Accelerated Phase of Myeloproliferative Neoplasms

Abstract

Background:

Myeloproliferative neoplasms (MPN) can transform to blast phase MPN (leukemic transformation; MPN-BP), typically via accelerated phase MPN (MPN-AP), in ~20–25% of the cases. MPN-AP and MPN-BP are characterized by 10–19% and ≥ 20% blasts, respectively. MPN-AP/BP portend a dismal prognosis with no established conventional treatment. Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is the sole modality associated with long-term survival.

Summary:

MPN-AP/BP has a markedly different mutational profile from de novo acute myeloid leukemia (AML). In MPN-AP/BP, TP53 and IDH1/2 are more frequent, whereas FLT3 and DNMT3A are rare. Higher incidence for leukemic transformation has been associated with the most aggressive MPN subtype, myelofibrosis MF); other risk factors for leukemic transformation include rising blast counts above 3–5%, advanced age, severe anemia, thrombocytopenia, leukocytosis, increasing bone marrow fibrosis, type 1 CALR-unmutated status, lack of driver mutations (negative for JAK2, CALR or MPL genes), adverse cytogenetics, and acquisition of ≥ 2 high molecular risk mutations (ASXL1, EZH2, IDH1/2, SRSF2, and U2AF1^Q157). The aforementioned factors have been incorporated in several novel prognostic scoring systems for MF. Currently, elderly/unfit patients with MPN-AP/BP are treated with hypomethylating agents (HA) with/without ruxolitinib; these regimens appear to confer comparable benefit to intensive chemotherapy but with lower toxicity. Retrospective studies in patients who acquired actionable mutations during MPN-AP/BP showed positive outcomes with targeted AML regimens, such as IDH1/2 inhibitors, and require further evaluation in clinical trials.

Key Messages:

Therapy for MPN-AP patients represents an unmet medical need. MF patients, in particular, should be appropriately stratified regarding their prognosis and the risk for transformation. Higher-risk patients should be monitored regularly and treated prior to progression to MPN-BP. MPN-AP patients may be treated with hypomethylating agents alone or in combination with ruxolitinib or provided with the option to enroll in rationally designed clinical trials exploring combination regimens, including novel targeted drugs, with an ultimate goal to transition to transplant.

Introduction

Aberrant proliferation of hematopoietic stem cells and pro-inflammatory cytokine release, megakaryocytic atypia, and bone marrow fibrosis are the main features of myelofibrosis (MF), the most aggressive among Philadelphia negative (Ph⁻) myeloproliferative neoplasms (MPNs). Clinically, MF manifests with anemia and thrombocytopenia, hepatosplenomegaly, extramedullary hematopoiesis, and constitutional symptoms that compromise quality of life [1, 2]. MF can present as primary (PMF) or as a complication of preceding polycythemia vera (PV) or essential thrombocythemia (ET) (called post-PV MF and post-ET MF, respectively) [3]. Dysregulated JAK/STAT signaling is a hallmark of MPN, primarily arising from JAK2 mutations in exon 14 (primarily the valine to phenylalanine point mutation at codon 617 in the pseudokinase domain of JAK2) or other activating JAK2 mutations in exon 12, the myeloproliferative leukemia (MPL) gene, and indels in exon 9 of the calreticulin (CALR) gene [4, 5].

MF has the worst prognosis among MPNs whereas PV and ET are relatively indolent subtypes [6]. The most concerning complication of PMF and post-PV MF/post-ET MF is leukemic transformation, which occurs in ~20–25% of the cases [7]. At 10 years, the estimated incidence of leukemic transformation is highest for MF (10–20%) followed by PV (~2.3–14.4%), and is lowest in ET (0.7–3%) [7, 8]. Accordingly, in a retrospective analysis that assessed the outcomes of 795 MPN patients, the annual incidence of leukemic transformation was the highest in PMF patients when compared to PV (P = 0.002) or ET (1.09% and 0.37%, respectively, P = 0.02). [9].

Progression from MPN in the chronic phase (MPN-CP) to MPN in the blast phase (MPN-BP) usually evolves through MPN in the accelerated phase (MPN-AP) [10]. MPN-CP, MPN-AP and MPN-BP are defined on the basis of blast percentage in the bone marrow or peripheral blood, i.e., 0–9%, ≥10–19% and 20% or more, respectively (Table 1) [11, 12, 13]. In MPNs, leukemic transformation is associated with dismal prognosis, and great efforts have been made to identify the risk factors and molecular events associated with disease evolution and progression to MPN-BP. In general, advanced age, severe anemia, leukocytosis, circulating blasts >2%, thrombocytopenia, advanced bone marrow fibrosis, cytogenetic abnormalities, and acquisition of ≥ 2 high-risk mutations were defined as clear risk factors for leukemic transformation, in many studies [2, 7, 14, 15, 16]. Given that MPN-AP precedes MPN-BP in most patients, the aforementioned factors can be assessed as risk predictors of transformation from CP to MPN-AP [17]. Clinical data and treatments for MPN-AP are limited because the majority of investigations have been conducted in patients with MPN-CP, and not MPN-AP/BP [18]. Clinical practice, however, has clearly demonstrated that patients with MPN-AP/BP have a dismal outcome (worse than de novo acute myeloid leukemia [AML]), which is attributed to the inherent nature of the disease and the dearth of standard management or disease-modifying treatments [19, 20]. Currently, the sole modality that confers prolonged survival and potentially cures the disease is consolidative allogeneic hematopoietic stem cell transplantation (allo-HSCT) but a limited number of patients qualify for this option [19, 21, 22, 23, 24, 25, 26, 27, 28, 29]. In this review, we aim to highlight the key concepts and findings that have emerged for MPN-AP, primarily on the basis of retrospective studies, mounting evidence from clinical practice; and the limited number of clinical trials that have assessed therapeutic schemes in this population, facing adverse prognosis and a dearth of treatment modalities heretofore.

Table 1.

MPN-CP, MPN-AP, and MPN-BP and blast counts in the bone marrow or peripheral blood.

MPN Phase	MPN Type	BM blast count	PB blast count
Chronic	PMF, PV, ET	>1–9%	>1–9%
Accelerated	PMF, PV, ET	10–19%	10–19%
Blast	PMF, PV, ET	≥ 20%	≥ 20%

MPN-CP, myeloproliferative neoplasm in chronic phase; MPN-AP, myeloproliferative neoplasm in accelerated phase; MPN-BP, myeloproliferative neoplasm in blast phase; BM, bone marrow; PB, peripheral blood; PMF, primary myelofibrosis; PV, polycythemia vera; ET, essential thrombocythemia.

