Blood Journal
Leading the way in experimental and clinical research in hematology

International Working Group (IWG) consensus criteria for treatment response in myelofibrosis with myeloid metaplasia, for the IWG for Myelofibrosis Research and Treatment (IWG-MRT)

  1. Ayalew Tefferi,
  2. Giovanni Barosi,
  3. Ruben A. Mesa,
  4. Francisco Cervantes,
  5. H. Joachim Deeg,
  6. John T. Reilly,
  7. Srdan Verstovsek,
  8. Brigitte Dupriez,
  9. Richard T. Silver,
  10. Olatoyosi Odenike,
  11. Jorge Cortes,
  12. Martha Wadleigh,
  13. Lawrence A. Solberg Jr,
  14. John K. Camoriano,
  15. Heinz Gisslinger,
  16. Pierre Noel,
  17. Juergen Thiele,
  18. James W. Vardiman,
  19. Ronald Hoffman,
  20. Nicholas C. P. Cross,
  21. D. Gary Gilliland, and
  22. Hagop Kantarjian
  1. From the Mayo Clinic, Rochester, MN; Istituto di Ricovero e Cura a Carattere Scientifico Policlinico San Matteo, Pavia, Italy; Hospital Clinic, Institut d'Investigacions Biomèdiques August Pi i Sunyer, Barcelona, Spain; the Fred Hutchinson Cancer Research Center, Seattle, WA; the Royal Hallamshire Hospital, Sheffield, United Kingdom; the M. D. Anderson Cancer Center, Houston, TX; Service d'Hématologie Clinique, Centre Hospitalier de Lens, France; Cornell Medical Center, New York, NY; the University of Chicago, IL; the Dana Farber Cancer Institute, Boston, MA; the Mayo Clinic, Jacksonville, FL; the Mayo Clinic, Scottsdale, AZ; the Department of Hematology and Blood Coagulation, Medical University of Vienna, Austria; the National Institutes of Health, Bethesda, MD; the Institute of Pathology, University of Cologne, Germany; the University of Illinois, Chicago and the Wessex Regional Genetics Laboratory, Salisbury, United Kingdom.


Myelofibrosis with myeloid metaplasia (MMM) is a clinicopathologic entity characterized by stem cell-derived clonal myeloproliferation, ineffective erythropoiesis, extramedullary hematopoiesis, and bone marrow fibrosis and osteosclerosis. Patients with MMM have shortened survival and their quality of life is compromised by progressive anemia, marked hepatosplenomegaly, and severe constitutional symptoms including cachexia. After decades of frustration with ineffective therapy, patients are now being served by promising treatment approaches that include allogeneic hematopoietic stem cell transplantation and immunomodulatory drugs. Recent information regarding disease pathogenesis, including a contribution to the myeloproliferative disorder phenotype by a gain-of-function JAK2 mutation (JAK2V617F), has revived the prospect of targeted therapeutics as well as molecular monitoring of treatment response. Such progress calls for standardization of response criteria to accurately assess the value of new treatment modalities, to allow accurate comparison between studies, and to ensure that the definition of response reflects meaningful health outcome. Accordingly, an international panel of experts recently convened and delineated 3 response categories: complete remission (CR), partial remission (PR), and clinical improvement (CI). Bone marrow histologic and hematologic remissions characterize CR and CR/PR, respectively. The panel agreed that the CI response category is applicable only to patients with moderate to severe cytopenia or splenomegaly.


