Clinical Benefit Derived from Decitabine Therapy for Advanced Phases of Myeloproliferative Neoplasms

BCR-ABL1-negative myeloproliferative neoplasms (MPN) are a heterogeneous group of clonal hematopoietic stem-cell (HSC) malignancies whose pathogenesis is linked to the hyperactivity of the JAK-STAT pathway [1]. Amongst the many sequelae affecting morbidity and mortality is the transformation to acute myeloid leukemia (AML), as defined by the presence of ≥20% myeloblasts in the peripheral blood or bone marrow. Chronic phase (CP) primary myelofibrosis (MF), polycythemia vera (PV), and essential thrombocythemia (ET) are associated with a risk of 10–20, 4, and 1%, respectively, of a transformation to blast-phase (BP) disease, predominantly of a myeloid phenotype [2, 3].

Retrospective studies have identified the prognostic factors associated with the development of MPN-BP. Noted patient-specific risk factors are advanced age and exposure to cytoreductive agents such as radioactive P-32 and alkylating agents [4, 5]. Disease-specific risk factors include anemia, red blood cell (RBC) transfusion dependence, a platelet count <100 × 10⁹/L, and ≥3% peripheral blood blasts [6-9]. Furthermore, certain disease genotypes in MF, including the absence of a driver mutation (JAK2, CALR, and MPL) and the presence of certain nondriver mutations including ASXL1, TET2, SRSF2, RUNX1, and TP53, have been shown to confer an increased risk of leukemic transformation [10-14]. Cytogenetic abnormalities involving chromosomes 5, 7, and 17p have also been associated with a 6-fold increased risk of evolution to MPN-BP, in addition to +1q [3, 10, 15].

MPN-BP is generally associated with poor outcomes, with a median overall survival (OS) of approximately 3–5 months [16-18]. Currently, there is no standard approach for treating this patient population. Intensive chemotherapy alone provides minimal benefit, with most studies showing an OS similar to supportive care [16, 19, 17, 20]. HSC transplantation (HSCT) is the only treatment modality shown to alter the course of the disease but it has, historically, been limited in practice, as most patients are not candidates due to their advanced age and/or significant competing comorbidities [21, 22]. Therefore, the lack of effective management options for advanced phases of MPN represents an urgent unmet clinical need.

Decitabine (deoxyazanucleoside 5-aza-2′-deoxycyti-dine) (DEC) is an S-phase-specific therapeutic activated by deoxycytidine kinase, that produces a pyrimidine analog that is incorporated into DNA and causes irreversible inhibition of DNA methyltransferase [23]. In 2010, Mascarenhas et al. [24]reported the clinical benefit of DEC in reducing spleen size and RBC transfusion requirements, and an associated median OS that was not reached at 9 months in a small cohort of MPN-BP patients. Badar et al. [25]later confirmed this mortality benefit in a single institutional retrospective study at MD Anderson, with a median OS of 6.9 months in MPN-BP patients, 9.7 months in accelerated-phase MPN (MPN-AP) patients (10–19% blasts), and 27 months in high-risk MF (MF-HR) patients (<10% blasts).

In this study, we aimed to further characterize the patient population with advanced phases of MPN treated with DEC alone or in combination with ruxolitinib, delineate the clinical and survival outcomes, and examine the safety profile of DEC in this setting.

We retrospectively reviewed the electronic medical records of all patients seen in the Myeloproliferative Disorders Program at Mount Sinai Hospital that were treated with DEC in 2012–2018. Patients with MPN-BP seen in consultation at our institution but treated by a local physician were not included in this analysis. Each cohort was assigned a particular disease status at DEC initiation. MPN-BP was defined by the presence of ≥20% blasts in the bone marrow or peripheral blood, while MPN-AP was defined as 10–19% blasts [26]. The MF-HR cohort was defined as intermediate-2 or high-risk disease by the Dynamic International Prognostic Scoring System (DIPSS) and was determined to be at a heightened risk for leukemic transformation by the presence of circulating blasts of 4–9% in the peripheral blood, 5–9% blasts in the bone marrow, or a myelodysplastic syndrome (MDS)/MPN overlap [9]. Patients were included in this study if they were receiving single-agent DEC or a combination therapy of DEC and ruxolitinib (Jakafi, Incyte).