Retrospective studies, risk factors, and prognostication in MPN-AP

Our team retrospectively analyzed the clinical characteristics of 370 patients with PMF (79% of the patients) and secondary MF (post-PV MF and post-ET MF) that had a median overall survival (OS) of 12 months or less, to determine the characteristics of MPN-AP and identify a set of progression-associated criteria [10]. Multivariable analysis of the putative high-risk AP characteristics associated with median OS ≤ 12 months was performed; baseline bone marrow or circulating blasts ≥10%, aberrations of chromosome 17 and platelets below 50×10⁹/L were associated with median OS of 10, 5, and 12 months, respectively. These high-risk features were also validated during the disease course; e.g., patients that were in the chronic phase and subsequently developed one of the aforementioned characteristics had similarly short survivals (median OS 12, 6 and 15 months, respectively). Moreover, the rate of leukemic transformation of patients who persistently were in MPN-CP was very low (1% at 3 years and 3 % at 10 years, P < 0.001) as compared to patients in MPN-AP. These findings clearly demonstrate that transformation to MPN-BP from an antecedent MPN transitions through the MPN-AP phase [10].

The role of increased blasts in the outcome of patients with MF was recently assessed by our group [30, 31]. On a cohort of 1316 MF patients with available peripheral and bone marrow blast percentages, we showed that patients with 4% peripheral blasts and ≥ 5% bone marrow blasts had comparable OS and clinical characteristics, and also resembled patients in AP (blasts ≥10–19%). The respective OS of patients with 0%, 1–3%, 4% peripheral blasts (and < 5% bone marrow blasts), 5–9% peripheral and/or bone marrow blasts and ≥10–19% peripheral and / or bone marrow blasts were 64, 48, 22, 24, and 13 months, respectively. Overall, 146 patients progressed to AML, with the highest rate among those with ≥10% blasts, as expected; however, patients with 5–9% blasts had an approximate twofold rate of leukemic transformation as compared to patients with lower blasts (AML incidence per 100 person-years was 7.7 and 24.7 cases for 5–9 % and ≥10% blasts, respectively, vs. 3.5 cases only for blasts ≤4%) [30, 31]. The estimated leukemia-free survival rate at 3 years was ~ 60% for those with < 5% blasts, and ~ 30% for those with ≥ 5% blasts [30]. Patients who had blasts <10% and were treated with ruxolitinib had superior overall survival compared to patients who did not receive ruxolitinib [31].

Geyer et al. conducted a retrospective study in which 92 patients with initial diagnosis of PMF, PV/ post-PV MF, and ET/ post-ET MF (median follow-up time was 11 years) and increased blasts were stratified in three categories according to blast counts: IB-1 (2–4% peripheral blood blasts and <5% bone marrow blasts), IB-2 (5–9% blasts in the bone marrow and/or peripheral blood), and AP (≥ 10–19% blasts in bone marrow or peripheral blood); the investigators also identified a control group comprising MPN patients that did not have increased blasts, leukocytosis or monocytosis [32]. The analysis demonstrated that patients in the IB-2 group (P = 0.00014) had a significantly inferior OS than both the IB-1 group and control patients; in contrast, the OS rates of group IB-1 and the controls were similar. As expected, patients in the combined IB-1/IB-2 groups (P = 0.0038) and the controls (P < 0.0001) had significantly longer OS than patients in AP [32]. The two aforementioned studies by Masarova at al. [30] and Geyer et al. [32] indicate that lowering the blast percentage threshold defining MPN-AP from 10% to 5% is justified [25, 33]. Notwithstanding designation of ≥ 20% bone marrow or peripheral blood blasts as indication of MPN-BP by the International Working Group-Myeloproliferative Neoplasms Research and Treatment (IWG-MRT) in 2013 [3], blasts in the range 10–19% have been associated with MPN-AP and clearly inferior outcomes [10, 30]; in clinical practice, the former blast range is currently applied as a criterion of advanced MPN disease and thus, could be included in the next revision of the criteria. In a retrospective study of 311 patients with PMF, aiming to assess the risk factors leading to leukemic transformation, peripheral blood blasts ≥ 3% (P < 0.0001) and a platelet count below 100 × 10⁹/L (P = 0.004) were determined to be strong and independent risk factors in both univariate and multivariate analyses [34]. When the previous factors were taken into consideration, the frequency of leukemic transformation increased from 5.8% in the subgroup with neither one of the two risk factors to 18.2% in the subgroup that had both [34]. In a recent large study including 1306 patients with PMF, Vallapureddy et al. demonstrated that peripheral blasts ≥3% were associated with increased risk of leukemic transformation (HR 3.3) in the first five years of diagnosis [35]. Other retrospective studies led to the conclusion that red blood cell transfusion dependence and high white blood cell (WBC) counts were correlated with elevated risk of transformation [36, 37].

Several prognostic models were developed to assess the outcome of PMF patients [38]. The initial International Prognostic Scoring System (IPSS) [39] relies on clinical parameters at diagnosis and was modified to the Dynamic IPSS (DIPSS) to assess prognosis based on risk factors at diagnosis and during the natural history of the disease [40]. In DIPSS, age over 65 years (1 point), hemoglobin < 10 g/dL (2 points), WBC count > 25 × 10⁹/L (1 point), peripheral blood blasts ≥1% (1 point), and constitutional symptoms (1 point) predicted poor survivals. In DIPSS, patients were stratified into 4 categories: low (0 points), intermediate-1 (1–2 points), intermediate-2 (3–4 points) and high-risk (5–6 points) [40]. The estimated hazard risk for progression to BP was 2 when the risk category progressed from low to intermediate-1, 3.8 from intermediate-1 to intermediate-2, and 3.2 from intermediate-2 to high. The risk to develop MPN-BP in the intermediate-2 and high categories was 7.8- and 24.9-fold higher than in the low-risk category, respectively [41]. On the basis of the evidence acquired from follow-up data of PMF patients, karyotype, transfusion-dependence and thrombocytopenia (platelet count <100 ×10⁹/L) were integrated in the DIPSS, thereby developing DIPSS-plus [42], which turned out to be superior in assessing the LFS compared to both IPSS and DIPSS. Thrombocytopenia and anemia were also included in the more integrated Mutation-Enhanced International Prognostic Scoring System 70 (MIPSS70) for transplant-eligible patients with PMF aged ≤70 years [43] and in the Myelofibrosis SECondary to PV and ET-Prognostic Model (MYSEC-PM) for post-PV MF / post-ET MF [44].