Myelofibrosis with myeloid metaplasia (MMM) is classified with polycythemia vera (PV) and essential thrombocythemia (ET) as a BCR/ABL-negative myeloproliferative disorder (MPD).1 Current diagnosis is based on morphologic assessment of bone marrow histology, peripheral blood examination, and complementary information from cytogenetic and molecular studies.2,3 The term MMM, for the purposes of the current communication, includes both chronic idiopathic myelofibrosis (ie, de novo presentation of the clinicopathologic phenotype, also known as agnogenic myeloid metaplasia) and cases in which the disease is preceded by either essential thrombocythemia (ie, postthrombocythemic myeloid metaplasia [PTMM]) or polycythemia vera (ie, postpolycythemic myeloid metaplasia [PPMM]). It also includes cases with a cellular (aka prefibrotic) phase of the disease. Previous studies have shown clinical and laboratory similarities between de novo MMM, PPMM, and PTMM, including the types and distribution of bone marrow cytogenetic abnormalities.4 Furthermore, the 3 MMM variants are now known to harbor a common mutant allele (JAK2V617F).5 Pathogenetically, MMM represents a stem cell-derived abnormal clonal myeloproliferation distinctly associated with reactive, cytokine-mediated marrow stromal changes including collagen fibrosis, osteosclerosis, and angiogenesis.3,6,7

Clinical course in MMM is characterized by progressive anemia, marked hepatosplenomegaly, profound constitutional symptoms including cachexia, and occasionally the development of nonhepatosplenic extramedullary hematopoiesis (EMH).8 Median survival ranges from fewer than 3 years to more than 10 years depending on the presence or absence of adverse risk factors: anemia, thrombocytopenia, leucopenia, leukocytosis, circulating blasts, constitutional symptoms, and poor-risk cytogenetic abnormalities.9-13 Causes of death include leukemic transformation, infections, bleeding, thrombosis, heart failure, liver failure, solid tumor, respiratory failure, and portal hypertension.14,15 Conventional treatment modalities, which include drugs,16 splenectomy,17 and involved field irradiation of an EMH site,18,19 neither alter the natural history of MMM nor provide durable symptom relief. The status quo is fortunately changing in the face of encouraging preliminary results from both allogeneic hematopoietic stem cell transplantation (HSCT)20-22 and novel drugs.23-25 Further progress is expected from the development and/or application of new agents that target disease-specific molecular abnormalities.26-28 It is therefore essential to establish uniform response criteria that allow valid interstudy comparisons and accurate assessment of the efficacy of newer therapies.

Materials and methods

The current document was compiled by an international panel of clinical and laboratory experts in MMM. The initial process involved review of response criteria used in (1) recent collaborative group29 or cell-based20,30 treatment trials, (2) a recent proposal by the European Myelofibrosis Network (EUMNET),31 and (3) published response criteria for myelodysplastic syndrome (MDS),32 which is a related chronic myeloid disorder. The strengths and limitations of each one of the aforementioned documents were discussed and new recommendations were proposed with the intent to (1) preserve some similarity to response categories in MDS,32 (2) ensure that a defined response has relevance to quality of life (QOL), and (3) incorporate treatment effects on bone marrow histology.33

Panel consensus regarding treatment objectives and definition of complete and partial remissions

Two primary objectives in the treatment of cancer are to prolong survival and to improve QOL. Among myeloid disorders, randomized studies have confirmed treatment-associated improvement in survival in both acute myeloid leukemia (AML)34-36 and chronic myeloid leukemia (CML).37,38 The same has not yet been definitively documented for either MDS or MMM, although this possibility exists in the context of modern therapeutic regimens, including allogeneic HSCT.20,21,39 The achievement of a strictly defined complete remission (CR) state has been shown to be a prerequisite for improvement in survival in many hematologic malignancies including AML and CML. Similarly, treatment-induced CR and partial remission (PR) in MDS have been associated with significantly longer survival.40,41 Therefore, the panel endorsed the inclusion of CR and PR categories in MMM that are similar but not identical to those used in MDS,32 with the intention to prove their validity in future studies (Tables 1, 2). In so doing, the CR designation in MMM fulfilled the requirement of bone marrow histologic remission, which distinguishes such a response category from those that imply clinical benefit without the potential to alter the natural history of the underlying disease. In contrast, CR and PR categories were not part of the MMM response criteria either published by the European Myelofibrosis Network (EUMNET)31 or previously published collaborative group clinical trials.29 Furthermore, the latter groups considered histologic remissions as a separate response category and not part of the major clinical response categories (Table 3). On the other hand, histologic remission has been an integral component of response criteria in the context of allogeneic HSCT.20,30

View this table:
Table 1.