Mutational profiles were determined from PCR detection of JAK2V617F, or when available, a next-generation sequencing (NGS) panel of 44 genes associated with myeloid malignancies (Genoptix, Carlsbad, CA, USA), from cells harvested from either the peripheral blood or a bone marrow aspirate when obtained [27].

Cytogenetic analyses were performed by the Tumor Cyto-Genomics Laboratory at Mount Sinai. As previously reported, an unfavorable karyotype was defined as the presence of +1q, inv(3)/t(3;3), -5/del(5q),-7/del7(7q),+8,11q23 rearrangements, and del(12p) [28-30].

Patients were assessed for clinical response on a monthly basis for up to 24 months on DEC therapy. Spleen length was categorized as minimal (0–5 cm), moderate (6–10 cm), or severe (11+ cm) by palpation; a spleen response was defined as a downgrade in spleen length category. RBC transfusion dependence was defined according to the criteria in Gale et al. [31] (2+ units/month over a 3-month period), and a response in transfusion dependence was an improvement from dependence to independence. A response in ECOG performance status was any decrease along the scale. The absence of blasts or a 50% reduction in peripheral blast numbers by manual review of the peripheral blood smear was considered a blast response. Complication rates in the same follow-up period were also assessed. Complications of interest included infections (bacterial, viral, or fungal), thromboses (arterial or venous) documented by radiological studies, and hemorrhage requiring hospitalization and/or transfusional support. Time frames for all events were documented.

Continuous patient-related, disease-related, and treatment-related variables were denoted as median and interquartile range, and categorical variables as n (%). Cumulative incidence functions (CIF) were used to estimate the cumulative probabilities of spleen reduction, transfusion independence, ECOG score <3, and blast percentage reduction over time in a competing-risk setting with death from any cause as the competing event. The Aalen estimator method, based on the theory of counting processes, was used to estimate the standard error of the CIF [32]. The method of Kaplan-Meier was used to estimate the OS distribution with patients censored at the last date known to be alive. The reverse Kaplan-Meier method introduced by Schemper and Smith [33]was used to estimate the median follow-up time, treating censored observations as “events” and patients that were deceased as “censored” . Statistical analyses were performed with SAS v9.4 (SAS Institute Inc., Cary, NC, USA) software package.

Overall, a total of 42 patients treated with DEC were identified, including 16 MPN-BP, 14 MPN-AP, and 12 MF-HR patients (Table 1). The median age of the MPN-BP and AP cohorts at the time of disease evolution was 66.5 and 67.3 years, respectively; MF-HR patients were older with a median age of 72.3 years. There was no gender preference and the Charlson Comorbidity Index (CCI) scores varied widely. 42.9% (18/42) of the merged cohort had an initial MPN diagnosis of MF, 19.0% (8/42) post-PV MF, 28.6% (12/42) post-ET MF, and 9.5% (4/42) MDS/MPN overlap. None of the MPN-BP patients had received >3 prior therapies, and only 2 (15.4%) MPN-AP patients had received >3 prior therapies. Of those who had received prior MPN therapy, hydroxyurea (47.6%) was the most common treatment prior to the current disease state requiring DEC therapy.

