-0610 was administered as (i) monotherapy in MF patients that were refractory or intolerant to JAK inhibitors (Arm 1, n = 43; ref. 153), (ii) add on therapy in those with suboptimal response to ruxolitinib (Arm 2, n = 78; ref. 154), or (iii) in combination with ruxolitinib in JAK inhibitor–naïve patients (Arm 3, n = 64; ref. 155); 3/14 (21%) and 13/36 (36%) transfusion-dependent patients achieved transfusion independence in Arm 1 (CPI-0610 monotherapy) and Arm 2 (CPI-0610 with ruxolitinib) cohorts, respectively (153, 154). Correspondingly, among nontransfusion-dependent patients, 13 of 22 (59%) in Arm 1 and 4 of 23 (17%) in Arm 2 had a rise in hemoglobin level of >1.5 g/dL (153, 154). Notably, none of the transfusion dependent patients in Arm 1 versus 21% in Arm 2 demonstrated spleen response. CPI-0610 in combination with ruxolitinib yielded spleen and symptom response in 63% and 59% of JAK inhibitor–naïve patients, respectively (155). Commonly reported adverse events included gastrointestinal toxicity (46%), thrombocytopenia (40%), respiratory infections (32%), and anemia (12%; refs. 153–155). Based on these findings, a randomized placebo-controlled phase III study (MANIFEST-2) with CP-0610 in combination with ruxolitinib in JAK inhibitor–naïve patients is ongoing (156).

Navitoclax is an oral small-molecule inhibitor of antiapoptotic B-cell lymphoma 2 (BCL2) family proteins, including BCL-X_L, BCL2, and BCL-W. In a phase II study with navitoclax in combination with ruxolitinib in patients with MF with suboptimal response to ruxolitinib, 34 patients were enrolled, with ongoing therapy in 17 patients (157, 158). Spleen and symptom response were reported in 27% and 30% of patients, respectively (157). Commonly reported adverse events included thrombocytopenia in 88%, which led to dose reduction in 56% of patients, diarrhea (71%), and fatigue (62%). As a next step, two phase III studies are ongoing, which include a placebo-controlled study of navitoclax in combination with ruxolitinib in JAK inhibitor treatment–naïve patients (TRANSFORM-1), and navitoclax in combination with ruxolitinib versus best available therapy in patients relapsed following JAK inhibitor therapy (TRANSFORM-2).

Luspatercept, an erythropoiesis maturating agent that binds to TGFβ superfamily ligand, reduces aberrant Smad2/3 signaling and enhances late-stage erythropoiesis, is FDA approved for myelodysplastic syndromes with ringed sideroblasts. In an ongoing phase II investigation in MF, four cohorts that included transfusion-dependent and -independent patients that were either receiving or not on ruxolitinib were enrolled (159). Among transfusion-dependent patients, 36% and 19% of patients receiving or not receiving ruxolitinib were rendered transfusion independent with median duration of response of 55 and 59 weeks, respectively (160). Most common adverse event of hypertension was noted in 13% of patients. Currently, luspatercept is undergoing phase III investigation in the INDEPENDENCE study, which includes patients with MF with transfusion requiring anemia and receiving JAK inhibitor therapy.

Imetelstat, a telomerase inhibitor, has proven safety and efficacy in a pilot study in patients with treatment-naïve and relapsed MF (51). Recently published data from the randomized phase II study in patients relapsed/refractory to JAK inhibitors showed modest activity in terms of spleen (10%) and symptom (32%) response in patients receiving imetelstat 9.4 mg/kg i.v. every 3 weeks (161). However, improvement in bone marrow fibrosis was noted in 41%, with driver mutation variant allele frequency reduction in 42% of patients, which correlated with superior overall survival (median survival; 29.9 months; ref. 161). Major toxicities included thrombocytopenia recorded in half of patients, followed by anemia (44%) and neutropenia in one-third of patients (161). Based on its selective impact on the malignant clone, a confirmatory phase II study of imetelstat 9.4 mg/kg vs. best available therapy excluding JAK inhibitors in relapsed MF is ongoing.

