Primary myelofibrosis




Primary myelofibrosis is one of several disorders in the spectrum of clonal myeloid diseases, malignant diseases that originate in the clonal expansion of a single neoplastic hematopoietic multipotential cell. Approximately 50 percent of cases have a mutation in the Janus kinase 2 (JAK2) gene. It is characterized, classically, by anemia, mild neutrophilia, thrombocytosis, and splenomegaly. Occasional cases may present with bi- or tricytopenias (~10%). Immature myeloid and erythroid precursors, teardrop-shaped erythrocytes, and large platelets are constant features of the blood film. The marrow contains an increased number of pathologic megakaryocytes and increased reticulin fibers and, often later, collagen fibrosis. This reactive, polyclonal fibroplasia is the result of cytokines (e.g., transforming growth factor [TGF]-) released locally by the numerous abnormal megakaryocytes. The disease may be complicated by portal hypertension as a result of a very large splenic blood flow and loss of compliance of hepatic vessels and by fibrohematopoietic tumors that can develop in any tissue and lead to symptoms by compression of vital structures. Treatment may include hydroxyurea for thrombocytosis and massive splenomegaly, androgens, erythropoietin, or red cell transfusions for severe anemia, local irradiation of fibrohematopoietic tumors or of the spleen, and splenectomy. Trials of newly developed JAK2 inhibitors show beneficial effects but require further study. Portosystemic shunt surgery may be required for gastroesophageal variceal bleeding. In younger patients, allogeneic hematopoietic stem cell transplantation can be curative and nonmyeloablative transplantation has been successful, at least up to age 65 years, and may become the approach of choice at any age. The disease may remain indolent for years or may progress rapidly by further deterioration in hematopoiesis, by massive splenic enlargement and its sequelae, or by transformation to acute myelogenous leukemia. Overall median survival is approximately 5 years.

Acronyms and Abbreviations

Acronyms and abbreviations that appear in this chapter include: AML, acute myelogenous leukemia; bFGF, basic fibroblast growth factor; CD, cluster of differentiation; CML, chronic myelogenous leukemia; FISH, fluorescence in situ hybridization; G-6-PD, glucose-6-phosphate dehydrogenase; G-CSF, granulocyte colony-stimulating factor; IL, interleukin; MRI, magnetic resonance imaging; PDGF, platelet-derived growth factor; TGF, transforming growth factor; TNF-R, tumor necrosis factor-receptor.

Definition and HistoryPrimary myelofibrosis is a chronic clonal myeloid disorder characterized by (1) anemia; (2) splenomegaly; (3) immature granulocytes, increased CD34+ cells, erythroblasts, and teardrop-shaped red cells in the blood; (4) marrow fibrosis; and (5) osteosclerosis. The disorder originally was described by Heuck1 in 1879 under the title “Two Cases of Leukemia and Peculiar Blood and Bone Marrow Findings.” In his monograph, Silverstein traced the history of the concepts set forth during the first half of the 20th century to explain the pathogenesis of this disease, including its origin in the marrow, the appearance of extramedullary hematopoiesis, and the relationship of fibrosis to hematopoietic changes.2 More than 20 designations for the disease have been proposed or used, and different designations were preferred in different countries.3Primary myelofibrosis has been designated the most recent “official” name of the disease by a working group on nomenclature.3 This compromise selection is debatable as the fibrosis is secondary, not primary, and the choice omits focusing on the central pathologic change: a clonal myeloid disease with singular neoplastic megakaryocytopoiesis.4 The discovery that the mutated Janus kinase 2 (JAK2) gene plays a role in the causation and behavior of myeloproliferative diseases5 and that approximately 50 percent of cases have a mutation of the JAK2 gene6 has led to better understanding of the pathogenesis of the disease and its relationship to other myeloproliferative diseases. The JAK2 mutation also represents an important new target for therapy.



