INTRODUCTION — Primary myelofibrosis (PMF) is a myeloproliferative neoplasm (MPN) associated with bone marrow fibrosis, cytopenias, constitutional symptoms, hepatosplenomegaly, and/or extramedullary hematopoiesis. PMF has the least favorable prognosis among the MPNs, and patients are at risk for premature death due to disease progression, leukemic transformation, thrombo-hemorrhagic complications, and infections. PMF has previously been described by various names, including agnogenic myeloid metaplasia, myelofibrosis with myeloid metaplasia, and chronic idiopathic myelofibrosis.
This topic will discuss assessment of prognosis, which is important for evaluating potential clinical outcomes and for selecting treatment of PMF.
Clinical manifestations and diagnosis, risk-stratified management, and pathogenesis of PMF and an overview of MPNs are discussed separately.
●(See "Clinical manifestations and diagnosis of primary myelofibrosis".)
●(See "Myelofibrosis (MF): Management of primary MF and secondary MF".)
●(See "Pathogenetic mechanisms in primary myelofibrosis".)
●(See "Overview of the myeloproliferative neoplasms".)
CLINICAL OUTCOMES IN PMF — PMF is the most aggressive of the BCR-ABL1-negative myeloproliferative neoplasms, which also includes polycythemia vera and essential thrombocythemia. PMF is associated with an estimated median overall survival (OS) of approximately six years [1]. As an example, median OS was 69 months (95% CI 61-76 months) in a multicenter retrospective study of 1054 patients diagnosed with PMF between 1980 and 2017 [2]. This study included patients who were managed with a wide variety of approaches, including observation, oral chemotherapy (primarily hydroxyurea), androgens, erythropoiesis-stimulating agents, steroids, interferon, anagrelide, immunomodulatory agents, splenectomy, and transplantation. Other prospective, retrospective, and population-based studies have reported comparable survival data [3-11].
Leukemic transformation is the most commonly identified cause of death in patients with PMF. In the multicenter study described above, transformation to acute leukemia accounted for 17 percent of deaths (86 of 517 deaths), followed by progression of PMF (10 percent), thrombosis and cardiovascular complications (7 percent), infection (6 percent), bleeding (3 percent), portal hypertension (2 percent), and second cancers (2 percent) [2].
PROGNOSTIC FACTORS — Clinical outcomes in PMF are associated with certain clinical factors and cytogenetic/molecular features.
Clinical factors — Outcomes in PMF are associated with age, presence of symptoms, blood counts, and other clinical and laboratory features, which have been incorporated into some prognostic models. The specific features and thresholds that are scored as adverse features for PMF vary from model to model. (See 'Assessing prognosis in PMF' below.)
Inferior survival in PMF is associated with the following clinical factors, based on multivariate analysis in large studies [2,12-16]:
●Older age
●Anemia
●Leukocytosis
●Thrombocytopenia
●Circulating blasts
●Bone marrow fibrosis
●Constitutional symptoms
●Need for transfusion
Inferior leukemia-free survival (LFS) was associated with platelet count <100,000/microL and circulating blasts ≥2 percent by multivariate analysis in a study of 641 patients with PMF [17].
Expression of various cytokines is increased in PMF. Inferior overall survival (OS) and progression-free survival (PFS) were associated with elevated levels of interleukin (IL)-8 and IL-2R in a study of 127 treatment-naïve patients with PMF [18]. C-reactive protein >7 mg/L was associated with inferior LFS in a separate study of 184 patients with PMF or myelofibrosis arising after other myeloproliferative neoplasms [19].