Cytogenetic abnormalities constitute an important risk factor in progression of MF-CP to MF-AP/BP [7]; however, cytogenetics were not included as a factor in the IPSS [39] and DIPSS [40]. Hussein et al. showed the prognostic relevance of specific cytogenetic abnormalities (e.g., +8 or complex karyotype) [45] and the favorable prognostic impact of sole 20q-, 13q- or +9 [46, 47, 48]. In a more recent study on 1002 patients with MF, the same investigators showed that 45% of the patients had abnormal karyotype; 6% of the latter subgroup had complex karyotype (3 or more abnormalities), and of those, about one third had a monosomal karyotype [48]. In a cohort of patients with PMF, the 2-year leukemic transformation rate associated with monosomal karyotype was 29.4%, and the median survival was 6 months versus 8.3% and 24 months, respectively, for complex karyotype without monosomies, demonstrating the dismal prognosis associated with certain monosomies [49]. The unfavorable karyotype category of the two-tiered cytogenetic model for PMF includes complex karyotype and one or two abnormalities, for example, +8, −5/5q-, −7/7q-, 12p-, i(17q), inv(3), and 11q23 rearrangement [42, 46] (Table 2). Notably, Tefferi et al. recently proposed a more refined three-tiered cytogenetic stratification, comprising very high-risk (VHR), unfavorable, and favorable karyotype categories [48] (Table 2). The VHR karyotype showed inferiority in LFS regardless of other risk factors, such as high-risk mutations, thrombocytopenia, and DIPSS [48]. The three-tiered cytogenetic model has been integrated in the five-tiered Mutation and karyotype-enhanced MIPSS70-plus version 2.0 [50] and the four-tiered Genetically Inspired Prognostic Scoring System (GIPSS) [51] for PMF. On the other hand, the two-tiered cytogenetic model (comprising favorable and unfavorable categories) has been incorporated in DIPSS-plus [42] and MIPSS70-plus [43]. The recent refined three-tiered cytogenetic model was developed based on a cohort of 1002 PMF patients at Mayo Clinic [48] and was further evaluated in the studies that led to the development of MIPSS70-plus version 2.0 [50] and GIPSS [51]. Validation of the three-tiered model is necessary as preliminary results from our group did not confirm the same discrimination power of the three-tiered model [52].

Table 2.

Two-tiered versus the revised three-tiered cytogenetic risk models and risk categories in PMF.

Two-tiered cytogenetic risk categories according to Caramazza et al. [46]	Cytogenetic risk category	Revised three-tiered cytogenetic risk categories according to Tefferi et al. [48]	Cytogenetic risk category
Complex karyotype Sole or two abnormalities that include:  • +8  • −7/7q-  • i(17q)  • −5/5q-  • 12p-  • inv(3)  • 11q23 rearrangement	Unfavorable	Complex karyotype without a VHR abnormality Sole abnormalities that include:  • +8  • −7q- Sole translocations not involving chromosome 1 Two abnormalities not including a VHR abnormality Single/multiple 5q− abnormalities Monosomal karyotype without a VHR abnormality Sole abnormalities not otherwise classified	Unfavorable
		Single/multiple abnormalities of:  • inv(3)/3q21  • −7  • 11q-/11q23  • 12p-/12p11.2  • i(17q) Autosomal trisomies, not including +8 or +9 (for example, +19, +21)	Very High Risk (VHR)
Normal karyotype Sole abnormalities of:  • +9  • 13q-  • 20q-    chromosome 1 translocation/duplication two abnormalities excluding unfavorable ones	Favorable	Normal karyotype Sole abnormalities of:  • +9  • 13q-  • 20q-    chromosome 1 translocation/duplication sex chromosome abnormality, including –Y	Favorable

PMF: primary myelofibrosis; VHR: very high-risk

A recent study that Marcellino et al. conducted on the impact of chromosomal abnormalities in advanced MPN showed that 43% of the patients with MPN-AP/BP gained +1q versus 9% and 2% in the MF and PV/ET cohorts, respectively [53], and +1q was recurrently detected in MPN-AP/BP [53, 54]. Notably, +1q was detected in the majority of MPN patients (84%) that harbored JAK2^V617F [53]. Furthermore, all the patients with MPN-AP/BP (33%), PMF (50%) and secondary MF (17%) exhibited loss of 17p13 (leading to TP53 deletion) versus none with PV/ET, findings clearly indicating the association of 17p deletions with advanced MPN [53].

A retrospective analysis of 649 patients with MF demonstrated that baseline bone marrow or peripheral blood blasts ≥ 5% (P = 0.02 and 0.01, respectively) and unfavorable karyotype (−5, −7, del17p, and/or complex karyotype, P = 0.04) were independent risk factors for leukemic transformation [55]. The subgroup that had one or both risk factors (80 patients, 12% of the cohort) had a median OS of 10 months and a considerably higher risk of leukemic transformation versus patients without either one of the two adverse factors; the one-year LFS was 82% compared to 98%, respectively (P = 0.001), at one year. During a 12-month follow-up, 13% (n = 10) of 80 patients who had either one of the two risk factors progressed to AML compared to 2% (n = 10) of 568 patients without the previous characteristics [55]. Acquisition of an abnormal karyotype has also been associated with higher incidence of disease progression in PV [16, 56]. In studies conducted at the MD Anderson Cancer Center, an abnormal karyotype was detected in about 14–20% of the patients with PV; in contrast, the frequency increased to 90% in the AP/BP patients, of whom 83% had complex karyotype [56]. In this study, another factor that was associated with higher risk of leukemia transformation, and later was confirmed by others, was the presence of moderate to severe bone marrow fibrosis and dysplasia [8, 16]. Expectedly, higher grades of bone marrow fibrosis correlated with advanced disease and adverse prognosis in PMF [57, 58] and have been incorporated in the MIPSS70-plus version 2.0 model [43].

