Guideline for the diagnosis and management of myelofibrosis
John T. Reilly,1 Mary Frances McMullin,2 Philip A. Beer,3 Nauman Butt,4 Eibhlean Conneally,5 Andrew Duncombe,6
Anthony R. Green,7 N. George Michaeel,8 Marie H. Gilleece,9 Georgina W. Hall,10 Steven Knapper,11 Adam Mead,10 Ruben
A. Mesa,12 Mallika Sekhar,13 Bridget Wilkins,8 Claire N. Harrison8 and Writing group: British Committee for Standards in
Haematology
1Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, 2Queen’s University Hospital, Belfast, UK, 3Terry Fox Laboratory,
BC Cancer Agency, Vancouver, BC, Canada, 4Wirral University Teaching Hospital, Wirral, UK, 5St. James’s Hospital, Dublin,
Ireland, 6University Hospital Southampton NHS Foundation Trust, Southampton, 7Cambridge Institute for Medical Research,
Cambridge, 8Guy’s and St Thomas’ NHS Foundation Trust, London, 9St. James University Hospital, Leeds, 10Oxford University
Hospitals NHS Trust, Oxford, 11University Hospital of Wales, Cardiff, Wales, UK, 12Mayo Clinic, Scottsdale, AZ, USA and 13Royal
Free Hampstead NHS Trust, London, UK
Summary
The guideline group regarding the diagnosis and management
of myelofibrosis was selected to be representative of UK-based
medical experts, together with a contribution from a single
expert from the USA. MEDLINE and EMBASE were searched
systematically for publications in English from 1966 until
August 2011 using a variety of key words. The writing group
produced the draft guideline, which was subsequently revised
by consensus of the members of the General Haematology and
Haemato-oncology Task Forces of the British Committee for
Standards in Haematology (BCSH). The guideline was then
reviewed by a sounding board of UK haematologists, the BCSH
and the British Society for Haematology Committee and
comments incorporated where appropriate. The criteria used to
state levels and grades of evidence are as outlined in the Proce-
dure for Guidelines commissioned by the BCSH; the ‘GRADE‘
system was used to score strength and quality of evidence. The
objective of this guideline is to provide healthcare professionals
with clear guidance on the investigation and management of
primary myelofibrosis, as well as post-polycythaemic myelofi-
brosis (post-PV MF) and post-thrombocythemic myelofibrosis
(post-ET MF) in both adult and paediatric patients.
Keywords: myelofibrosis, myeloproliferative disorders, trans-
plantation, treatment.
Aim
The purpose of this guideline is to provide a practical, rather
than a research, approach to the diagnosis, investigation and
management of patients with primary, as well as post-poly-
cythaemic myelofibrosis (post-PV MF) and post-thrombocy-
themic myelofibrosis (post-ET MF). The criteria used to state
levels and grades of evidence are as outlined in the Procedure
for Guidelines commissioned by the BCSH; the ‘GRADE’
system was used to score strength and quality of evidence
(Table I).
Clinical features
Myeloproliferative myelofibrosis can present as a de novo dis-
order (PMF) or evolve secondarily from previous polycytha-
emia vera or essential thrombocythaemia (Post-PV MF or
Post-ET MF respectively); the term myeloproliferative neo-
plasm (MPN)-associated myelofibrosis has been suggested to
encompass all of these entities. Regardless of whether myelo-
fibrosis is primary or secondary, the disease is characterized
by a clonal haemopoietic stem cell proliferation associated
with a characteristic stromal pattern, a leuco-erythroblastic
blood film and elevated levels of various inflammatory and
pro-angiogenic cytokines.
The clinical features of myelofibrosis are variable and
include progressive anaemia, leucopenia or leucocytosis,
thrombocytopenia or thrombocytosis and multi-organ extra-
medullary haemopoiesis, most commonly causing hepatome-
galy and symptomatic splenomegaly. Patients with advanced
disease experience severe constitutional symptoms, the conse-
quences of massive splenomegaly (pain, early satiety, splenic
infarction, portal hypertension and dyspnoea), progressive
marrow failure, pulmonary hypertension, transformation to
leukaemia and early death.