Age and Sex

Primary myelofibrosis characteristically occurs after age 50 years.2,7–15 The median age at diagnosis is approximately 65 to 70 years,7,11–13,16 but the disease can occur from the neonatal period to the ninth decade of life.2,11,13,17–19 In infants, the disorder can mimic the classic disease or show certain features but not others, such as absence of hepatosplenomegaly.18 Familial infantile myelofibrosis mimics the adult disease and in some cases is transmitted by autosomal recessive inheritance.20–22 The occurrence of primary myelofibrosis in children usually is in the first 3 years of life.19,23,24 In young children, girls are afflicted with the disease twice as frequently as boys.18 In young and middle-aged adults, the disease is similar to that in older subjects, although the proportion of indolent cases may be higher.17,19,25 In adults, the disease occurs with about equal frequency in men and women.7,11–15 Like virtually all clonal myeloid diseases,26 primary myelofibrosis can cluster in families, suggesting transmission of an unidentified predisposition gene.27–29 A large Swedish study found a significant relative risk (five- to sevenfold) for a familial occurrence of another myeloproliferative disease neoplasm, although not specifically primary myelofibrosis. The latter finding may relate to the small number of cases of primary myelofibrosis in that study.16 The incidence of the disease is approximately 0.5 cases per 100,000 population per year in northern European countries.30–33 A survey in Olmstead County, MN, reported an incidence of 1.5 case per 100,000 population per year and a median age of onset of 67 years. This median age is in keeping with several reports based on other studies (see above in this section).34

Etiology and Pathogenesis

Exogenous Factors

Exposure to benzene35–37 or very-high-dose ionizing radiation38 preceded the development of primary myelofibrosis in a very small proportion of patients with the disease. The former inciting agent, in exposures greater than 40 ppm-years, is associated with an increased relative risk of acute myelogenous leukemia (AML). Radiation is a well-established environmental cause of AML and chronic myelogenous leukemia (CML; see Chaps. 89 and 90).

Immune Mechanisms

Reports of myelofibrosis in patients with lupus erythematosus, have suggested the possibility of immunologic-mediated hyperplasia of marrow connective tissue2 (see “Immune and Inflammatory Manifestations” below). These forms of myelofibrosis are different from the monoclonal multipotential hematopoietic stem cell disease, which is the principal subject considered in this chapter.

Clonal Hemopathy, Animal Models, and Activating Mutations

The disease arises from the neoplastic transformation of a single hematopoietic multipotential cell, a conclusion derived from the presence of clonal cytogenetic abnormalities in patients with an identifiable chromosomal abnormality and in studies in women with primary myelofibrosis who also were heterozygous for isotypes A and B of glucose-6-phosphate dehydrogenase (G-6-PD).39,40 Although the nonhematopoietic tissues of these patients expressed both isotypes, each patient had blood cells with only one G-6-PD isotype. The findings strongly imply the blood cells of each patient arose from only one transformed stem cell. Furthermore, chromosome studies of colonies of hematopoietic progenitor cells in primary myelofibrosis established that the same clonal cytogenetic abnormality is present in erythroblasts, neutrophils, macrophages, basophils, and megakaryocytes.41 These studies were confirmed by (1) examining X-linked restriction fragment length polymorphisms in women with primary myelofibrosis with heterozygosity for X chromosome-linked genes42,43 and (2) verifying the presence of a mutation of codon 12 of the N-ras gene in five blood cell lineages of a patient with the disease.44,45 Lymphocyte derivation from the clone has been noted using mutation in codon 12 of the RAS gene as the marker.44 Using fluorescence in situ hybridization (FISH) analysis, T and B lymphocytes were found to be derived from clonal expansion of a multipotential hematopoietic cell in 3 of 4 patients with primary myelofibrosis with a 13q– or 20q– clonal cytogenetic abnormality.46 Primary myelofibrosis can be distinguished from secondary myelofibrosis in women by clonality studies.47 The advent of the JAK2 V617F mutation has permitted this marker to be used in assessing clonality. Mutated JAK2-containing cells were identified in all blood lineages and in the common lymphomyeloid cell.48