Dobrowolski et al. demonstrated that correlation of leukemic transformation risk and persistent basophilia in PMF in the fibrotic stage was statistically significant [59]. Other groups showed that monocytosis was a strong adverse indicator of progression to the accelerated phase and inferior survival in patients with PMF [60, 61, 62]. A recent study conducted in 454 patients with PMF revealed that monocytosis was associated with leukemic transformation: an absolute monocyte count ≥ 1×10⁹/L and > 3×10⁹/L increased the hazard risk by two- and four-fold, respectively [62]. Furthermore, an association was found between monocytosis and older age, increased circulating blasts, higher DIPSS, and DIPSS-plus scores, and the presence of ASXL1 [62].

Radioactive phosphorus (³²P), pipobroman and busulphan, used in PV treatment, have been associated with increased risk of leukemogenesis [7, 63]. In a retrospective analysis of 83 patients with MPN followed for nearly 8 years, 14%, 30% and 4% of the patients who were treated with hydroxyurea monotherapy, hydroxyurea and busulphan, or observed only developed AML/myelodysplastic syndrome (MDS).[64]. In a national study on a large Swedish cohort of patients with MPN (n = 11,039), transformation to AML/MDS (n = 162) significantly increased with high doses of ³²P and alkylating agents (and the risk increased 2.9-fold with ≥ 2 treatments), whereas there was no association between the use of hydroxyurea and elevated risk of AML/MDS [65]. Researchers from Mayo Clinic also analyzed the effect of specific treatments on LFS and concluded that erythropoiesis-stimulating agents and danazol were both associated with significantly shorter LFS on univariate analysis [34]; however, these findings were not confirmed in subsequent studies by other researchers. Other proposed factors contributing to disease progression to MF-AP and MF-BP include high levels of circulating pro-inflammatory mediators, such as interleukin-8 and C-reactive protein (> 7 mg/L) [66, 67].

Molecular signatures of MPN-AP/BP

Approximately 90% of the patients with PMF harbor one of the three “driver” mutations. Janus kinase 2 (JAK2) V617F, calreticulin (CALR) exon 9 indels and myeloproliferative leukemia virus oncogene (MPL) W515L/K were detected in about 50–60%, 20–30% and 5–10% of the patients, respectively [68]. Nearly 10% of PMF patients have “triple negative” molecular status regarding “driver” mutations, which is a high-risk profile and has been associated with significantly worse LFS [69, 70]. In a retrospective analysis of 1581 patients with MPNs, including 428 with PMF, patients with “triple negative” status for the three driver mutations (JAK2, CALR, MPL) had significantly worse LFS compared to harboring JAK2 (HR 0.4; 95% CI, 0.2–0.7), CALR (HR 0.1; 95% CI, 0.05–0.35), and MPL (HR 0.3; 95%CI, 0.1–0.9) mutants [71]. Accordingly, OS was considerably longer in patients harboring CALR versus “triple-negative” (HR, 5.1; 95% CI) and JAK2 mutants (HR, 2.5; 95% CI) [71]. An international study of 570 patients with PMF showed that type 1 CALR mutations conferred the longest survival in the absence of ASXL1 (median 10.4 years); conversely, the shortest survival was detected in mutated-ASLX1 and unmutated-CALR cases [72]. A favorable correlation has been identified between CALR type 1 mutations and longer survival [70]. Unmutated- CALR (type 1) status is one of the factors incorporated in MIPSS70 [43] (1 point), MIPSS70-plus version 2.0 (2 points) [50], GIPSS (1 point) [51], and MYSEC-PM (2 points) [44].

A retrospective study assessing a gene panel including 27 MPN-related mutations in 182 patients with PMF showed that 81% (n = 147 patients) harbored other mutations/variants apart from JAK2/CALR/MPL [69]. The most frequent mutations were ASXL1 (36%), SRSF2 (18%), TET2 (18%), and U2AF1 (16%) [69]. SRSF2, RUNX1, CEBPA and SH2B3 were associated with higher risk of leukemic transformation (HR was 4.9, 8.7, 5.4, and 5.8, respectively) [69]. In a larger analysis of 879 MF patients comprising 2 cohorts (European study group, n = 483, and validation group from Mayo Clinic, n = 396), the most common “non-driver” mutations were ASXL1 (21.7%), TET2 (9.7%), SRSF2 (8.5%), DNMT3A (5.7%), EZH2 (5.1%), CBL (4.4%) and IDH1/2 (2.6%) [73]. The study led to the conclusion that ASXL1, IDH1/2, and SRSF2 were independently correlated with leukemogenesis; ASXL1, EZH2 and SRSF2 were independent predictors of shortened survival; and ASXL1, SRSF2, EZH2 and IDH1/2 were identified as high molecular risk (HMR) mutations [73]. The cumulative risk of transformation to MPN-BP was considerably higher, and LFS was significantly lower in the HMR group as compared to the low-risk group (no HMR mutations) [73]. The aforementioned HMR mutations were incorporated in MIPSS70 [43], and U2AF1^Q157 mutations were added to the MIPSS70-plus version 2.0 [50] after the negative prognostic value of U2AF1^Q157 mutations in PMF was demonstrated [74]. Another study illustrated the association between significantly shorter LFS and the presence of ≥2 HMR mutations (HR 6.2) [71], a criterion that was incorporated in the MIPSS, MIPSS70-plus, MIPSS70-plus version 2.0, and GIPSS models [38, 68, 75].

Moreover, an extended exome-sequencing analysis of 69 genes in more than 2,000 MPN patients demonstrated that mutations in the RAS-pathway, and spliceosome and epigenetic regulators had a strong association with MF-AP [14]; and patients harboring NRAS/KRAS mutations had a higher incidence of progression to AML [76]. A recent study on 1306 patients with PMF corroborated association of ASXL1 (HR = 2.0), SRSF2 (HR 3.0) and IDH1 (HR = 4.3) with the highest transformation risk [35]. A few studies showed that TP53 mutations were strongly associated with progression to BP for all classic MPN subtypes [16, 54, 75, 77]. Accordingly, Marcellino et al. demonstrated that advanced forms of MPNs are accompanied by chromosomal abnormalities that lead to dysregulation of TP53 [53]. TP53 mutations co-occurred with JAK2^V617F in patients with MPN-BP and increased during leukemic transformation [78]. Lundberg et al. demonstrated the strong association of TP53 loss with leukemogenesis as opposed to the presence of heterozygous TP53 at low allelic burden during MPN-CP [79]. In another recent study, multivariate analysis showed that RUNX1 mutations correlated with shorter survival in a cohort of 248 patients with MPN-BP [80].