Genetics
Cytogenetic features — Cytogenetic abnormalities are common in PMF, but highly variable. The largest study of PMF reported an abnormal karyotype in 45 percent of 1002 patients at the time of diagnosis [14]. Certain cytogenetic features are independently associated with OS in PMF, and they have been grouped into three risk categories for use in prognostic models [17,20-23] (see 'Assessing prognosis in PMF' below):
●Very high risk (VHR): single or multiple abnormalities of -7, i(17q), inv(3)/3q21, 12p-/12p11.2, 11q-/11q23, or other autosomal trisomies not including +8/+9 (eg, +21, +19)
●Favorable: normal karyotype or sole abnormalities of +9, 13q-, 20q-, chromosome 1 translocation/duplication, or sex chromosome abnormality including -Y
●Unfavorable: all other abnormalities
Multivariate analysis of 641 patients with PMF (median age 63 years, 64 percent male) reported that, relative to the favorable category as a reference value, these cytogenetic categories were associated with the following hazard ratios (HR) for OS and LFS [17]:
●VHR: OS HR 2.5 (95% CI 1.7-3.5); LFS HR 2.4 (95% CI 1.02-5.5)
●Unfavorable: OS HR 1.9 (95% CI 1.4-2.5); LFS HR 2.7 (95% CI 1.5-4.9)
Driver mutations — Constitutive activation of the JAK/STAT pathway appears to be important to the pathogenesis of PMF, and mutations of three genes involved with this pathway, JAK2 (Janus kinase 2), CALR (calreticulin), and MPL (myeloproliferative leukemia virus oncogene, which encodes the thrombopoietin receptor) have been identified as drivers of this malignancy. (See "Pathogenetic mechanisms in primary myelofibrosis", section on 'JAK/STAT pathway'.)
Approximately 90 percent of patients with PMF have one of the following, mutually exclusive driver mutations [17,20,24]:
●JAK2: 57 to 65 percent
●CALR: 20 to 25 percent
●MPL: 5 to 10 percent
●"Triple-negative" (ie, none of the driver mutations): 10 to 12 percent
Clinical outcomes are generally best for CALR-mutated disease and worst for triple-negative disease. In a retrospective study of 428 patients with PMF, median OS for patients with a mutation of JAK2, CALR, MPL, or triple-negative disease were 5.9, 15.9, 9.9, and 2.3 years, respectively [1]. In another study of 641 patients, multivariate analysis reported the following HRs for OS; none of these factors was independently associated with LFS [17]:
●JAK2: 2.2 (95% CI 1.6-3.0)
●Absence of type 1-like CALR: 2.0 (95% CI 1.5-2.8)
●MPL: 1.6 (95% CI 1.1-2.6)
●Triple negative: 2.3 (95% CI 1.5-3.5)
High molecular risk (HMR) mutations — Mutations of certain genes are considered to be high molecular risk (HMR) mutations for PMF, and inferior outcomes are related to the number of HMR mutations.
More than 80 percent of patients with PMF harbor a variant of at least one of the following genes [20,22,24]:
●ASXL1 (38 percent)
●SRSF2 (14 percent)
●U2AF1Q157 (8 percent)
●EZH2 (7 percent)
●IDH1/2 (4 percent)
Several studies have reported that a greater number of HMR mutations is associated with increasingly poor prognosis [20,22,24]. As an example, in a study of 797 patients with PMF, median survivals for those with none, one, or two or more mutations in these genes, were 12.3, 7.0, and 2.6 years, respectively [24].
Multivariate analysis of 641 patients reported that mutations of ASXL1, SRSF2, and U2AF1Q157 were associated with inferior survival in PMF; HRs ranged from 1.6 to 2.2 [17]. ASXL1 and SRSF2 were also independently associated with inferior LFS (HR: 2.1 and 4.3, respectively). Loss of heterozygosity for TP53 was associated with leukemic transformation in a study that included 34 patients with PMF [25].
ASSESSING PROGNOSIS IN PMF — There is no consensus regarding the optimal prognostic tool for PMF, and the preferred model may vary by clinician and/or institution. In general, prognostic models that incorporate cytogenetic and molecular features are more effective for distinguishing risk groups than those that utilize clinical features alone. In settings where molecular techniques are not available, it is acceptable to use models that are based solely on clinical features and/or karyotype. (See 'Alternative approaches' below.)
Our preferred approach — We favor a two-step approach for assessing prognosis in PMF (algorithm 1):
●We initially assess patients using GIPSS (genetically inspired prognostic scoring system) (calculator 1), an easy-to-use tool that requires only cytogenetic and molecular data [17]. GIPSS stratifies patients into four distinct prognostic groups (ie, low, intermediate [int]-1, int-2, high risk). (See 'GIPSS' below.)