International Working Group (IWG) consensus criteria for treatment response in myelofibrosis with myeloid metaplasia

View this table:
Table 2.

Peripheral blood and bone marrow requirements for complete (CR) and partial (PR) remissions as well as other remission categories in myelofibrosis with myeloid metaplasia (MMM) and myelodysplastic syndrome (MDS)

View this table:
Table 3.

Response criteria for myelofibrosis with myeloid metaplasia, which was recently proposed by the European Myelofibrosis Network (EUMNET)

In addition to reporting CR and PR, the current panel of experts strongly recommends the gathering and reporting of baseline and follow-up cytogenetic/molecular and other biologic information during clinical trials, as has been the case in CML and AML.42-45 Because many patients with MMM do not display a specific molecular or cytogenetic marker, and furthermore, it is unclear whether these abnormalities are primary or secondary events, the panel felt that it was premature, at the present time, to incorporate such information in formal response categories. However, the panel underscores the fact that none of the previously published or currently proposed response criteria in MMM have been validated in a prospective fashion, and the current lack of effective drug therapy limits one's ability to retrospectively examine the issue at hand. Accordingly, our main goal in the current proposal was not to undermine the potential value of cytogenetic/molecular responses but to establish widely applicable uniform response criteria that bear some similarity to those currently in use for MDS and AML (Table 2).32,46 Nevertheless, we would like to reiterate that monitoring disease status by cytogenetic and molecular analysis provides important supplementary information and should be clearly communicated along with other responses.

Panel consensus on the definition of “clinical improvement” as a third response category

In addition to strictly defined CR and PR categories, and as has been the case with the designation of hematologic improvement as a response category in MDS,32 the panel recommended the inclusion of an additional response category that reflected a measurable effect on QOL. The major parameters of relevance in MMM, in this regard, include symptomatic anemia (hemoglobin level, < 100 g/L [10 g/dL]; frequency of 35%-45%),9-11 marked splenomegaly (palpable at > 5 cm from left costal margin [LCM]; frequency of > 50%),9,11 and constitutional symptoms/cachexia (eg, fatigue, fever, night sweats, loss of lean body mass; frequency of approximately 20%).11 The panel felt that a suitable designation for this symptom-directed response in MMM would be clinical improvement (CI) rather than hematologic improvement because it includes splenomegaly response in addition to response in other hematologic parameters. It is to be noted that such responses have in the past been assigned a myriad of response categories that have not been uniform, across different groups of investigators, in both nomenclature and the levels of change that were required to constitute a response.29,31,47 Furthermore, previous response categories of clinical improvement did not always reflect an associated QOL benefit (Table 3).

Other disease manifestations with the potential to affect QOL include severe to moderate thrombocytopenia (platelet count, < 50 × 109/L) or neutropenia (absolute neutrophil count, < 1 × 109/L), painful hepatomegaly, and clinically overt nonhepatosplenic EMH (eg, pleural effusion, ascites, pulmonary hypertension, and spinal cord and nerve root compression).18 The choice of a relatively higher platelet count threshold for clinically relevant thrombocytopenia was based on the fact that MPD platelets are often qualitatively defective. On the other hand, a treatment effect on mild anemia (hemoglobin level, ≥ 100 g/L [10 g/dL]), mild asymptomatic splenomegaly (spleen palpable at ≤ 5 cm below LCM), nonblastic leukocytosis, asymptomatic thrombocytosis, mild leukopenia or thrombocytopenia, serum lactate dehydrogenase level, circulating myeloblasts, CD34+ cells, or endothelial progenitor cells rarely translates into meaningful health outcome, and may not warrant consideration for definition of response outside the context of CR and PR.9,15,48-52 Similarly, the panel felt that symptomatic thrombocytosis in MMM was too infrequent to warrant its inclusion as a parameter of response.