Diagnosis
The diagnosis of PMF, as defined by the World Health
Organization (WHO; Thiele et al, 2008), is based on a
Correspondence: Professor J T Reilly, Department of Haematology,
H Floor, Royal Hallamshire Hospital, Glossop Road, Sheffield S10
2JF, UK. E-mail: j.t.reilly@sheffield.ac.uk
ª 2012 Blackwell Publishing Ltd, British Journal of Haematology doi:10.1111/j.1365-2141.2012.09179.x
guideline
combination of clinical, morphological, cytogenetic and
molecular features. Furthermore, the diagnoses of Post-PV
MF and Post-ET MF have recently been clarified by the
International Working Group for Myelofibrosis Research and
Treatment (IWG-MRT) with the criteria being adopted by
the WHO (Barosi et al, 2008). However, the robustness and
utility of these criteria have been questioned. For example,
key diagnostic difficulties may arise in differentiating a
myelodysplastic syndrome (MDS) with fibrosis from PMF
and between ET and some early forms of PMF (Wilkins
et al, 2008; Beer et al, 2010, 2011). Furthermore, a raised lac-
tate dehydrogenase level has recently been shown to lack
specificity for primary myelofibrosis (Beer et al, 2010). In
view of these limitations, it is recommended that the diag-
nostic criteria proposed by Campbell and Green (2006) for
PMF (Table II), as well as for Post-PV MF and Post-ET MF
(Table III) be adopted.
Molecular investigations
JAK2
Mutational screening should be carried out routinely in
patients with PMF, as the JAK2 V617F mutation is present in
approximately 45–68% of cases (Tefferi et al, 2005, 2008;
Campbell et al, 2006; Barosi et al, 2007; Guglielmelli et al,
2009). While a high JAK2 V617F allele burden, is associated
with a higher transformation rate to MF in both PV and ET
(Vannucchi et al, 2007; Passamonti et al, 2010a), quantitative
assays are currently of no value in determining therapy for
MPN patients.
BCR-ABL1
Presence of BCR-ABL1 rearrangement (diagnostic of chronic
myeloid leukaemia) excludes PMF and testing should be per-
formed if atypical features are present on the trephine
biopsy, or if the patient lacks a mutation in JAK2 or MPL.
Other mutations
MPL W515L mutations were first described in 4/45 (9%)
cases of JAK2 V617F mutation-negative PMF (Pikman et al,
Table II. Diagnostic criteria for primary myelofibrosis: diagnosis
requires A1 + A2 and any two B criteria.
A1 Bone marrow fibrosis � 3 (on 0–4 scale).
A2 Pathogenetic mutation (e.g. in JAK2 or MPL),
or absence of both BCR-ABL1 and reactive
causes of bone marrow fibrosis
B1 Palpable splenomegaly
B2 Unexplained anaemia
B3 Leuco-erthroblastosis
B4 Tear-drop red cells
B5 Constitutional symptoms*
B6 Histological evidence of extramedullary
haematopoiesis
*Drenching night sweats, weight loss >10% over 6 months, unex-
plained fever (>37·5°C) or diffuse bone pains.
Table III. Diagnostic criteria for post-PV and post-ET myelofibrosis:
diagnosis requires A1 + A2 and any two B criteria.
A1 Bone marrow fibrosis � 3 (on 0–4 scale)
A2 Previous diagnosis of ET or PV
B1 New palpable splenomegaly or increase in spleen size of � 5 cm
B2 Unexplained anaemia with 20 g/l decrease from baseline
haemoglobin
B3 Leuco-erythroblastic blood film.
B4 Tear-drop red cells
B5 Constitutional symptoms*
B6 Histological evidence of extramedullary haematopoiesis.
*Drenching night sweats, weight loss >10% over 6 months, unex-
plained fever (>37·5°C) or diffuse bone pains.