●For patients assigned to the GIPSS int-1 or int-2 categories (approximately two-thirds of patients), we suggest further evaluation with MIPSS70+ v2.0 (mutation-enhanced international prognostic scoring system plus karyotype, 70+, version 2.0) [26]. MIPSS70+ v2.0 is a more complex tool that incorporates clinical, cytogenetic, and molecular features, but it provides greater discrimination for patients with intermediate prognostic risk. MIPSS70+ v2.0 scoring is described below. (See 'MIPSS70+ v2.0' below.)
We prefer this two-step approach because it leverages the simplicity of GIPSS for identifying patients who are at highest and lowest risk for death and transformation to leukemia. GIPSS and MIPSS70+ v2.0 are equally effective for identifying patients in the highest and lowest risk categories [17].
In contrast, MIPSS70+ v2.0 is more effective for stratifying patients with intermediate prognosis, and it reclassifies substantial proportions of GIPSS int-1 and int-2 patients. As examples, GIPSS int-1 patients may be reassigned to MIPSS70+ v2.0 low, intermediate, or high risk groups, while GIPSS int-2 may be reassigned as MIPSS70+ v2.0 very high, high, or intermediate risk [17]. Reclassification of GIPSS int-1 or int-2 may have a significant effect on therapeutic decisions, as discussed separately. (See "Myelofibrosis (MF): Management of primary MF and secondary MF", section on 'Risk stratification'.)
Some clinicians may favor initial use of MIPSS70+ v2.0 for all patients (ie, rather than sequential application of GIPSS and MIPSS70+ v2.0). (See 'MIPSS70+ v2.0' below.)
GIPSS — GIPSS (genetically inspired prognostic scoring system) (calculator 1) is our preferred tool for initial assessment of prognosis in PMF. (See 'Our preferred approach' above.)
GIPSS is an easy-to-use prognostic tool that is based entirely on the karyotype and a limited number of mutations (ie, GIPSS does not include any clinical features) [17]. GIPSS has been validated for predicting overall survival (OS) and leukemia-free survival (LFS) in large patient populations, for which it was shown to be noninferior to other PMF prognostic models.
Importantly, for patients who are scored in one of the GIPSS intermediate (int) categories, we suggest subsequent application of the MIPSS70+ v2.0 prognostic model (algorithm 1). (See 'MIPSS70+ v2.0' below.)
GIPSS utilizes the following features:
●Karyotype (definitions of these categories are provided above) (see 'Cytogenetic features' above):
•Very high risk (VHR) = 2 points
•Unfavorable = 1 point
●Driver mutations:
•Absence of type 1-like CALR = 1 point
●High molecular risk (HMR) mutations:
•ASXL1 mutation =1 point
•SRSF2 mutation = 1 point
•U2AF1Q157 mutation = 1 point
Points are summed to assign a patient to one of four prognostic groups with distinct five-year OS and median OS (respectively, in parentheses), according to an international study of 641 patients [17]:
●Low risk: zero points (94 percent, 26.4 years)
●Int-1: 1 point (73 percent, 10.3 years)
●Int-2: 2 points (40 percent, 4.6 years)
●High risk: ≥3 points (14 percent, 2.6 years)
GIPSS scoring was comparable to MIPSS70+ v2.0 for prediction of 10-year mortality, based on logistic regression by area under the curve (AUC) analysis and Akaike information criterion (AIC) [17]; both GIPSS and MIPSS70+ v2.0 were superior to DIPSS (dynamic international prognostic scoring system) by these criteria.
MIPSS70+ v2.0 — We suggest using MIPSS70+ v2.0 (mutation-enhanced international prognostic scoring system plus karyotype, version 2.0) (calculator 2) to refine the prognosis of patients who are assigned the GIPSS int-1 or int-2 categories (algorithm 1), as described above. Some clinicians may favor initial use of MIPSS70+ v2.0 for all patients with PMF. (See 'Our preferred approach' above.)
MIPSS70+ v2.0 is a more complex model that utilizes three clinical risk factors, five HMR mutations, and absence of type 1-like CALR mutation to stratify patients into five risk categories [26]. Sequential application of MIPSS70+ v2.0 to patients of intermediate risk is described above.
