Risk Factors for Progression to Blast Phase and Outcome in 589 Patients with Myelofibrosis Treated with Ruxolitinib: Real-World Evidence

Introduction. Blast phase (BP) is the terminal and most incurable phase of myelofibrosis (MF) and occurs in a not negligible fraction of patients (pts). In the pre-ruxolitinib (RUX) era, peripheral blasts, thrombocytopenia, unfavorable cytogenetics, and high risk category were identified as predictors of BP. RUX is the standard of care for symptomatic MF; however, information on clinical/laboratory correlates of BP in RUX-treated pts is not available.

Aims. The primary objective of the study is to assess real-world data on incidence, risk factors and outcome of BP in RUX-treated MF pts.

Methods. A multicentre observational retrospective study on RUX-treated MF pts was conducted in 20 European Hematology Centers. Data were extracted from an electronic database that included consecutive pts treated with RUX from June 2011. Data cut-off was June 2019. Risk category was assessed at RUX start according to the Dynamic International Prognostic Score System (DIPSS) or the Myelofibrosis Secondary to PV and ET Collaboration Prognostic Model (MYSEC-PM) in pts with post-Polycythemia Vera (PV)/post-Essential Thrombocythemia (ET) MF (secondary MF, SMF). A time-to-event (BP) analysis was conducted with Fine & Gray model with death/time of stem cell transplant as competing risks. Variables tested for association with BP were: age≥65yr, sex, transfusion-dependency, PLT<150x10⁹/l, peripheral blasts ≥3%, marrow fibrosis grade, CALR-unmutated genotype, unfavorable karyotype, spleen length (≥10 cm), total symptoms score (≥20), previous hydroxyurea (HU), alkylating agents, and interferon (IFN) use, time from MF diagnosis to RUX start, and PV/ET duration. Cumulative Incidence Function among risk categories for DIPSS and MYSEC-PM was calculated applying the Gray's model.

Results. Overall, 589 MF pts were included and observed for 1833 pt-yrs from RUX start (median, 35.4 mos). Diagnosis was PMF in 304 pts (51.6%), PPV-MF in 164 pts (27.8%) or PET-MF in 121 (20.6%); 58.4% males. Molecular status was: JAK2^V617F (82.5%), CALR (11.3%) and MPL^W515K/L (1.1%); 5.1% were triple negatives. Overall, 368 (62.5%) pts received ≥1 cytoreductive therapy before RUX, specifically: HU, n. 357; alkylating agents, n. 47; anagrelide, n. 33; and IFN, n. 29. Median time from MF diagnosis to RUX start was 1.3 yrs. DIPSS for the whole cohort was: INT-1 (52.9%), INT-2 (40.1%), and HIGH (7%). DIPSS distribution in PMF pts was: INT-1 (47.8%), INT-2 (45.7%), and HIGH (6.5%), while SMF pts were categorized at LOW (11.1%), INT-1 (43.1%), INT-2 (31.2%) and HIGH (14.6%) risk according to the MYSEC-PM.

Overall, 65 (11%) developed BP. In 61 pts, BP caused RUX withdrawal after a median time of 1.2 yrs (0.7-6.2); in 4 pts BP occurred after RUX stop (median time: 2.4 yrs). BP incidence rate was 3.6 x100 pt-yrs and was comparable in PMF and SMF (p=0.1). In univariate analysis, the probability of BP evolution for the PMF cohort was significantly reduced by previous IFN use (p=0.001). In SMF, predictors for BP in univariate analysis were PLT <150 x10⁹/l (p=0.001), blasts ≥3% (p=0.002), grade 3 marrow fibrosis (p=0.03) and PV/ET duration ≥ 10 yrs (p=0.02); previous IFN significantly reduced the risk of BP (p=0.02). In multivariable analysis, PLT <150 x10⁹/l (HR 2.4, 95% CI 1.1-5.4, p=0.03), blasts ≥3% (HR 3.3, 95% CI 1.4-7.5, p=0.004) and previous IFN (HR 0.1, 95% CI 0.02-0.8, p=0.04) remained significant.

High DIPSS risk significantly predicted BP in PMF (p=0.04, HR [95% CI]: 2.6 [1.1-6.5]) but not in SMF (p=0.40). In this latter cohort, only the MYSEC-PM was associated with BP (p=0.02, HR 1.7 [95% CI]: [1.1-2.8]) (Fig.1). Estimated HRs, in reference to the lower score category, were: 1.10 for INT-1, 1.82 for INT-2, and 4.04 for HIGH risk. HR for HIGH risk, comparing to all lower risk groups, was 3.53 (95% CI: 1.53-8.11).

Overall, 54 (81.8%) BP pts died and median survival was 2.8 mos. Survival after BP was not influenced by type of MF, previous response to RUX, and type of salvage treatment.

Conclusions. Thrombocytopenia and peripheral blasts at RUX start identified pts at higher risk of BP in SMF, while previous IFN use was associated with reduced BP evolution in both PMF and SMF, suggesting a possible disease-modifying action of this agent. Also, this analysis supports the ability of MYSEC-PM in predicting BP in pts with SMF. Despite RUX use, outcome after BP remained dismal, confirming the need for newer treatment strategies.

View largeDownload slide

View largeDownload slide

Close modal