Until very recently,31 published response criteria in MMM did not exist. Both ongoing and recently completed clinical trials used in-house versions, with little uniformity among different studies. The recent effort from EUMNET31 to standardize response criteria in MMM is useful but may not have addressed the complexity and assignment of QOL relevance to response and did not incorporate histologic remission as part of the major clinical response categories, as has been the tradition in other myeloid disorders including MDS and AML (Table 3). The current panel of experts, which includes the lead authors from the EUMNET publication,31 proposes new recommendations with the specific intent to simplify response definitions, ensure that response is linked to meaningful health outcome and QOL, and preserve some similarity with response criteria used for MDS, a related chronic myeloid disorder.32

As elaborated in the previous section, the new set of recommendations is based on 3 response categories: CR, PR, and clinical improvement (CI) (Table 1). Both CR and PR designations require trilineage peripheral blood count remission; leukocyte differential including the absence of nucleated red blood cells, circulating blasts, and immature myeloid progenitor cells (in the absence of splenectomy); and complete resolution of disease-related symptoms and signs including palpable hepatosplenomegaly. In addition, CR requires the attainment of an operationally defined complete bone marrow histologic remission in terms of cellularity (normocellular for age), myeloblast percentage (≤ 5%), and osteomyelofibrosis (grade of ≤ 1; Table 4).33 Because of a high degree of subjectivity and lack of standardization in peripheral blood and bone marrow smear interpretation, the panel favored not including demonstration of normal erythrocyte or megakaryocyte morphology as a CR requirement. The requirements for PR are similar to CR but without the need to demonstrate bone marrow histologic remission. In this regard, the panel considers resolution of leukoerythroblastosis, a PR requirement, as a reflection of a favorable effect on bone marrow histology. However, the panel strongly recommends a repeat (ie, after treatment) bone marrow biopsy in PR patients and recognizes the possibility that less than complete histologic remission might occur in some of these patients. Accordingly, the MMM-relevant CR and PR response categories encompass the requirements needed for MDS-relevant CR or PR categories without being identical to them (Table 2).32

View this table:
Table 4.

European consensus on grading bone marrow fibrosis (fiber density should be assessed in hematopoietic cellular areas)

The CI response category in MMM (Table 1) is linked to positive experience in symptom relief and is therefore applicable only in the presence of symptomatic anemia (hemoglobin level, < 100 g/L [10 g/dL]), marked splenomegaly (palpable spleen size, > 5 cm from LCM), or either severe to moderate thrombocytopenia (platelet count, < 50 × 109/L) or neutropenia (absolute neutrophil count, < 1 × 109/L). The panel fully recognizes the fact that the current response criteria do not capture all treatment effects. Multiple factors were discussed as the basis for this decision. For example, the measurement of constitutional symptoms is not always accurate because it involves subjective interpretation of a nonstandardized patient report. Furthermore, constitutional symptoms are closely linked to the presence of marked hepatosplenomegaly and display a similar direction of response to treatment. Similarly, symptomatic nonhepatosplenic EMH is an infrequent complication with exquisite response to low-dose involved field irradiation.18 On the other hand, it is unusual for either mild thrombocytopenia/neutropenia or thrombocytosis/leukocytosis to directly affect QOL in MMM. Therefore, such parameters along with serum LDH, leukoerythroblastosis, and circulating CD34+ cell count should be monitored during clinical trials but not used to define formal response categories outside the context of CR and PR. In this regard, the new IWG criteria differ from those prescribed by the EUMNET, which are outlined in Table 3. Similarly, unlike both the EUMNET and the IWG for MDS, the current IWG for MMM does not consider reduction in transfusion requirement as constituting a response that is significant enough to be included in any of the major response categories. The reason for this decision concerns the difficulty in acquiring accurate quantification of baseline and posttreatment transfusion requirements. On the other hand, whereas both IWG and EUMNET for MMM require a minimum of 20-g/L (2 g/dL) increase in hemoglobin level for an erythroid response, the corresponding hemoglobin increment level used by the IWG for MDS is 15 g/L (1.5 g/dL). It is also important to note that the 3 sets of response criteria differ in the parameters they use in defining platelet and neutrophil responses as well as disease progression.