MIPSS70+ v2.0 utilizes the following features, and assigns points as follows [26]:
●Clinical risk factors:
•Severe anemia (Men: hemoglobin [Hb] <9 g/dL; women: Hb <8 g/dL): 2 points
•Moderate anemia (Men: Hb 9 to 10.9 g/dL; women: Hb 8 to 9.9 g/dL): 1 point
•Circulating blasts ≥2 percent: 1 point
•Constitutional symptoms: 2 points
●Karyotype (see 'Cytogenetic features' above):
•Very high risk (VHR): 4 points
•Unfavorable: 3 points
●Mutations (see 'High molecular risk (HMR) mutations' above):
•≥2 high molecular risk (HMR) mutations: 3 points
•One HMR mutation: 2 points
•Absence of type 1-like CALR: 2 points
Points are summed to assign a patient to one of five MIPSS70+ v2.0 prognostic categories with distinct 10-year OS and median OS, based on a cohort study of 406 patients [26]:
●Very low risk: 0 points (86 percent, not reached)
●Low risk: 1 to 2 points (50 percent, 10.3 years)
●Intermediate risk: 3 to 4 points (30 percent, 7 years)
●High risk: 5 to 8 points (10 percent, 3.5 years)
●Very high risk: ≥9 points (<3 percent, 1.8 years)
Compared with previous versions of this model (ie, MIPSS70 and MIPSS70-Plus), MIPSS70+ v2.0 established an additional prognostic category (ie, very high risk), to which one-sixth of patients were assigned without compromising the predictive power of the model [26].
Alternative approaches
Why select an alternative approach? — As discussed above, we favor prognostic tools that incorporate cytogenetic and molecular features, because they are the most effective models for risk stratification in PMF. (See 'Our preferred approach' above.)
Although tools that utilize only clinical and/or cytogenetic features are not as effective for stratifying patients into prognostic groups as our preferred models (ie, GIPSS and MIPSS70+ v2.0) [17,26], alternative models are acceptable in the following settings:
●If molecular analysis is not available, we favor use of DIPSS-Plus. (See 'DIPSS-Plus' below.)
●If neither cytogenetics nor molecular analysis is available, DIPSS is an acceptable prognostic model. (See 'DIPSS' below.)
DIPSS — Dynamic international prognostic scoring system (DIPSS) is our preferred prognostic tool in settings where neither cytogenetics nor molecular analysis is available. DIPSS is based entirely on clinical features (age, leukocyte count, hemoglobin, circulating blasts, constitutional symptoms) [27]. DIPSS is a dynamic tool, meaning that it can be applied at any time in the course of PMF (ie, not only at the time of diagnosis). DIPSS differs from IPSS, an earlier prognostic tool, in that it assigns greater weight to severe anemia, based on its greater hazard ratio for survival when compared with the other clinical features [2,27].
DIPSS was created from a database of 525 patients with PMF, and reported the following median values for OS [27]:
●Low risk: not reached
●Intermediate-1: 14.2 years
●Intermediate-2: 4 years
●High risk: 1.5 years
DIPSS also effectively stratifies these categories for progression to leukemia [12].
DIPSS-Plus — DIPSS-Plus (table 1) is our preferred prognostic tool in settings where molecular analysis is not available [13]. DIPSS-Plus differs from DIPSS in that it adds three additional prognostic features (ie, transfusion-requiring anemia, unfavorable karyotype, and platelet count <100,000/microL).
DIPSS-Plus was evaluated in 793 patients with PMF (figure 1), and reported the following values for median OS [13]:
●Low risk: 185 months
●Intermediate-1: 78 months
●Intermediate-2: 35 months
●High risk: 16 months
DIPSS-Plus is inferior to both GIPSS and MIPSS70+ v2.0 for discriminating between PMF prognostic categories [17].
SPECIAL POPULATIONS
Younger patients — There is no consensus regarding what constitutes a young patient in this setting, but PMF generally presents in middle-aged or older adults; one study reported that less than 10 percent of patients with PMF were ≤50 years old [4]. Few studies have focused exclusively on younger patients, but many large studies of PMF have identified younger age (generally <65 years old) as a favorable prognostic factor for survival in PMF [1-4,10,12-15]. (See "Clinical manifestations and diagnosis of primary myelofibrosis", section on 'Epidemiology'.)
We favor the approach described above to evaluate prognosis in younger patients with PMF. No studies have applied contemporary prognostic models (eg, GIPSS, MIPSS70+ v2.0) exclusively to populations of younger patients, but most large studies of PMF included patients ≤70 years old (eg, MIPSS70, MIPSS70-Plus, MIPSS70+ v2.0) [16,26,28]. (See 'Our preferred approach' above.)