The neoplastic hematopoietic stem cells in primary myelofibrosis containing the JAK2 V617F mutation behave differently from the same cell population in polycythemia vera when studied in nonobese diabetic severe combined immunodeficient mice. Although studied in a nonhuman environment, the findings provide some explanation for the presence of JAK2 V617F mutations in three phenotypically different myeloproliferative diseases: polycythemia vera, essential thrombocythemia, and clonal myelofibrosis.49

Animal models followed the derivation of the murine myeloproliferative leukemia virus, carrying the oncogene v-mpl, which in mice produced a syndrome having features of a mixed idiopathic myelofibrotic–polycythemic disorder (see Chap. 113).50 The availability of v-mpl led to the isolation of the thrombopoietin receptor and its ligand thrombopoietin.51 Later models of myelofibrosis and osteosclerosis, mimicking some of the important features of human primary myelofibrosis, were induced in mice by retroviral-mediated overexpression of thrombopoietin.52,53 The concomitant high levels of fibroblastic factors (transforming growth factor [TGF]-1 and platelet-derived growth factor [PDGF]) resulted in intense fibrosis.54 In this model, increased osteoprotegerin was thought to be the principal cause of osteosclerosis.55 The disease was cured by murine hematopoietic stem cell transplantation.52

A syndrome in mice that results from the GATA-1 (low) mutation also leads to a phenotype that closely simulates human myelofibrosis. The mice gradually develop anemia, teardrop poikilocytes, myeloid immaturity, marrow fibrosis, extramedullary hematopoiesis, and overexpression of profibrotic cytokines in marrow.56 GATA-1 is a transcription factor required for normal megakaryocyte development. GATA-1 deficiency in mice leads to increased megakaryocytic proliferation, followed by myelofibrosis and osteosclerosis, as a result of exaggerated elaboration of fibroblast-inducing and osteoblast-stimulating factors.57,58

The discovery in 2005 that a somatic mutation in JAK2 was associated with the three major myeloproliferative diseases—polycythemia vera, essential thrombocythemia, and primary myelofibrosis—has rapidly led to a fuller understanding of the pathogenesis of these diseases.44,59 A dominant, gain-of-function mutation in the gene JAK2 residing on the short arm of chromosome 9, which encodes the JAK2 tyrosine kinase, is present in approximately 50 percent of patients with primary myelofibrosis, in approximately 95 percent of patients with polycythemia vera (see Chap. 86), and in approximately 40 percent of patients with essential thrombocythemia (see Chap. 87), but is absent in healthy individuals.60,61 In confirmation, the expression of the mutated human JAK2 gene transferred into mice can induce a myeloproliferative disease with features characteristic of the human disorders.62–64 Homozygosity results from allelic duplication as a result of uniparental disomy of chromosome 9p, not loss of the normal allele corresponding to the mutation.65

It is not yet precisely known how JAK2 V617F, the most prevalent mutation, links the three diseases and what modifiers explain the dramatically different phenotype and expected survival of the patients with polycythemia and primary myelofibrosis. At least four other modifying factors have been proposed to account for the different phenotypes and the apparent absence of the mutation in a high proportion of patients with primary myelofibrosis: (1) gene dosage, (2) germ line modifiers, (3) predisposition alleles, and (4) additional somatic mutations.59,61,66 An example of each influence follows. There is an increasing JAK2 V617F allele burden from essential thrombocythemia, to polycythemia vera, to primary myelofibrosis.67 For example, the JAK2 V617F allele burden may be a key determinant of the degree of myeloproliferation and myeloid metaplasia reflected by significantly higher levels of white blood cell counts, CD34+ cell counts, lower platelet counts, and a higher frequency of splenomegaly in homozygous polycythemia vera patients compared to their heterozygous counterparts. These findings are consistent with JAK2 V617F-positive chronic myeloproliferative disorders as a biologic continuum with phenotypic presentation in part influenced by JAK2 V617F mutational load.67 Also, single nucleotide polymorphisms may influence the phenotype that results from the JAK2 mutation. Myeloproliferative neoplasm predisposition alleles could provide a selective advantage for the development of mutations in the JAK2 signaling pathway. As yet unidentified, pre-JAK2 alleles, which arise in cells prior to JAK2 mutation, may contribute to the phenotype displayed.61