The mutational profiles of MPN-AP and MPN-BP are similar given that AP precedes BP in the majority of the patients [10]. However, the mutational landscape of MPN-AP/BP has marked differences from that of both de novo AML and MPN-CP [17, 78]. For example, IDH1/2 and TP53 mutations have a higher frequency in MPN-BP versus MPN-CP [16, 78, 85]; and mutations in FLT3 and DNMT3A predominate in do novo AML, whereas both MPN-AP and MPN-BP are more frequently associated with mutated ASXL1, IDH1, IDH2, TET2, SRSF2 and TP53 [16, 78, 80, 81, 82, 83, 84, 85, 86]. In another deep sequencing study, more than 50% of the PV patients harbored the adverse mutations ASXL1, IDH2, and SRSF2, whereas in ET patients, IDH2, TP53, SH2B3, SF3B1, U2AF1, and EZH2 occurred more frequently, besides JAK2/CALR/MPL [87].

Canadian researchers performed targeted gene sequencing of patients with MPN-AP/BP (N=122, 14 in AP and 108 in BP) and correlated the mutational profiles with clinical outcomes [88]. Of 122 specimens that were analyzed, 34 (28%) patients had PMF, 40 (33%) had PV/post-PV MF, and 38 (31%) had ET/post-ET MF. The most common mutations were: JAK2^V617F (55%), ASXL1 (30%), TET2 (25%), SRSF2 (22%), RUNX1 (20%), TP53 (17%), DNMT3A (16%), IDH1 and IDH2 (13% each) (Fig. 1) [88]. Twenty six percent of the patients were “triple-negative” for “driver” mutations, and had the shortest latency period between MPN-CP and diagnosis of MPN-AP/BP (2.8 years), whereas patients harboring CALR had the longest time (10.9 years) [88]. Shorter OS was associated with mutated TP53 (HR = 1.89, P = 0.03) and four or more mutations (P = 0.02) [88]. OS was significantly reduced and the risk to leukemic transformation increased when two or more somatic mutations (e.g., epigenetic regulators) were detected in MPN patients [79]. Increased peripheral blasts, low serum albumin and more than three cytogenetic abnormalities were correlated with shorter survival [10, 88]. Patients in AP/BP that harbored ASXL1 were more likely to have threefold higher WBC counts (P = 0.04) [88], and patients with ASXL1-mutated MPN had significantly lower hemoglobin levels than the wild-type subgroup [79]. Furthermore, acquisition of ASXL1 along with JAK2^V617F cooperated in accelerating progression to MF in PV patients [89]. In a cohort comprising 42 patients with high-risk MF and MPN-AP/BP, the most frequent co-occurring mutations were ASXL1 and JAK2^V617F [90].

Figure 1.

Mutational landscape (excluding “driver” mutations) in patients with MPN-AP/BP (N=122, 14 in AP and 108 in BP). Data shown in the plot were reported by McNamara et al. [88].

Recently, our group showed that concomitant presence of JAK2^V617F and JAK2 “variants” was associated with increased risk of leukemic transformation in patients with MF [91]. Analysis of the twenty-eight gene panels of bone marrow specimens from 2154 patients with MPN (n = 608), AML/MPN (n = 59) and de novo AML (n = 1487) revealed JAK2 “variants” (non-V617F/non-exon 12 JAK2 variants), in 114 patients [91]. Among all tested patients, 23 (20.2%) had a co-occurring JAK2^V617F mutation. The three most frequent variants were N1108S, L393V, and R1063H across the three subgroups. JAK2 “variants” were considerably more frequent in the MPN-BP subgroup (~15 %) versus the MPN (~5%) and de novo AML (5.2%) subgroups. Patients harboring both JAK2^V617F and a JAK2 “variant” had an increased cumulative risk to progress from MPN to AML compared to those with JAK2 ^V617F only (P = 0.003) [91]. Moreover, further mutational analysis demonstrated that TP53 was found in ~51% of the patients harboring a JAK2 “variant” [91].

Therapeutic Modalities

As noted in the previous sections, MPN-AP usually precedes MPN-BP [10], and both phases are inherently associated with a poor prognosis. Historically, the median OS for MPN-BP has been in the approximate range of 3–9 months [7, 19, 24, 92]. Treatment options and management for MPN-AP/BP are largely similar, and at present, there is no standard approved therapy [25, 92]. At our institution, we usually treat MPN-AP/BP patients with hypomethylating agents (decitabine, azacitidine) in combination with ruxolitinib - ruxolitinib is typically added at the dose of 10 mg twice daily to control splenomegaly and MPN symptoms to the degree possible, and this dose/schedule is maintained despite lowering of the blood count as expected from hypomethylating agents-, which has become standard practice as it appears to provide at least equal benefit (but lower toxicity) to intensive chemotherapy. In the cases of MPN-BP patients with IDH1 and IDH2 mutations, we administer combinations with ivosidenib or enasidenib or enroll them in clinical trials. In rare instances of FLT3-mutated MPN-BP, gilteritinib is included in the regimen. Patients who can achieve complete or partial remission or return to a second chronic phase are referred for allo-HSCT. Our experience and that of other investigators have shown that early detection of targetable mutations (e.g., IDH1, IDH2) and timely initiation of the appropriate treatment play an important role in efficacious treatment of patients with advanced-phase MPN. In patients who are fit for transplantation and do not harbor targetable mutations, we recommend allo-HSCT after intensive chemotherapy and achievement of CR or return to MPN-CP [94].

Allogeneic Hematopoietic Stem Cell Transplant

The importance of accurate risk stratification of MF patients [68, 75, 93, 94] and timely referral of MF patients for allo-HSCT evaluation in all categories except for low-risk were demonstrated in a recent study [95]. Allo-HSCT is recommended in patients with intermediate-2 or high risk MF according to the IPSS, DIPSS, or DIPSS-plus criteria or intermediate-1 disease in the presence of unfavorable karyotype according to the DIPSS-plus criteria [42], high molecular risk mutations (e.g., triple-negative MPN driver mutations, mutated ASXL1, IDH1/2), transfusion-dependent anemia, and peripheral blasts > 2% [95]. The National Cancer Comprehensive Network (NCCN) recommends consideration of blasts in the range 10–19% in the peripheral or bone marrow for MPN-AP diagnosis [96]. However, a few recent studies showed the association of blasts ≥ 3% with leukemic transformation [35] and ≥ 5% with leukemic transformation and inferior survival [31–33]. As detailed in the section on risk factors of advanced phase MPN, Masarova et al. demonstrated the prognostic value of bone marrow and peripheral blood blasts ≥5% in patients with MF; the findings of the former studies clearly underscore that blasts ≥5% should be taken into consideration along with unfavorable karyotypes and high molecular risk mutations in patients with MF [30, 31].