Overall, 83.3% (35/42) of patients included in this study had JAK2 mutational testing available and 38.1% (16/42) of patients had NGS results available at the time of the MF-HR, MPN-AP, or MPN-BP diagnosis. Of those who had JAK2 mutational testing, 60.0% harbored JAK2V617F, with a median variant allele fraction (VAF) of 56.7% (range, 2.4 to 94.4). Additional available genomic data included 18.8% (3/16) MPL mutations, 25.0% (4/16) CALR mutations, and 12.5% (2/16) triple-negative (JAK2, MPL, and CALR). 68.8% (11/16) of available patients had a high molecular risk mutation (HMR), including 18.8% (3/16) with a TP53 mutation. Mutated JAK2 and ASXL1 were the most commonly co-occurring mutations (Fig. 1). Additionally, 64% (27/42) of patients tested had abnormal karyotype and 38% (16/42) were unfavorable. Unfavorable karyotype was most common in MPN-BP patients with 56.3% (9/16) compared to 25.0% (5/14) in MPN-AP patients and 16.7% (2/12) in MF-HR patients. The most common karyotypic abnormality was -7/7q, which was present in 15.5% (7/42) of patients (1 MPN-BP, 5 MPN-AP, and 1 MF-HR), followed by the 1q abnormality in 7.1% (3/42) of patients (all MPN-BP). Co-occurrence of +1q and del(5q) occurred in 1 MPN-BP patient and with +8 in another patient. Co-occurrence of t(3;3) and monosomy 7 was identified in 1 MPN-BP patient and del(12p) and monosomy 7 in another MPN-BP patient.

Over half of the patients were RBC transfusion-dependent (54.8%, 23/42), most frequently in the MPN-BP cohort (68.8%, 11/16). ECOG score was also highest in MPN-BP patients, with 17.6% (3/17) of patients noted to have an ECOG score of 3.

All patients received DEC 20 mg/m² intravenously for 5 consecutive days every 4 weeks. The median number of cycles of DEC was 1.5 in MPN-BP, 4.5 in MPN-AP, and 6.5 in MF-HR patients (Table 2). Overall, 35.7% (15/42, i.e., 6 BP, 6 AP, and 3 HR) of the patients received concurrent ruxolitinib therapy. Four patients with MPN-BP were enrolled in the Myeloproliferative Disorder Research Consortium 109 (MPD-RC 109) prospective trial of combination DEC and ruxolitinib (ClinicalTrials.gov ID: NCT02076191).

The median follow-up duration was 23.6 (2.1–NE) months for MPN-BP patients, 12.4 (2.1–48.8) months for MPN-AP patients, and 38.7 (1.8–61.5) months for MF-HR patients. The median OS of the MPN-BP patients was 2.6 months, with the longest-living patient alive at 24 months. The discrepancy between median follow-up and OS in the MPN-BP group is due to the method used to estimate median follow-up (reverse Kaplan-Meier estimator), which does not consider deaths as events when calculating follow-up. Thus, early deaths in the MPN-BP group did not influence the median follow-up. Patients who received at least 2 cycles of DEC in the MPN-BP group had an OS of 6.7 months with a median follow-up duration of 23.6 (3.8–29.8) months. Median OS was not reached for MPN-AP and MF-HR patients, with 1 and 2 patients alive at 60 months, respectively (Fig. 2, 3). Across the integrated cohort of patients, DEC monotherapy (27 patients, i.e., 11 BP, 7 AP, and 9 HR) was associated with a median OS of 12.9 months while DEC combination therapy with ruxolitinib (15 patients, i.e., 6 BP, 6 AP, and 3 HR) was associated with a median OS of 21.0 months (online suppl. Fig. 1; for all online suppl. material, see www.karger.com/doi/10.1159/000506146).

Overall, the probability of spleen length reduction from moderate or massive splenomegaly to minimal splenomegaly within 12 months of initiating DEC therapy was 28.6% (95% CI 9.5–51.3%). The 12-month probability of transfusion independence among patients who were transfusion-dependent at the start of DEC was 23.5% (95% CI 9.0–41.8%) and the probability of an improvement in ECOG score was 28% (95% CI 11.7–47%). Lastly, the probability of attaining a complete or 50% reduction in peripheral blast count within 12 months of DEC treatment was 54.6% (95% CI 35.7–70%) (online suppl. Table 1). Patients who achieved at reduction in blast count of at least 50% within 24 months of initiating DEC were more likely to be spleen responders (p = 0.05). No such association existed between blast reduction and transfusion independence (p = 0.22) or OS (p = 0.62). Follow-up cytogenetics were not sufficiently available to assess for karyotypic response.