One study of 123 patients <55 years old reported 128 month median survival (95% CI 90-172 months) [5]. This study identified hemoglobin <10 g/dL, presence of constitutional symptoms, and circulating blasts ≥1 percent as independent features associated with inferior survival, but it did not include cytogenetic and molecular findings. Using those criteria, two categories were described: Low risk (one adverse factor) had median survival of 176 months, and high risk (two or three factors) had a median survival of 33 months.
Prognosis in transplantation for PMF — Allogeneic hematopoietic cell transplantation (HCT) is the only potentially curative therapy for PMF. Prognostic variables associated with outcomes for patients who undergo allogeneic HCT for PMF are described separately. (See "Myelofibrosis (MF): Management of primary MF and secondary MF", section on 'Transplant-eligible patients'.)
Pregnancy — There is scant information concerning the management and prognosis of pregnant women with PMF. Fetal mortality is high, although some pregnancies have been brought to term [29-31].
SUMMARY
●Description – Primary myelofibrosis (PMF) is a myeloproliferative neoplasm (MPN) that manifests as bone marrow fibrosis, cytopenias, constitutional symptoms, hepatosplenomegaly, and/or extramedullary hematopoiesis.
●Survival – PMF has an estimated median overall survival (OS) of approximately six years and is associated with the least favorable prognosis among the BCR::ABL1-negative MPNs, which also include polycythemia vera and essential thrombocythemia. Patients with PMF are at risk for premature death due to disease progression, leukemic transformation, thrombo-hemorrhagic complications, and infections. (See 'Clinical outcomes in PMF' above.)
●Prognostic factors – Both clinical and molecular features are associated with prognosis:
•Clinical – Inferior survival in PMF is associated with certain clinical prognostic factors, including older age, constitutional symptoms (ie, fever, sweats, weight loss), anemia, leukocytosis, thrombocytopenia, circulating blasts, the degree of bone marrow fibrosis, and transfusion dependence. (See 'Clinical factors' above.)
•Genetic – Prognosis in PMF is also influenced by specific cytogenetic features, driver mutations (ie, JAK2, CALR, MPL), and certain high molecular risk (HMR) mutations (eg, ASXL1, SRSF2, U2AF1Q157, EZH2, IDH1/2). (See 'Genetics' above.)
●Assessment of prognosis – Our preferred approach (algorithm 1) for evaluating prognosis in PMF is to initially assess patients using GIPSS (genetically inspired prognostic scoring system) (calculator 1), an easy-to-use tool that utilizes only cytogenetic and molecular data. GIPSS stratifies patients into four prognostic groups: low, intermediate (int)-1, int-2, and high risk. For patients who are assigned to either GIPSS int-1 or int-2, we suggest subsequently evaluating the patient with MIPSS70+ v2.0 (mutation-enhanced international prognostic scoring system) (calculator 2), which incorporates clinical, cytogenetic, and molecular features. (See 'Our preferred approach' above.)
We prefer this two-step approach because GIPSS is a simple and effective tool for identifying patients who are at highest and lowest risk for death and transformation to leukemia. However, MIPSS70+ v2.0 is more effective at stratifying risk for patients of intermediate prognosis, and it reclassifies a significant percentage of GIPSS int-1 and int-2 patients, which may affect treatment decisions. (See 'GIPSS' above and 'MIPSS70+ v2.0' above.)
Use of MIPSS70+ v2.0 alone for all patients with PMF is an acceptable option for determining prognosis. (See 'MIPSS70+ v2.0' above.)
●Alternative approaches – Other methods to assess prognosis are acceptable in the following settings (see 'Why select an alternative approach?' above):
•If neither cytogenetics nor molecular analysis is available, DIPSS, which utilizes clinical features only, is an acceptable prognostic model. (See 'DIPSS' above.)
•If molecular analysis is not available, we favor use of DIPSS-Plus. (See 'DIPSS-Plus' above.)
ACKNOWLEDGMENT — The editors of UpToDate acknowledge the contributions of Stanley L Schrier, MD as Section Editor on this topic, his tenure as the founding Editor-in-Chief for UpToDate in Hematology, and his dedicated and longstanding involvement with the UpToDate program.
Do you want to add Medilib to your home screen?