The myelofibrosis transplant scoring system (MTSS) was recently developed to predict the outcome of patients with primary and secondary MF after allo-HSCT on the basis of pre-transplant factors [93]. Patients who can sustain intensive chemotherapy and achieve complete or partial remission or return to a second chronic phase have the option of consolidative allo-HSCT [21, 22, 24, 28, 29]. Up to the present, standardized criteria for response assessment to treatment of MPN-AP/BP have not been well established. Several studies have shown that complete remission (CR) prior to transplant was the most critical factor affecting survival in patients with MF that had transformed to AML [21, 22, 28, 29]. For example, Alchalby et al. showed that approximately three times more patients who underwent allo-HSCT in CR (69%) were alive compared to patients with residual disease (22%) after 3 years [22]. In general, there is no consensus about the depth of the response but typically, less than 5% blasts in the bone marrow and absence of peripheral blasts is considered a morphologic leukemia-free state [97]. To assess the response of patients with MPN-AP/BP to treatment, clinical investigators have been applying the European LeukemiaNet (ELN) criteria for AML published in 2017 [97], the proposed criteria for MPN-BP in 2012 [98] or both. The traditional ELN criteria assess elevation of the blast counts in the bone marrow [97], which is a hallmark of the disease. However, in advanced-phase MPN, a discordance can be found in the blast count of the bone marrow versus the peripheral blood due to extramedullary hematopoiesis. The proposed criteria for MPN-AP/BP assess both the AML and MPN components of the disease (cytogenetic/molecular responses and bone marrow fibrosis of the underlying MPN and other symptoms, such as splenomegaly) [98].

A retrospective study of 39 patients who had MPN-AP/BP and underwent allo-HSCT manifested that approximately one third of the patients achieved long-term survival, and the OS rate was ~29% at 2 years [21]. In a recent multicenter retrospective study, the OS of 551 MF patients who underwent allo-HSCT versus 1,377 non-HSCT patients were compared; allo-HSCT conferred long OS benefit to the former cohort, primarily in patients who had intermediate-1 and higher DIPSS risk scores and had been treated for more than one year [21,95]. Conversely, for the first year, the OS was longer in the non-HSCT cohort that had low-risk DIPSS score and received other treatments due to mortality related to transplant (frequently caused by graft-versus-host disease) [95]. Treatment of six patients in MPN-BP with a combination of intensive chemotherapy ([7 + 3] type; cytarabine with idarubicin or daunorubicin) and continuous ruxolitinib resulted in four responders and three patients who were able to undergo allo-HSCT [99]. Induction therapy can result in remission but the responses are not durable unless the treatment is followed by allo-HSCT [19, 23, 28, 88, 100]. Many patients are precluded from allo-HSCT due to age, poor performance status, other comorbidities, as well as lack of a suitable donor or inability to achieve a reasonable response [21, 25, 92, 95].

Hypomethylating Agents as Monotherapies and in Combination with Ruxolitinib

Hypomethylating agents or inhibitors of DNA methyltransferases, such as azacitidine and decitabine, initially as monotherapies or in combination with ruxolitinib (JAK1/2 inhibitor) have shown synergistic preclinical [78] and clinical activity [7] with comparable response rates to intensive chemotherapy but lower toxicity in patients with MPN-AP/BP [101]. In a previous study, 54 patients with MPN who progressed to MPN-BP (n = 26) or myelodysplastic syndrome (n = 28) were treated with azacitidine; the entire cohort had a median OS of 11 months, and the CR rate for the MPN-BP subgroup was 12% [102]. A retrospective analysis of 19 MPN patients (developing MPN-BP) that were treated with azacitidine showed that five patients (26.3%) achieved CR, one had a partial response, four had stable disease, and three displayed hematological improvement; since the time of BP evolution, the median cumulative survival was 9.9 months [103]. Andriani et al. conducted another recent retrospective study in 39 patients who were in MPN-AP/BP and were treated with azacitidine in the frontline setting at 10 hematologic centers in Italy [104]. Hematologic responses were noted in 24 patients: 8 (20.5%) and 7 (17.9%) achieved complete and partial responses, respectively, whereas hematologic improvement was noted in 9 (23.1%); 4 responders underwent allo-HSCT. The overall response rate was 61.5%, the entire cohort had a median OS of 13.5 months; and the median OS for responders was about 4 months longer (17.6 months) [104]. Another phase 1b study evaluated the safety and maximum tolerated dose of ruxolitinib in combination with azacitidine in a small cohort of patients comprising seven with MPN-AP (including MDS) and seven with MPN-BP. Of the 6 evaluable patients, one achieved complete remission, and one reached partial remission (MPN-BP patient), whereas two had stable disease [105]. A retrospective, multicenter study of 122 MPN patients (from eight French hospitals) who progressed to MPN-AP/BP demonstrated that the MPN-AP subgroup treated with azacitidine exhibited the longest OS in the cohort (13.6 months) [24]. In a phase 2 investigation of ruxolitinib and azacitdiine at our institution (NCT01787487), among 46 enrolled MF patients, 3 had MF-AP with 10% blasts or more; and one patient achieved clinical improvement in the spleen and Total Symptom Score (per International Working Group-Myeloproliferative Neoplasms Research and Treatment criteria) with a blast reduction to 2%, lasting for 28 months [106].