Table I. Evidence statements and grades of recommendations.
Strength of recommendations
Strong (grade 1): Strong recommendations (grade 1) are made when there is confidence that the benefits do or do not outweigh harm and
burden. Grade 1 recommendations can be applied uniformly to most patients. Regard as ‘recommend’.
Weak (grade 2): Where the magnitude of benefit or not is less certain a weaker grade 2 recommendation is made. Grade 2 recommendations
require judicious application to individual patients. Regard as ‘suggest’.
Quality of evidence
The quality of evidence is graded as high (A), moderate (B) or low (C). To put this in context it is useful to consider the uncertainty of
knowledge and whether further research could change what we know or our certainty.
(A) High: Further research is very unlikely to change confidence in the estimate of effect. Current evidence derived from randomized clinical
trials without important limitations.
(B) Moderate: Further research may well have an important impact on confidence in the estimate of effect and may change the estimate. Current
evidence derived from randomized clinical trials with important limitations (e.g. inconsistent results, imprecision – wide confidence intervals or
methodological flaws – e.g. lack of blinding, large losses to follow up, failure to adhere to intention to treat analysis), or very strong evidence
from observational studies or case series (e.g. large or very large and consistent estimates of the magnitude of a treatment effect or
demonstration of a dose-response gradient).
(C) Low: Further research is likely to have an important impact on confidence in the estimate of effect and is likely to change the estimate.
Current evidence from observational studies, case series or just opinion.
Guideline
2 ª 2012 Blackwell Publishing Ltd, British Journal of Haematology
2006), an incidence confirmed by other studies (Pardanani
et al, 2006; Guglielmelli et al, 2007). MPL mutation-positive
patients were older, more frequently female and presented
with more severe anaemia (Guglielmelli et al, 2007). Testing
for MPL exon 10 mutations can be considered in cases that
are negative for JAK2 V617F. Mutations in TET oncogene
family member 2 (TET2) occur in approximately 15% of
cases of PMF and are associated with older age and anaemia
but there is no correlation with overall survival (OS) or risk
of leukaemic transformation (Tefferi et al, 2009a) and TET2
testing is not recommended on a routine basis. The clinical
significance of mutations in other genes, including IDH1,
IDH2, ASXL1, SH2B3, IKZF1, CBL and NRAS, remain
unclear. EZH2 mutations are seen in about 5% of cases and
have been associated with a poor prognosis (Guglielmelli
et al, 2011), but routine screening of this large gene is not
warranted at present.
PDGFRA and PDGFRB rearrangements, however, should
be excluded in the presence of significant eosinophilia, as
PDGFRA/B-rearranged MPNs are highly sensitive to imatinib
therapy.
Recommendations
• JAK2 V617F mutation screening should be carried out
routinely in patients with PMF. Quantitative results are
not required for clinical management.
• BCR-ABL1 rearrangemzent should be excluded in cases
with atypical trephine biopsy features, or if the patient
lacks a mutation in JAK2 or MPL.
• PDGFRA and PDGFRB rearrangements should be
excluded in the presence of significant eosinophilia
(Screening for other mutations remains a research tool
and routine screening cannot be justified, apart from in
cases of diagnostic difficulty where detection of a clonal
abnormality would be informative)
(Evidence level 2, Grade B).
Prognosis
Therapeutic decision-making in PMF, especially determining
the need for allogeneic stem cell transplantation (allo-SCT)
requires an accurate assessment of an individual patient’s
prognosis. Until recently, the most widely used prognostic
system has been the so-called Lille Score (Dupriez et al,
1996). To address deficiencies with the latter, Cervantes et al
(2009) published the International Prognostic Scoring System
(IPSS), based on the analysis of 1054 patients, which esti-
mates survival from the time of diagnosis, based on five risk
factors: age >65 years, haemoglobin concentration <100 g/l,
leucocyte count >25 9 109/l, circulating blasts � 1% and the
presence of constitutional symptoms. Based on the presence
of 0 (low risk), 1 (intermediate risk-1), 2 (intermediate risk-
2) or 3 or more (high risk) of these variables, four risk
groups were identified, with no overlap in survival curves,
with median survivals of 135, 95, 48 and 27 months, respec-
tively (Table IV).