A mutation in the thrombopoietin receptor gene, MPL, has been found in some mutant JAK2-negative patients with primary myelofibrosis. The finding of an activating JAK2 (~50% of patients) or MPL (~10% of patients) mutation, which employs JAK2 for signaling, reinforces the critical role of unregulated activation of the JAK-STAT (signal transducer and activator of transcription) signaling pathway in the pathogenesis of primary myelofibrosis.68–70

Isolated genetic findings in individual patients have included (13q14) deletions, mutation or overexpression of the retinoblastoma gene,71,72 NF1 (17q11) deletions,73 RAS mutations in approximately 1 in 20 patients studied, and occasional patients with mutations in KIT.71 Mutational analysis of the class III receptor tyrosine kinase genes KIT, FMS, and FLT3 in 40 to 60 patients with idiopathic myelofibrosis found only 2 mutation in FMS.71 Uniparental disomy has been found on chromosomes 9p (site of JAK2) and 1p.72

HMGA2, a gene on chromosome 12, normally is not expressed in humans and is implicated in mesenchymal tumors. HMGA2 was expressed in 12 of 12 patients with idiopathic myelofibrosis studied, implying that, if confirmed, expression of this gene in myeloid cells may play a role in the disease.73

Centrality of CD34+ Cell Egress and Neoplastic Megakaryocytopoiesis

Increased neoplastic megakaryopoiesis is the most prominent alteration in this clonal disease and is responsible for most of its major manifestations. Constitutive mobilization and circulation of CD34+ cells are a prominent feature of the clonal expansion. This phenomenon is the result of epigenetic methylation of the CXCR4 promotor, a resultant decrease in CXCR4 messenger ribonucleic acid (mRNA), decreased expression of CXCR4 on CD34+ cells, and their resultant enhanced migration into the blood in primary myelofibrosis patients.74

Circulating CD34+ cells in patients with primary myelofibrosis generate about 24-fold the number of megakaryocytes in culture than do CD34+ cells from normal subjects, express increased levels of BCL-XL, and have delayed apoptosis.75,76 Media conditioned with CD61-positive cells (presumptive megakaryocytes) elaborated greater quantities of growth factors and proteases, including TGF- and metalloprotease-9, than did CD61-positive cells generated from normal CD34+ cells.

Circulating CD34+ cells in patients with primary myelofibrosis also had a higher expression of eight genes (CD9, GAS2, DLK1, CDH1, WT1, NFE2, HMGA2, and CXCR4) than did normal CD34+ cells. These genes or subsets of them are likely related to disease pathogenesis and were shown to be related to specific manifestations in patients (e.g., CD9 and DLK1 with platelet count, WT1 with severity score).77

Enhanced Angiogenesis

Microvessel density and marrow blood flow are increased in patients with myelofibrosis. These changes may be related to an increase in circulating endothelial cell progenitors.78