DEC was well-tolerated with minimal clinically significant adverse events noted (online suppl. Tables 2 and 3). Bacterial infections were the most common complication, with the highest risk for these being in the first month, and the prevalence was not significantly different in patients who received concomitant ruxolitinib and those who did not. Viral infections (n = 2), however, occurred only in patients receiving concurrent ruxolitinib. No cases of arterial thrombosis were observed, and all venous thrombotic events (n = 3) were deep venous thromboses that occurred within the first 6 months of DEC therapy. Similarly, hemorrhagic events (n = 2), i.e., uncontrolled epistaxis requiring hospitalization or cauterization, occurred within the first 3 months of DEC therapy and were associated with extreme thrombocytopenia. Additionally, only 2 patients in the cohort had to discontinue treatment due to treatment-related adverse events. The most common reasons for discontinuation were death (n = 10), progression of disease (n = 7), transplantation (n = 3), and no improvement while receiving DEC therapy (n = 3).

MPN-BP has historically been associated with a dismal outcome, with HSCT as the only therapeutic approach offering a potential for cure and long-term survival [21, 22, 34]. Supportive care only is associated with a median OS of 3–5 months, and AML-like induction chemotherapy regimens do not meaningfully improve progression-free survival or OS in the absence of HSCT consolidation [16, 19, 17, 20]. DEC is FDA-approved for the treatment of MDS and is frequently used off-label for the treatment of AML. DEC has also been shown to have clinical activity in MPN-BP, with a favorable OS of 6–9 months, and has emerged as a promising outpatient treatment option for this secondary AML population [24, 25]. The objective of this single-center, retrospective study was to clarify the clinical benefit and tolerability of DEC across a group of patients with MPN in advanced phases including MF-HR, MPN-AP, and MPN-BP cohorts.

Surprisingly, the MPN-BP cohort had an associated median OS of only 2.6 months with DEC therapy. This was lower than reported by prospective studies and is comparable to historical cohorts of supportive care only. There are several potential explanations for this discrepancy. Importantly, patients treated by their local physician were not included in this study. Because these patients are likely to have a good performance status and be eligible for outpatient DEC therapy, the MPN-BP cohort included in our study was enriched with higher-risk, sicker patients with a poor performance status, many of whom required rapid initiation of DEC. This is reflected in the poor baseline performance status of our MPN-BP patients (50% with ECOG 2/3), and the fact that 56% required hospitalization at time of DEC initiation. Secondly, half of the MPN-BP cohort (n = 8) had only 1 cycle of DEC before they died. In a large phase-III study on DEC administration to patients with AML, the median time to best response was 4.3 months [35]. Therefore, it is reasonable to conclude that a single cycle of DEC did not have sufficient time to achieve a therapeutic effect and that these deaths were secondary to rapidly progressive disease. In fact, this poor OS of patients with MPN-BP suggests that many patients do not live long enough to garner any benefit from DEC.

