Incidence of blast phase in myelofibrosis according to anemia severity

Abstract Myelofibrosis (MF) is a clonal malignancy frequently characterized by anemia and in 10%–20% of cases it can evolve into blast phase (BP). Anemia in MF is associated with reduced survival and ‐in primary MF‐ also with an increased probability of BP. Conventional treatments for anemia have limited effectiveness in MF. Within a dataset of 1752 MF subjects largely unexposed to ruxolitinib (RUX), BP incidence was 2.5% patients per year (p‐y). This rate reached respectively 4.3% and 4.5% p‐y in case of patients with common terminology criteria for adverse events (CTCAE) grade 3/4 and grade 2 anemia, respectively, that represented together 32% of the cohort. Among 273 MF cases treated with RUX, BP incidence was 2.89% p‐y and it reached 4.86% p‐y in subjects who started RUX with CTCAE grade 2 anemia (one third of total). Within patients with red blood cell transfusion‐dependency at 6 months of RUX (21% of the exposed), BP rate was 4.2% p‐y. Our study highlights a relevant incidence of BP in anemic MF patients, with a similar rate whether treated with or without RUX. These findings will help treating physicians to make decisions on the safety profile of innovative anemia treatments.

Anemia is a characteristic feature of MF, found in around 35%-40% of cases at diagnosis, becoming more frequent with the disease progression, and impacting deeply on patients' quality of life [12][13][14][15]. In addition, by using the currently approved JAK inhibitors (JAKis), such as ruxolitinib (RUX) or fedratinib (FED), hemoglobin (Hb) reduction is an expected early on-target effect, with grade 3/4 anemia occurring in 38-45% of cases [2]. Anemia is listed among the major survival risk factors in MF prognostic models developed before the widespread use of JAKis [5,7,12]. In this respect, red blood cell (RBC) transfusion requirement represented a detrimental factor for survival in 209 RUX-treated MF patients [16]. In large series of PMF patients, risk factors for BP evolution have been investigated and, among those, anemia has been recognized as relevant [9,17,18]. On this point, information in SMF or in RUX-treated cases is more limited [6,11].
In current practice, treatment of MF-associated anemia mostly consists of erythropoiesis-stimulating agents, danazol and RBC transfusions [2,19]. However, the interest of studying anemia-improving molecules with new targets is growing in MF, although the impact of such treatments on BP occurrence is unknown [2,19].
In this study, we reported the incidence of BP according to anemia severity in large real-world cohorts of PMF and SMF patients, treated with or without JAKis, mainly RUX. This could serve as a reference for assessing BP occurrence in populations of MF patients receiving innovative anemia-oriented treatments.

Study populations and definitions
A total of 2381 MF patients entered the analysis, and we generated two different cohorts (Table S1). Cohort 1 (C1) was composed of (1) PLT count (x10ˆ9/L) anemia). Lastly, we used the Hb 9.5 g/dL value as a threshold [23,24].

Patients' characteristics
The main features at MF diagnosis and follow-up events of the 1752 C1 subjects, overall and distinguished by anemia presence and degree, are detailed in Table 1 and in Table S2 and fatalities were reported mostly in patients presenting with a higher degree of anemia (p < 0.0001).  Table 3 reports the rate and incidence of BP transformation in C1, distinguished by the presence and the degree of anemia at MF diagnosis. At a median time of 2.2 (IQR, 0.9-3.9) years from MF diagnosis (Table 1), BP evolution was reported in 185 (10.6%) C1 patients, with an incidence rate of 2.5% patients per year (p-y) (95%CI, 2.2-2.9).

BP incidence per anemia grade in the RUX-unexposed population (C1)
The latter was higher in patients with greater degrees of anemia.  Figure 1 shows that BP-FS was significantly associated (p < 0.0001) with the presence and severity of anemia at MF diagnosis, whichever of the three classifications is used ( Figure 1A-C). F I G U R E 1 Association between blast phase-free survival and anemia degree at time of myelofibrosis diagnosis in 1752 patients of Cohort 1. Kaplan-Meier curves describe the blast phase (BP)-free survival of patients within the different anemia classes, as for common terminology criteria for adverse events (CTCAE, (A)), sex-and severity-adjusted (B) and as for hemoglobin (Hb) 9.5 g/dL-threshold (C) categorization.

TA B L E 3
Blast phase prevalence and incidence of 1752 myelofibrosis patients (Cohort 1), in the overall dataset and based on anemia degree at diagnosis. Abbreviations: 95%CI, 95% confidence interval; CTCAE, common terminology criteria for adverse events; Hb, hemoglobin; n, number; p-y, persons-year; RBC-TD, red blood cells-transfusion dependency. Considering the 41 RBC-TD patients at time of RUX start, Table   S6 reports that BP risk after six, 12, 18, and 24 months of treatment was higher compared to the 232 no-TD subjects. Of note, in the RCB-TD group, 75% of all BP transformations occurred within 1 year from RUX start. Within the no-TD cohort, around 74% of evolutions were detected after 1 year of treatment. TA B L E 5 Comparison of blast phase incidence between ruxolitinib-unexposed Cohort 1 and Cohort 2 patients, in the overall datasets and based on anemia degree according to the CTCAE classification. As detailed in Table 5, we found no difference in the IRR of BP between RUX-unexposed C1 patients and RUX-exposed C2 subjects, both when considering the whole populations and the different CTCAE anemia classes.