An oral inhibitor of lysine-specific demethylase-1 inhibitor (LSD1), which is involved in differentiation and maturation of megakaryocytes, has been shown to improve blood counts, splenomegaly, cytokine profile, mutant allele burden and fibrosis in murine models (162). In a phase II study of IMG-7289 monotherapy in patients with MF resistant to approved therapies, 62 patients have been enrolled, majority had received prior ruxolitinib (n = 56), one-third were transfusion-dependent, and 94% harbored ASXL1 mutation (163). Eighty-one percent of patients demonstrated spleen response, one-third reported symptom response, and 72% of patients had stable or >1 g/dL improvement in hemoglobin levels (163). In addition, 26% of patients demonstrated at least grade 1 improvement in fibrosis, 42% with reduction, and 50% with stability of driver and high-risk mutation allele burden.

An oral highly selective PI3Kδ inhibitor is under phase II investigation in patients with MF with suboptimal response to ruxolitinib that may arise due to persistent PI3K/AKT activation (164). Patients with MF underwent randomization to add-on parsaclisib once daily/once weekly or parsaclisib once daily while remaining on a stable ruxolitinib dose. Fifty-three patients have been treated; 33 and 20 patients received parsaclisib once daily/once weekly and once daily, respectively. At 12- and 24-week assessments, median percent spleen change was −2.3 and −15.4 at week 12; and −2.5 and −25.4 at week 24 with once daily/once weekly and once daily dosing, respectively, while median percent change in symptom score at week 12 was −14.0 in once daily/once weekly, and −39.6 in once daily (164). With once daily/once weekly and once daily dose, 18% and 30% developed grade 3 thrombocytopenia; 21% and 0% had grade 4 thrombocytopenia, while hemoglobin levels remained stable. Additional grade 3/4 treatment-related, adverse events included disseminated tuberculosis, enteritis, fatigue, hypertension, abnormal liver tests and stomatitis. Importantly, half of patients interrupted parsaclisib due to adverse events (164). On the basis of the above findings, a randomized study of add-on parsaclisib versus placebo in patients with suboptimal response to ruxolitinib is underway (165), together with first-line phase III investigation in patients with JAK and PI3K inhibitor—naïve MF (166).

At present, AHSCT remains the only treatment modality in MF that secures disease-free remission state and prolonged survival; furthermore, palliative value beyond ruxolitinib has remained out of reach for most of the drugs that are currently under investigation, and the possibility of incremental value is likely to be countered by additional side effects (Table 1). The scenario warrants urgent attention to newer therapeutic targets and, more importantly, identification of repurposable drug candidates, to accelerate the discovery process (167). In a recent edition of Cancer Research, Dutta and colleagues described their observations from JAK2V617F knock-in mice, which included upregulation of CDK6 expression in hematopoietic progenitors and in vivo therapeutic activity for CDK4/6 inhibitor palbociclib alone or in combination with ruxolitinib, including reversal of bone marrow fibrosis and reduction of spleen size (73). The observations regarding PPARγ and CDK6 as therapeutic targets are particularly noteworthy considering the current availability of FDA-approved drug antagonists (168, 169). A similar approach might be needed to address the unmet need in the treatment of blast-phase MF, which occurs in approximately 10%–20% of patients during their clinical course and where outcome under current therapy, including AHSCT, is dismal (170). Now more than ever, there is a need for genetic characterization of clinical observations followed by the relevant in vitro and in vivo animal studies that hopefully lead to development of novel therapies that target the malignant clone, and not only its secondary constitutional effects.

A. Pardanani reports personal fees from Adelphi Values outside the submitted work. J.D. Crispino reports grants from ALSAC during the conduct of the study. No disclosures were reported by the other authors.