The MPN-AP and MF-HR cohorts did not reach a median OS. While these patients are anticipated to have better outcomes than MPN-BP, this difference in OS may also suggest there is an optimal benefit of DEC when it is initiated early in the course of disease evolution. For comparison, in the single-center study of Masarova et al. [36], the OS of patients with MPN-AP and MF-HR (as defined in our study) was 24 (range 52–66) and 28 (range 18–38) months, respectively. Theyshowed that patients with bone marrow blasts >5%, or peripheral blood blasts 4–9%, had a survival comparable to MPN-AP patients, and should, therefore, be considered as candidates for cytoreductive therapy with DEC or HSCT. Furthermore, blast count influence on OS was independent of DIPSS score and driver-mutation status. In agreement with their results, the MPN-AP and MF-HR patients treated with DEC in our study had similar survival (p = 0.85). Clinical benefit with DEC treatment was seen in terms of reduction in spleen size, transfusion burden, peripheral blood blast count, and an improvement in performance status. DEC was well tolerated by all the patients. Of note, the patients in our cohort received DEC for 5 days, not a 10-day regimen as is common in many European centers. This dosing schedule is standard at our institution and was confirmed to be equivalent to 10 days in a recent head-to-head comparison in older patients with newly diagnosed AML [37].

It is important to highlight that 36% (15/42) of the cohort also received concurrent ruxolitinib therapy with a median OS of 21 months compared to 12.9 months in the DEC monotherapy group; however, this difference was not statistically significant (p = 0.77). This observation is notable given the recent final results of the MPD-RC 109 multicenter, phase-1/2 trial of combination DEC and ruxolitinib therapy in MPN-AP/BP patients. The median OS of the combined dose escalation cohorts was 7.9 months in the phase-1 study; at the recommended phase-2 dose of 25 mg twice daily of ruxolitinib in cycle 1 and then 10 mg twice daily in subsequent cycles, a median OS of 9.7 months in the MPN-BP group was demonstrated [38, 39].

Our study is not without its limitations. Given its retrospective nature, details related to certain outcome measures may have been missed. For example, the cause of death was unknown in 10 patients. Additionally, a number of patients were lost to follow-up, thus limiting the survival analyses. Furthermore, the small sample size precluded multivariate analysis of OS. Finally, we had incomplete mutational data for the majority of our patients, thus reducing our ability to fully characterize this cohort.

DEC, alone or in combination with ruxolitinib, appears to have clinical benefit and the potential to extend the survival of patients with MPN when it is initiated early in the evolution of the disease to MPN-BP. For hospitalized patients or those with poor performance status, DEC is unlikely to significantly alter the natural course of the disease. Bacterial infection was the most common treatment emergent adverse event. The combination of DEC and ruxolitinib may confer additional survival benefit over DEC alone and warrants further prospective evaluation. Overall, DEC is a viable treatment option, with existing preclinical rationale and MPN murine modeling to support its use in patients with disease progression, with the goal of extending their survival. Optimally, it should be administered prior to the development of MPN-BP.

The authors wish to acknowledge Ami Patel for reviewing the manuscript.

This study was approved by the Program for the Protection of Human Subjects at the Icahn School of Medicine at Mount Sinai. The authors state that this research complied with all internationally accepted standards for research practice and is in compliance with the Helsinki Declaration.

J.M. received clinical research funding paid to the institution from Incyte, Novartis, Roche, Promedior, Merck, CTI Biopharma, Janssen, PharmaEssentia, Celgene, Merus, and Arog, and is a clinical trial steering committee member for Roche, Incyte, Celgene, and CTI Biopharma. M.K. received research funding from Incyte, Celgene, Constellation, and Blueprint Medicines, and consulting fees from La Jolla Pharmaceutical. R.H. serves on the advisory boards of Novartis and La Jolla Pharmaceuticals. The remaining authors have no conflicts of interest to disclose.

The authors wish to acknowledge the support of the Biostatistics Shared Resource Facility, Icahn School of Medicine at Mount Sinai, and an NCI Cancer Center Support Grant P30 CA196521–01.

S.Z. and D.T. conceived the study, performed data collection, and analysis. J.M., R.H., and M.K. conceived the study. V.N. performed cytogenetic interpretation and analysis. L.L. and E.M. performed statistical analysis. All authors composed and reviewed the manuscript.

Selena Zhou and Douglas Tremblay are co-first authors with equal contribution in this study.