DISCUSSION
Anemia is a hallmark feature of PMF and SMF, being included not only within their diagnostic criteria but also among the major prognostic variables [1,5,7,12,22,25].
In our large real-world population of 1752 PMF and SMF subjects, median Hb value at diagnosis was 11.4 g/dL, and 31.9% of subjects presented CTCAE grade 2 or 3/4 anemia. Around 37% had sex-adjusted moderate to severe anemia, while Hb was ≤9.5 g/dL in about 24% of cases. We confirmed a higher incidence of anemia in overt-versus pre-PMF, and in PET-versus PPV-MF patients.
In the DIPSS cohort, 47% of patients developed anemia after a median of 3.3 years from PMF diagnosis [12], and in the MYSEC population Hb values tended to decrease with a longer time to progression from PV [28]. Of 296 subjects included in the randomized phase 3 COMFORT studies, 134 (45.3%) had baseline Hb < 10 g/dL or required at least one RBC unit within 12 weeks prior to RUX initiation [29]. Similarly, in our real-world cohort of 273 PMF and SMF cases, around 40% of subjects presented CTCAE grade 2 or 3/4 anemia at the time of RUX start.
In both our study populations, anemia was correlated with unfavorable features, like a cytopenic phenotype and the higher prognostic risk categories, as previously described [5,7,12,22,25,30,31]. We also confirmed that disease-related RBC-TD or anemia adversely impact on survival in RUX-treated patients [16,29].
Overall, around 10-20% of MF cases evolve into BP [4,6,8], with a consequent dismal survival [32,33]. In an Italian study on 661 PMF cases, the cumulative 10-years incidence of BP was 12% and 23% for pre-and overt-PMF cases, respectively [4]. Within the MYSEC project, BP was reported in 20.5% patients after a median follow-up of 3.5 years [20]. In a retrospective European cohort of 589 RUX-treated PMF and SMF subjects, 14.5% progressed to BP after a median treatment time of 3 years, with a reported incidence of 3.7% p-y [11]. Our results are aligned with previously reported rates of BP in the general MF population and in those receiving RUX [4,11,20], and they confirm that RUX leaves unaffected the probability of BP evolution in MF [11]. Difference in BP incidence between our RUX-exposed cohort and the aforementioned European dataset could probably depend on population size and risk category distribution [11].
Risk factors for BP transformation have been studied mostly in PMF and before the widespread use of JAKis [9,17,18,25,34]. In this respect, Hb values < 10 g/dL are among the variables included in the DIPSS or in the molecularly annotated PMF prognostic models, that have been shown to predict BP transformation [17,22,25]. In over 1300 PMF subjects, the incidence of BP was higher in patients with sex-adjusted moderate to severe anemia compared to lower grades [9].
Considering patients that had to receive at least one RBC unit to maintain Hb ≥8.5 g/dL at any time after PMF diagnosis, BP probability was significantly increased compared to subjects that did not receive RBC units [18]. Within 589 RUX-treated MF patients, a greater risk of BP was associated with the higher DIPSS/MYSEC-PM categories, while baseline RBC-TD did not seem to play a role [11].
In our study, we confirmed an increased probability of BP progression in patients with relevant grades of anemia at PMF or SMF diagnosis. In RUX-unexposed patients, BP incidence reached 4.3% and 4.5% p-y for CTCAE grade 3/4 and grade 2 anemia, respectively. The same rate corresponded to 5.5% and 4.1% p-y for sex-adjusted severe and moderate anemia. Consequently, BP-FS was significantly reduced in these subgroups compared to patients with milder anemia.
In the RUX-exposed cohort, the trend was similar with an incidence of BP estimated to be 4.86% and 5.16% p-y in CTCAE grade 2 anemia and sex-adjusted moderate anemia, respectively. In patients with baseline RBC-TD, the incidence of BP was 4.63% p-y, slightly higher compared to non-TD patients.
Based on these findings, treating anemia in MF represents a challenge, both due to its impacts on quality of life and its associated risk of BP progression, and avoiding the development of RBC-TD remains a real unmet clinical need [2]. and it is under evaluation in a phase 2 study [41]. Reduction of hepcidin overexpression is the target of the ALK inhibitor INCB000928 [42], and of the humanized monoclonal anti-hemojuvelin DISC-0974 [2].
Finally, ACVR1 inhibition has been found to be critical, with JAKis as momelotinib and pacritinib able to ameliorate Hb levels despite JAK inhibition [43][44][45][46][47]. It remains to be determined if these new molecules interfere with BP occurrence.
Our study reiterates the high frequency of anemia in a large cohort of MF patients, including more contemporary patients treated with JAKi, and assesses the impact of varying degrees of anemia on BP progression. Although conventional therapies for anemia have limited efficacy, new molecules under study appear to have the potential to raise Hb levels. Our study assessed the natural occurrence of post-MF BP, that is left unaffected by RUX treatment, and these data will be a useful reference for physicians to make decisions on the efficacy and safety profile of innovative anemia treatments.

AUTHOR CONTRIBUTIONS
BM and FP were involved in all stages of manuscript development, including drafting and critical reviewing of the manuscript. LB performed statistical analysis. All authors gave a substantial contribution to acquisition of data, revised critically the draft, and approved the submitted and final version of the paper.

ACKNOWLEDGMENTS
The Varese and the Florence group have been supported by Ministero

ETHICS STATEMENT
The study was approved by the Institutional Review Board of Insubria (Varese, Italy) and conducted in accordance with the principles of the Declaration of Helsinki. The authors have confirmed patient consent statement is not applicable for this submission.

DATA AVAILABILITY STATEMENT
Data will be available for sharing upon request.