In 2015, we analyzed the efficacy of decitabine in 13 patients with MPN-AP at our institution [107]. Sixty two percent (8/13) of the patients with MPN-AP benefited from the treatment; the median duration of response was 6.5 months with a median OS of 9.7 months [107]. In another phase 2 study at our institution, 18 patients with MPN-BP were treated with ruxolitinib monotherapy, and 3/18 patients obtained CR and complete remission with incomplete hematological recovery (CR_i) [108]. In a phase 1/2 study (NCT02257138), we treated 12 patients who had MPN-BP with a combination of ruxolitinib and decitabine. We determined that 50 mg ruxolitinib twice daily with the standard regimen for decitabine (20 mg/m² for 5 days on a 4–6 week schedule) were tolerable and safe [109, 110] (Table 3). In the phase 2 study, among 18 patients with MPN-BP, treated with the recommended doses of ruxolitinib and decitabine, 61% were responders (11% CR + 50% CR_i). Median OS was 8.4 months (range, 0.4 – 42.6 months) for all the patients and 9.4 months (range 3.1–42.6) for responders [110]. A multicenter phase 1 study (NCT02076191) that was designed and monitored by the Myeloproliferative Neoplasms Research Consortium, also assessed the combination of ruxolitinib and decitabine, in patients with MPN-AP/BP (21 patients, n = 8 in AP and n = 13 in BP) [111] (Table 3). The response rate and median OS were 66.7% and 16 months in the MPN-AP subgroup; and 45.5% and 7.2 months in the MPN-BP subgroup [111]. The optimum determined regimen was 10 mg ruxolitinib twice daily (after one cycle at 25 mg twice a day), and 20 mg/m² per day of decitabine for 5 days every 4 weeks [111]. Mascarenhas and colleagues [112] published the results of the multi-center phase 2 trial evaluating the aforementioned regimen during the production of this article.

Table 3.

Selected clinical Trials on treatments for MPN-AP/BP.

Treatment Regimen	Disease	Clinical Trial
Ruxolitinib + decitabine	MPN-AP/BP†	NCT02076191 (phases 1 and 2) #
Ruxolitinib + decitabine	MPN-BP†	NCT02257138 (phase 1/2)
Ruxolitinib or fedratinib + decitabine	MPN-AP/BP†	NCT04282187 (phase 2)
Ruxolitinib + enasidenib	IDH2-mutated MPN-BP§	NCT04281498 (phase 2)
Ivosidenib + venetoclax +/− azacitidine	IDH1-mutated MPN-BP‡	NCT03471260 (phase 1b/2)
Ivosidenib + azacitidine	IDH1-mutated MPN-BP‡	NCT03173248 (phase 3; AGILE trial)
Guatecitabine (SGI-110)*	MPN-AP‡	NCT03075826 (phase 2)
Ivosidenib or Enasidenib + induction and consolidation therapy	IDH1- or IDH2-mutated MPN-BP‡	NCT02632708 (phase 1)
Gilteritinib	FLT3-mutated MPN-BP (rare) ‡	NCT03836209 (phase 2)
KRT-232 + decitabine or low dose cytarabine	MPN-BP‡, **	NCT04113616 (phase 1b/2)

^†The trials were conducted exclusively for patients in MPN-AP and/or MPN-BP.

^‡Enrollment of patients with MPN-AP and or MPN-BP was allowed in the trial.

^#The phase 1 and 2 studies were conducted by the MPN-Research Consortium.

^§The MPN-Research Consortium is conducting the study. Patients with IDH2-mutated MPN-AP/BP or MF-CP (4–9% circulating blasts) are enrolled.

^*Decitabine –P– guanosine

^**Patients harboring TP53 mutation are excluded from this study.

Recently, Zhou et al. reported a retrospective study of 42 patients, comprising 14 with MPN-AP, 16 with MPN-BP and 12 with HR MF, who were treated with decitabine monotherapy or in combination with ruxolitinib [90]. Among the patients in AP, seven received decitabine monotherapy and six were treated with ruxolitinib/decitabine combination. Over a median follow-up of 12.4 months (range, 2.1–48.8), the median OS of HR MF and MPN-AP patients was not reached; however, only two and one patients were alive at 60 months, respectively. After initiation of decitabine therapy, the probability of complete or partial (50%) reduction in peripheral blasts was 54.6% in the entire cohort [90]. The investigators suggested that early initiation of decitabine or ruxolitinib/decitabine treatment in the accelerated phase conferred benefit to the patients; and that the combination of ruxolitinib with decitabine was superior to decitabine monotherapy (the integrated cohort of HR MF, MPN-AP and MPN-BP patients had a median OS of 21.0 months with the combination versus 12.9 months with decitabine only) [90].

Another study of 180 patients with MPN-AP/BP showed no difference in the outcomes with intensive chemotherapy compared to non-intensive therapy (hypomethylating agents, low-dose cytarabine or treatment in a clinical study); the median OS of the entire group was 5.8 months. Intensive chemotherapy treatment was advantageous only in patients with less than 4 mutations and wild-type TP53 (median OS of 8.1 months for the latter group) [88].

Treatments with Targeted Inhibitors

In preclinical studies on MPN-BP cell lines, significant synergistic activity was noted between ruxolitinib in combination with bromodomain and extraterminal (BET) protein inhibitors, targeting epigenetic proteins [113]. The former preclinical studies provide grounds to conduct clinical studies assessing BET inhibitors/ruxolitinib in patients with MPN-AP/BP. In the ongoing global phase 2 MANIFEST trial (NCT02158858), patients with advanced MF-CP who had either suboptimal response to ruxolitinib treatment or were JAK-inhibitor naïve [114] had notable clinical responses (spleen volume reduction and improvement in anemia, BM fibrosis, total symptom score, and transfusion-dependence) after therapy with CPI-0610 (a selective BET inhibitor) alone or in combination with ruxolitinib [115].

Co-treatment with ruxolitinib and BET proteolysis-targeting chimeras (BET-PROTACs, such as ARV-825) has also exhibited strong synergism against MPN-BP cells, and ARV-825 induced apoptosis in ruxolitinib-resistant MPN-BP cells in vitro [116, 117]. Accordingly, in another recent preclinical study, multi-targeted inhibition of the β-catenin-TCF7L2-JMJD6-c-Myc axis, through co-treatment with the BET-PROTAC ARV-771 and the inhibitor BC2059, overcame resistance of patient-derived MPN-BP blasts to BET inhibitors and improved survival of mice engrafted with BET inhibitor-resistant MPN-BP cells [118]. Inhibitors of the heat shock protein 90 (HSP90) degrade JAK2, and concomitant BET and HSP90 inhibition had synergistic lethal activity against MPN-BP cells that were resistant to ruxolitinib [113, 119].