Using the same five prognostic variables, Passamonti et al
(2010b) subsequently modified the IPSS for use at any time
during the disease course, producing the so-called Dynamic
IPSS (DIPSS; Table IV). A further refinement, DIPSS Plus,
(Gangat et al, 2011) shows that the addition of three addi-
tional independent risk factors, transfusion dependence,
Table IV. Prognostic criteria.
Variable IPSS DIPSS
Age > 65 years ✓ ✓
Constitutional symptoms ✓ ✓
Haemoglobin <100 g/l ✓ ✓
Leucocyte count > 25 9 109/l ✓ ✓
Circulating blasts � 1% ✓ ✓
1 point each 1 point each but Hb = 2
DIPSS-Plus: add 1 point to the DIPSS RISK GROUP* (low = 0; intermediate 1 = 1, intermediate 2 = 2 and high risk = 3) in addition for:
Platelet count <100 9 109/l
RBC transfusion need
Unfavourable karyotype +8, �7/7q�,i(17q),inv(3), �5/5q�, 12p�, 11q23 rearrangement
Risk group
IPSS DIPSS DIPSS-Plus
Predictors (n) Median survival (years) Predictors (n) Median survival (years) Predictors (n) Median survival (years)
Low 0 11·3 0 Not reached 0 15·4
Intermediate-1 1 7·9 1 or 2 14·2 1 6·5
Intermediate-2 2 4·0 3 or 4 4 2–3 2·9
High � 3 2·3 5 or 6 1·5 � 4 1·3
*Note that this is the risk group NOT the sum of points.
Guideline
ª 2012 Blackwell Publishing Ltd, British Journal of Haematology 3
unfavourable karyotype (including +8, �7/7q�, i(17q), inv
(3), �5/5q�, 12p�. 11q23 rearrangements and complex
karyotypes) and platelet count <100 9 109/l, gives the four
prognostic groups an even greater discrimination, with corre-
sponding median survival estimates of 185, 78, 35 and
16 months (Table IV). Although a number of molecular
findings have been shown to adversely affect prognosis,
including low JAK2 V617F allele burden (Guglielmelli et al,
2009) and EZH2 mutational status (Guglielmelli et al, 2011),
such parameters have not yet been incorporated into practi-
cal prognostic scoring systems, although the findings suggest
that future improvements are likely.
Recommendation
• Therapeutic decisions in PMF, especially regarding the
use of allo-SCT, should be based on the patient prognosis
as determined by the DIPSS Plus as this is validated for
any timepoint of the disease and is more discriminating
in median survival prediction than the IPSS score.
• Whilst the IPSS, DIPSS and DIPSS Plus have not been
validated for post-PV MF and post-ET MF, it is suggested
that they still be used in this setting (Evidence level 2,
Grade B).
Treatment
Splenomegaly and extramedullary haemopoiesis
Medical treatment remains the treatment of choice for most
patients with symptomatic splenomegaly. However, no cur-
rent therapies deliver robust sustained responses, particularly
for patients with massive splenomegaly:
Hydroxycarbamide. This is the most widely used agent,
despite limited published data supporting its efficacy. An
early study suggested a response rate of approximately 45%
(Lo¨fvenberg & Wahlin, 1988), although the degree of splenic
reduction was not detailed. Similar findings were reported in
a recent study of 18 MPN patients with symptomatic spleno-
megaly (Martinez-Trillos et al, 2010). Overall, complete
responses are rare and doses of more than 1·5 g/d may be
required to achieve clinical effect. Benefit is usually seen
within 8–10 weeks of treatment, although side effects,
especially significant cytopenias, may be problematic at effec-
tive doses.