Cytogenetic and molecular responses

The panel also recommends baseline and follow-up evaluation of either bone marrow or peripheral blood cytogenetic and molecular studies (eg, JAK2V617F mutation analysis) to further define quality/depth of CR, clarify equivocal histologic remissions, and obtain preliminary information on the value of minimal residual disease monitoring. Our cytogenetic response criteria are similar to those used by the IWG for MDS32 and differ from the EUMNET criteria, which use the terms major and minor instead of complete and partial, respectively (Table 3); a complete cytogenetic response is defined as failure to detect a cytogenetic abnormality in cases with a pre-existing abnormality, and a partial cytogenetic response should be defined as 50% or greater reduction in abnormal metaphases, with at least 20 bone marrow-derived metaphases being analyzed in both cases. Molecular response is currently applicable only to cases that are JAK2 V617F positive, and we suggest that major molecular response be defined as the absence of the mutation in peripheral blood granulocytes of previously positive cases using the most sensitive methods currently available (allele-specific polymerase chain reaction [PCR], real-time PCR, or pyrosequencing, all of which have sensitivity of approximately 2%-5%). The sensitivity of the assay used should be established in each center and stated as a percentage. Finally, while it is appropriate to incorporate QOL assessment in clinical trials, such practice is unlikely to enrich the experience from phase 1/2 studies and is best reserved for large phase 3 studies. In this regard, a recent QOL study in MPD involving 1179 patients is being analyzed and will be used to construct guidelines for assessing QOL in MMM.

Confounders of response and study eligibility

The panel favors a treatment-free period of at least 4 weeks for patient participation in a clinical trial in order to minimize the confounding effect of previous therapy on study drug response. For example, a patient receiving hydroxyurea therapy might experience an improvement in anemia and other cytopenias as well as worsening of splenomegaly, cytosis, and constitutional symptoms as a result of drug discontinuation. This effect might be inaccurately attributed to the study drug unless an adequate period of time has elapsed between the discontinuation of hydroxyurea and initiation of protocol therapy. A similar scenario is possible in patients in whom growth factor treatment is discontinued close to study accrual. The panel recognizes the difficulty in withholding therapy for 4 weeks in the presence of active disease, in some patients who are otherwise fit for investigational therapies. The alternative possibility of allowing study participation without altering baseline disease-directed therapy might undermine accurate assessment of both drug effect and toxicity. However, because the aforementioned issues are unlikely to be a factor in the context of CR or PR, the panel was comfortable in considering, in some instances, patient participation in clinical trials regardless of the presence or absence of active therapy. Clinical trials that allow patients to continue pre-existing drug therapy should require the use of stable drug doses for at least 3 months prior to protocol entry and should include patient stratification that would allow assessment of CI based on baseline treatment status.

Another issue during clinical trials in MMM involves the distinction between potentially reversible drug-induced myelosuppression and disease progression. This is highlighted by the experience in CML with imatinib53,54 and in MDS with either lenalidomide55 or hypomethylating agents,40,41 where drug-associated myelosuppression often preceded response. It is therefore conceivable that an effective drug might be discontinued prematurely if its effect on blood counts is spuriously attributed to disease progression. This is the basis for not including anemia and cytopenias as markers of disease progression during protocol therapy. Instead, study drug-associated alterations in blood count should be used primarily for purposes of dose modification. However, a decrease in hemoglobin level of 20 g/L (2 g/dL) or more, acquiring transfusion dependency, or a 100% increase in transfusion requirement, each persisting for more than 3 months after drug discontinuation, should ultimately be considered as a marker of disease progression unless proved otherwise. On the other hand, the on-protocol criteria for disease progression include progressive splenomegaly, bone marrow biopsy-proven leukemic transformation per WHO criteria, or a persistent (ie, ≥ 8 weeks duration) increase in circulating blasts to 20% or higher (Table 2). Finally, as elaborated before, the current recommendations for response criteria do not include measurements of drug effect on constitutional symptoms and asymptomatic splenomegaly that is palpable at 5 cm or fewer from LCM; this minimizes inaccuracies stemming from subjective interpretations and avoids the need for objective confirmation by ultrasound.