Dysfunction of Hematopoiesis

Neoplastic myeloproliferation usually is the dominant marrow abnormality in the granulocytic and megakaryocytic lineages resulting in intensely cellular marrows and mild to moderate blood granulocytosis and thrombocytosis. Ineffective or hypoplastic hematopoiesis, resulting from exaggerated apoptosis of very early precursors, can be present initially or emerge later as the dominant pathogenetic process, leading to granulocytopenia and/or thrombocytopenia. Anemia is a frequent finding and results from a combination of decreased erythropoiesis, shortened red cell survival, and the effects of splenomegaly on the distribution of red cells in the circulation. Hemolysis can be a prominent factor in some cases. Megakaryocytosis and intense dysmorphogenesis of megakaryocytes are constant features of the disease. Even in intensely fibrotic marrows with severe decreases in erythroid and granulocytic precursors, clusters of megakaryocytes are easily found interspersed between collagen bundles. The term “megakaryocytic myelosis,” one of the many synonymous terms for the disease, catches the constancy of this finding. The dominance of megakaryopoiesis may relate to the average fivefold overexpression of FKBP51 in megakaryocytes in primary myelofibrosis and the marked predisposition of CD34+ cells to differentiate into megakaryocytes (see “Centrality of CD34+ Cell Egress and Neoplastic Megakaryocytopoiesis” above). FKBP51 increases resistance to apoptosis, possibly by an effect through the calcineurin pathway.79 The disease has all the hallmarks of chronic megakaryocytic leukemia.7 Although elevated levels of thrombopoietin (and interleukin [IL]-6 and IL-11) are found in the serum of patients with primary myelofibrosis, their etiologic role in the human disease is unresolved.80 A marked increase in the thrombopoietin receptor MPL is observed on the platelets and megakaryocytes of a proportion of patients with primary myelofibrosis.81 Expression of the polycythemia rubra vera gene PRV-1, also is increased on neutrophils in some patients with the disease.82,83 The latter group may include patients whose disease is evolving from polycythemia vera to myelofibrosis. In contrast, congenital overexpression of thrombopoietin produces a syndrome that resembles essential thrombocythemia. Despite the animal models of thrombopoietin-induced myeloproliferation and osteomyelofibrosis and the apparent abnormality of MPL receptor sites on human megakaryocytes, autonomous megakaryocyte growth, characteristic of human primary myelofibrosis marrow in culture, has not been associated with either an autocrine effect of MPL ligand (thrombopoietin) or of a mutation in MPL.


Four of the five major types of collagen84 are present in normal marrow: type I in bone, type III in blood vessels, and types IV and V in basement membranes. The fine reticulin fibers that appear after silver impregnation of marrow are principally type III collagen. They do not stain with trichrome dyes. The thicker collagen fibers are principally type I collagen and stain with trichrome dyes, but do not impregnate with silver. The amount of the very fine fibrous network barely perceptible in normal marrow that is stained by silver impregnation techniques85 increases in the marrow of patients with primary myelofibrosis (Table 91–1).86 The fibrous network contains collagen and occasionally progresses to include thick collagen bands that are evident with trichrome stains. Collagen types I, III, IV, and V are increased in myelofibrosis, but type III collagen is increased uniformly and preferentially.87–90 The latter occurrence accounts for the increased plasma concentration of procollagen III amino-terminal peptide, a component of collagen type III, which is cleaved during collagen biosynthesis.86,91,92 Serum prolyl-hydroxylase and marrow and plasma fibronectin also increase in patients with idiopathic myelofibrosis or myelofibrosis from other causes.88,89

Table 91–1. Fibroplasia in Idiopathic Myelofibrosis
I. Marrow Stroma
A. Increased amount of
1. Total collagen (hydroxyproline)87,91
2. Type I collagen87–89,93
3. Type III collagen87–89,93
4. Type III procollagen88–91,93,94
5. Type IV collagen88,95,96
6. Matrix metalloproteinase-1497
7. Bone morphogenetic protein98
8. Laminin88,95,99
9. Fibronectin100,101
10. Tenascin102
11. Vitronectin103
12. Microenvironment TGF-,104 bFGF,104 and substance P105
B. Decreased amount of
1. Collagenase97
II. Plasma
A. Increased concentration of
1. Prolylhydroxylase106
2. C-terminal peptide of procollagen type I90
3. N-terminal peptide of procollagen type III89,91,107,108
4. Type IV collagen89,99
5. Laminin89,99
6. Fibronectin101
7. Hyaluronan109

Marrow fibrosis in primary myelofibrosis is most closely correlated with increased dysmorphic megakaryocytes in the marrow. Even densely fibrotic marrow with little residual granulopoiesis or erythropoiesis usually has numerous megakaryocytes scattered throughout the fibrotic areas.86,92,110 The increased pathologic emperipolesis (the entry of neutrophils and other marrow cells into the canalicular system of megakaryocytes) of neutrophils in megakaryocytes, evident in human primary myelofibrosis and in mouse models, suggests this may be an additional mechanism of -granule injury and release of TGF- and PDGF.111 Animal models also indicate that marrow monocytes and macrophages may play a subsidiary role in the induction of fibrosis.111–113 Secretion of PDGF, basic fibroblast growth factor (bFGF), and TGF- from monocytes that are part of the clone have the potential to act as myeloproliferative growth factors and profibrotic cytokines.104