An Italian study investigated the efficacy of low-dose mel-
phalan (2·5 mg/thrice weekly) in PMF and documented a
similar response rate to that shown in the hydroxycarbamide
studies highlighted above, although with the added informa-
tion that normalization of spleen size was achieved in only
4·5% in patients with massive splenomegaly (� 15 cm; Petti
et al, 2002). Furthermore, there was no survival benefit in
those that responded. Bulsulphan may also produce clinical
benefit (Manoharan & Pitney, 1984; Chang & Gross, 1988),
but myelosuppression and an increased risk of acute leukae-
mia are potential adverse factors.
Immunomodulatory agents. Immunomodulatory drugs have
been evaluated in a number of small studies. Low-dose tha-
lidomide (50 mg/d), for example, combined with a tapering
dose of prednisolone resulted in an overall response rate of
33% (Mesa et al, 2003), although subsequent follow-up data
showed that only 8% of patients obtained a clinical improve-
ment in splenomegaly (Thapaliya et al, 2011), as defined by
the more stringent IWG-MRT criteria (Tefferi et al, 2006a).
Responses were also reported for anaemia (22%) and throm-
bocytopenia (50%). Lenalidomide has also been shown to
produce a response rate of 33% in a study that included
some patients who had failed on prior thalidomide therapy
(Tefferi et al, 2006b).
Interferon-alpha. Both standard and pegylated preparations
of interferon-alpha appear to have little clinical effect in
reducing splenomegaly and, as a result, their use is not rec-
ommended (Tefferi et al, 2001a; Jabbour et al, 2007; Ianotto
et al, 2009). They do, however, have a role as myelosuppres-
sive agents (see ‘Myelosuppression Therapy’ section).
Cladribine. This purine analogue, previously known as 2-
chlorodeoxyadenosine, has been shown to reduce clinically
significant post-splenectomy hepatomegaly and post-splenec-
tomy thrombocytosis in 56% and 50% of patients respec-
tively, although myelosuppression is a significant side-effect.
The drug, which can be administered for up to 4 monthly
cycles, frequently resulted in durable responses, which lasted
for a median of 6 months after discontinuation of therapy
(Faoro et al, 2005).
JAK inhibitors. JAK inhibitors may have a future role in the
management of splenomegaly (see ‘Novel Therapies’ section)
and are the only therapies to have been evaluated in the con-
text of randomized clinical trials.
Recommendations: medical management of splenomegaly
First Line:
• Hydroxycarbamide (in the absence of cytopenias).
• Thalidomide and prednisolone (in presence of cytopenias)
– consider lenalidomide (if anaemic with platelet count
>100 9 109/l).
Second Line:
• Consideration should be given to the use of JAK inhibitors
either as part of a clinical trial, or via patient access proto-
cols. These agents are now approved in the USA for first
line therapy which is appropriate following approval
(Evidence level 1, Grade A).
Guideline
4 ª 2012 Blackwell Publishing Ltd, British Journal of Haematology
Surgical management
The place of splenectomy in the management of myelofibro-
sis is well established (Barosi et al, 1993; Tefferi et al, 2000).
Routine splenectomy is inappropriate and the procedure
should be restricted to carefully selected patients with refrac-
tory haemolysis, symptomatic splenomegaly, significant sple-
nic infarction, severe portal hypertension and severe
hypercatabolic symptoms. Once a patient is considered a
candidate for splenectomy, an extensive pre-operative evalua-
tion is required to determine if the cardiac, hepatic, renal,
metabolic and haemostatic risks are acceptable. Even in the
best units, splenectomy is associated with morbidity and
mortality rates of approximately 31% and 9%, respectively
(Tefferi et al, 2000). Laparoscopic splenectomy is not advised
in PMF on account of bleeding complications (Feldman
et al, 2008). Splenic artery embolization is not without risk
and there is no evidence to support its use.
Patients with portal hypertension and bleeding varices
should have dynamic circulatory studies performed during
surgery, because portal hypertension due to splenomegaly is
ameliorated by splenecto