Notwithstanding limited advancements, the overall clinical experience of treating patients with MPN-AP/BP clearly shows that a major unmet need still remains. As new targeted therapies are transforming the field of AML [18], molecularly targeted monotherapies or combinations with other agents engender promise in MPN-AP/BP [120]. Subgroups of patients harboring targetable mutations that are more frequent in MPN-AP/BP compared to the chronic phase [121] (e.g., IDH1/2 occur at ~ 20% frequency in MPN-BP versus ~ 4% in PMF [85]) may benefit from treatments with the corresponding inhibitors. For example, IDH1- and IDH2-mutated MPN-AP/BP can be treated with the oral IDH1 and IDH2 inhibitors ivosidenib and enasidenib, respectively, and their combinations with other drugs. Preclinical studies demonstrated synergism between enasidenib and ruxolitinib in double-mutant IDH2/JAK2^V617F MPN and MPN-BP patient-derived cells [122]. Notably, Patel et al. and Chifotides et al. recently conducted two retrospective studies in small cohorts of patients with IDH2-mutated MPN-AP/BP [123] and IDH1/2-mutated MPN-BP [124], respectively. In the aforementioned studies, the patients were treated with enasidenib monotherapy (and azacitidine in one case) [123] and IDH1/2-inhibitor monotherapies or IDH1/2-inhibitor-based combinations with other drugs (hypomethylating agents, ruxolitinib, venetoclax, or intensive chemotherapy) [124], respectively; the patients exhibited durable clinical responses with acceptable tolerability and survivals that compared favorably with historical reports. A phase 2 clinical trial evaluating the combination of enasidenib and ruxolitinib in patients with IDH2-mutant MPN-AP/BP and MF-CP (4–9% circulating blasts) is currently underway (NCT04281498; Table 3). Moreover, in another ongoing phase 1 clinical trial (NCT02074839), 6/11 patients with IDH1-mutated MPN-BP and co-occurring JAK2^V617F at baseline, were treated with ivosidenib monotherapy in the relapsed or refractory setting, and achieved CR [125].

Venetoclax monotherapy has exhibited limited efficacy in myeloid malignancies but combination strategies with other therapeutics are very promising in AML and are becoming the standard of care in patients aged > 75 years [126, 127, 128]. However, recent publications in a small number of patients indicate that venetoclax may not be an effective treatment in MPN-AP/BP patients (because Bcl-xl is elevated in MPN cells [129] and sensitivity to venetoclax correlates negatively with Bcl-xl levels) except for patients harboring IDH1/2 mutations due to the production of R-2-hydroxyglutarate. Tremblay et al. [130] and Gangat et al. [131] conducted retrospective studies on small cohorts of MPN-BP patients treated with the combination of venetoclax and hypomethylating agents (decitabine or azacitidine): the median overall survivals were not extended compared to the historically poor survivals reported for MPN-BP. Notwithstanding the short overall survivals, in both studies, a few patients achieved CR and CR_i, and underwent allo-HSCT. Notably, in a few recent preclinical and clinical studies, the investigators reported an increased sensitivity of IDH1/2-mutated human AML cells and prolonged survival in IDH1/2-mutated AML patients with venetoclax [132, 133, 134, 135]. In the report by Gangat et al., the responses of 2 patients with IDH2-mutated BP-MPN revealed an analogous sensitivity to venetoclax-based combination treatment [131]. In another recent report, a patient with IDH2-mutated MPN-BP responded to venetoclax monotherapy [136]. A phase 1/2 trial assessing ivosidenib/venetoclax with or without azacitidine in IDH1-mutated patients with advanced malignancies, including MPN-AP/BP, is underway (NCT03471260; Table 3) [137]. Analysis of interim data from this study evaluating IDH1-mutated patients who were treated with ivosedinib/venetoclax showed efficacy of the combination and promising response rates [138]. Considering that MPN-AP/BP patients have been largely excluded from clinical trials heretofore, in this era, this population will have the option to enroll in rationally designed clinical studies evaluating novel promising regimens.

A phase 1/2 clinical study of KRT-232, an HDM2 inhibitor, in combination with either decitabine or low dose cytarabine is underway for patients with AML, including MPN-BP (NCT04113616; Table 3). In this study, patients with TP53 mutations are excluded due to lack of activity of HDM2 inhibitors on mutated TP53. Approximately 20% of the patients with MPN-AP/BP harbor TP53 [79]. Mutant TP53 AML [139] and TP53 MPN-AP/BP [88, 140] patients have a poor prognosis and do not respond well to chemotherapy treatment. Novel agents are being developed for patients with mutated TP53. Notably, preliminary results with anti-programmed death ligand 1 (anti-PD-L1) agents have not shown clinically relevant results in patients with MPN-AP/BP [141].

Conclusions

In summary, MPN-AP is an aggressive disease that is considered a continuous transitional phase between chronic phase MPN and MPN-BP, and portends poor prognosis. Clinical practice and the limited number of studies that have been conducted heretofore demonstrated that advanced clinical features, such as rising blast counts (≥ 3–5%), acquisition of HMR mutations (ASXL1, IDH1/2, SRSF2, EZH2, U2AF1^Q157, and TP53) and complex karyotype or high-risk cytogenetics (e.g., 17p deletion and gain of 1q), type 1 CALR-unmutated status, severe anemia, and high-grade bone marrow fibrosis likely herald impending leukemic transformation. Regular clinical follow-up and evaluation of factors indicating disease progression in high-risk patients is recommended, aiming to implement early therapeutic interventions and direct the patients to allo-HSCT. In this respect, the prognostic models that were developed to assess the risk score of MPN patients can also predict the risk of transformation, and have evolved significantly in the last 10+ years. Hindrance of MPN progression to the AP/BP and timely referral of eligible patients for allo-HSCT are desirable goals whenever possible. Treatment with hypomethylating agents in combination with ruxolitinib provides an option for MPN-AP patients, instead of intensive chemotherapy, followed by an allo-HSCT. As more targeted therapies emerge for de novo AML, significant advancements are made in the molecular landscape of MPN-AP/BP; and patients with MPN-AP and actionable mutations will have the opportunity to be treated with new modalities and personalized schemes, e.g., IDH1/2 inhibitors alone or in combination with other drugs. In this respect, wide implementation of next generation sequencing is important. Additional novel inhibitors that are in clinical development, e.g., BET inhibitors, provide promising venues that merit further assessment.