Members of the International Working Group for Myelofibrosis Research and Treatment include the following (in alphabetical order): Giovanni Barosi, Istituto di Ricovero e Cura a Carattere Scientifico Policlinico San Matteo, Pavia, Italy; John K. Camoriano, Mayo Clinic, Scottsdale, AZ; Peter J. Campbell, University of Cambridge, United Kingdom; Francisco Cervantes, Institut d'Investigacions Biomèdiques, August Pi i Sunyer, Barcelona, Spain; Jorge Cortes, M. D. Anderson Cancer Center, Houston, TX; Nicholas C. P. Cross, Wessex Regional Genetics Laboratory, Salisbury, United Kingdom; H. Joachim Deeg, Fred Hutchinson Cancer Research Center, Seattle, WA; Brigitte Dupriez, Service d'Hématologie Clinique, Centre Hospitalier de Lens, France; Francis Giles, M. D. Anderson Cancer Center, Houston, TX; D. Gary Gilliland, Dana Farber Cancer Institute, Boston, MA; Heinz Gisslinger, Medical University of Vienna, Austria; Ronald Hoffman, University of Illinois, Chicago; Catriona H. M. Jamieson, University of California, San Diego; Hagop Kantarjian, M. D. Anderson Cancer Center, Houston, TX; Hans M. Kvasnicka, Institute of Pathology, University of Cologne, Germany; Ross Levine, Brigham and Women's Hospital, Boston, MA; Ruben A. Mesa, Mayo Clinic, Rochester, MN; Stephen D. Nimer, Memorial Sloan-Kettering Cancer Center, New York, NY; Pierre Noel, National Institutes of Health, Bethesda, MD; Olatoyosi Odenike, University of Chicago, IL; Animesh Pardanani, Mayo Clinic, Rochester, MN; John T. Reilly, Royal Hallamshire Hospital, Sheffield, United Kingdom; Richard T. Silver, Cornell Medical Center, New York, NY; Lawrence A. Solberg Jr, Mayo Clinic, Jacksonville, FL; Ayalew Tefferi, Mayo Clinic, Rochester, MN; Juergen Thiele, Institute of Pathology, University of Cologne, Germany; Deborah Thomas, M. D. Anderson Cancer Center, Houston, TX; James W. Vardiman, University of Chicago, IL; Srdan Verstovsek, M. D. Anderson Cancer Center, Houston, TX; and Martha Wadleigh, Dana Farber Cancer Institute, Boston, MA.


Jointly organized by the Mayo Clinic (Rochester, MN) and the M. D. Anderson Cancer Center (Houston, TX).


  • Reprints:

    Ayalew Tefferi, Mayo Clinic, 200 First Street SW, Rochester MN 55905; e-mail: tefferi.ayalew{at}
  • Prepublished online as Blood First Edition Paper, May 4, 2006; DOI 10.1182/blood-2006-03-009746.

  • A complete list of the members of the IWG-MRT appears in “Appendix.”

  • Supported by the Joe W. and Dorothy Dorsett Brown Foundation (Metairie, LA) and the William Waugh and Judy Olin Higgins Research Fund of the Cancer Research and Treatment Fund (New York, NY).

  • The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 U.S.C. section 1734.

  • Submitted March 13, 2006.
  • Accepted April 23, 2006.


View Abstract