The increased content of marrow collagen types I and III results from release of fibroblast growth factors, which include PDGF,114,115 epidermal growth factor,116 endothelial cell growth factor,116 TGF-,103,117,118 and bFGF,109,119 each of which is present in megakaryocyte granules. Other factors, such as tumor necrosis factor alpha, IL-1, and IL-1, which can be released from marrow cells, also can stimulate fibroblasts.120,121 Platelet factor 4, also derived from megakaryocytes, inhibits collagenase and could contribute to collagen accumulation,110 although studies showing a poor correlation between plasma platelet factor 4 concentration and marrow fibrosis have dampened enthusiasm for the role of this factor.122 Substance P, a peptide that acts as a neurotransmitter and a modulator of immune and hematopoietic functions, is increased in the fibrotic marrow and colocalizes with fibronectin. It is angiogenic and is a fibroblast mitogen.104 Its precise role in the complex interactions among fibroblasts, cytokines, and matrix protein deposition is not clear. The high urinary excretion of platelet-derived calmodulin, a putative fibroblast growth factor, in patients with myelofibrosis has added this compound to the array of factors that may contribute to the fibroplasia.120 The plasma level of matrix metalloprotein III is decreased and the level of tissue inhibitor of metalloproteinase is increased in patients with idiopathic myelofibrosis.123 The expression of matrix metalloproteinase-14 in marrow increases by nearly two orders of magnitude as fibroplasia progresses during the course of the disease; and, megakaryocytes and endothelial cells are the major sources of this protein.97 Neutrophil collagenase (matrix metalloproteinase-8) content is decreased early in the disease.97 Bone morphogenetic proteins (BMPs) also have been implicated as a contributory factor in fibroplasia. BMP1, 6, and 7, and BMP-receptor 2 are increased in marrow in myelofibrosis as a result of release from megakaryocytes and stromal cells. These proteins are activators of latent TGF-1 and processors of collagen precursors. In addition, TGF-1 induces release of BMP6.98

This complex combination of alterations contributes to matrix deposition. The pathogenetic role of released growth factors in fibroplasia is not completely understood. Generalizations from in vitro experiments or correlation between two variables provide only a limited perspective. For example, TGF- can stimulate or inhibit fibroblast growth, depending on the repertoire of other growth factors in the environment.117,118

Fibroplasia is associated with an increase in the number and size of marrow sinuses,101 the number of endothelial cells,125 an increase in vascular volume in the marrow,103 and an increase in blood flow through the marrow.95,126,127 These processes are responsible for the increase in marrow collagen types IV and V and laminin synthesized by endothelial cells in the marrow of patients.116

The fibroblastic proliferation in marrow is not an intrinsic part of the abnormal clonal expansion of hematopoiesis.128 In cases of primary myelofibrosis in which G-6-PD isoenzyme studies or chromosome karyotyping establish monoclonal growth of hematopoietic cells, marrow fibroblasts contain both G-6-PD isoenzymes and do not share the clonal chromosome abnormality.129 The findings strongly imply that the fibroblasts differentiate from a primordial cell different from the neoplastic hematopoietic stem cell in primary myelofibrosis and that fibroblast proliferation and enhanced collagen synthesis are secondary results of abnormal hematopoiesis.

Extramedullary Hematopoiesis

Extramedullary hemopoiesis is consistently present in liver and spleen, where it contributes to organ enlargement.7–9 Escape of progenitor cells from marrow and their lodgment in other organs contributes to extramedullary blood cell formation. Reversion of the liver and spleen to their fetal hematopoietic functions (metaplasia) is not a major factor in extramedullary hematopoiesis, and quantitatively significant, effective hematopoiesis does not occur outside of the marrow (see “Fibrohematopoietic Extramedullary Tumors” below