World Health Organization-defined eosinophilic disorders: 2012 update on diagnosis, risk stratification, and management


  • Conflict of interest: Nothing to report


Disease overview: The eosinophilias encompass a broad range of nonhematologic (secondary or reactive) and hematologic (primary, clonal) disorders with potential for end-organ damage.

Diagnosis: Hypereosinophilia has generally been defined as a peripheral blood eosinophil count greater than 1,500/mm3 and may be associated with tissue damage. After exclusion of secondary causes of eosinophilia, diagnostic evaluation of primary eosinophilias relies on a combination of morphologic review of the blood and marrow, standard cytogenetics, fluorescent in situ-hybridization, flow immunocytometry, and T-cell clonality assessment to detect histopathologic or clonal evidence for an acute or chronic myeloid or lymphoproliferative disorder.

Risk stratification: Disease prognosis relies on identifying the subtype of eosinophilia. After evaluation of secondary causes of eosinophilia, the 2008 World Health Organization establishes a semimolecular classification scheme of disease subtypes including “myeloid and lymphoid neoplasms with eosinophilia and abnormalities of PDGFRA, PDGFRB, or FGFR1,” chronic eosinophilic leukemia, not otherwise specified' (CEL, NOS), lymphocyte-variant hypereosinophilia, and idiopathic hypereosinophilic syndrome (HES), which is a diagnosis of exclusion.

Risk-adapted therapy: The goal of therapy is to mitigate eosinophil-mediated organ damage. For patients with milder forms of eosinophilia (e.g., <1,500/mm3) without symptoms or signs of organ involvement, a watch and wait approach with close-follow-up may be undertaken. Identification of rearranged PDGFRA or PDGFRB is critical because of the exquisite responsiveness of these diseases to imatinib. Corticosteroids are first-line therapy for patients with lymphocyte-variant hypereosinophilia and HES. Hydroxyurea and interferon-alpha have demonstrated efficacy as initial treatment and steroid-refractory cases of HES. In addition to hydroxyurea, second line cytotoxic chemotherapy agents and hematopoietic cell transplant have been used for aggressive forms of HES and CEL with outcomes reported for limited numbers of patients. Although clinical trials have been performed with anti IL-5 (mepolizumab) and anti-CD52 (alemtuzumab) antibodies, their therapeutic role in primary eosinophilic diseases and HES has yet to be established. Am. J. Hematol. 2012. © 2012 Wiley Periodicals, Inc.

Disease Overview


The incidence and prevalence of HES is not well characterized. Using the International Classification of Disease for Oncology (Version 3), and coding of 9964/3 (HES including chronic eosinophilic leukemia), the Surveillance, Epidemiology and End Results (SEER) database from 2001 to 2005 revealed that the age-adjusted incidence rate was approximately 0.036 per 100,000 [1]. The incidence of eosinophilias with recurrent genetic abnormalities (PDGFRA/B, FGFR1) comprises a minority of these patients. The median frequency of the FIP1L1-PDGFRA fusion in patients with hypereosinophilia across eight published series enrolling more than 10 patients was 23% (range 3-56%) [2]. Larger studies conducted in developing countries indicate that the FIP1L1-PDGFRA fusion occurs in approximately 10-20% of patients with idiopathic hypereosinophilia [3–5]. Although usually diagnosed between the ages of 20 and 50, idiopathic hypereosinophilia or CEL may arise at the extremes of age, with infrequent cases being described in infants and children [6–8]. In the SEER database of 131 incident cases between 2001 and 2005, the male-to-female ratio was 1.47, and rates increased with age to a peak between 65-74 years [1]. For reasons that are unknown, the overwhelming majority of patients with FIP1L1-PDGFRA or myeloproliferative variants of HES are male [3, 9, 10], whereas other eosinophilia subtypes exhibit no clear gender bias.

Definition of eosinophilia and classification

The upper limit of normal for the range of % eosinophils in the peripheral blood is 3–5% with a corresponding absolute eosinophil count (AEC) of 350–500/mm3 [11, 12]. The severity of eosinophilia has been arbitrarily divided into mild (AEC from the upper limit of normal to 1,500/mm3), moderate (AEC 1,500–5,000/mm3) and severe (AEC >5,000/mm3) [11–13].

The classification of eosinophilic diseases was revised in the 2008 World Health Organization scheme of myeloid neoplasms (Table I). In recognition of the growing list of recurrent, molecularly defined primary eosinophilias, a new major category was created, “Myeloid and lymphoid neoplasms with eosinophilia and abnormalities of platelet-derived growth factor receptor alpha (PDGFRA), platelet-derived growth factor receptor beta (PDGFRB), or fibroblast growth factor receptor 1 (FGFR1)” (Table I) [14]. Within the major WHO category of myeloproliferative neoplasms (MPNs), “chronic eosinophilic leukemia-not otherwise specified” (CEL-NOS) is one of eight disease entities within this group (Table I) [15]. CEL-NOS is operationally defined by absence of the Philadelphia chromosome or a rearrangement involving PDGFRA/B and FGFR1, and the exclusion of other acute or chronic primary marrow neoplasms associated with eosinophilia such as acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), systemic mastocytosis (SM), the classic MPNs (chronic myeloid leukemia, polycythemia vera, essential thrombocythemia, and primary myelofibrosis), and MDS/MPN overlap disorders (e.g., chronic myelomonocytic leukemia, CMML) (Table II). CEL-NOS is histologically characterized by an increase in blasts in the bone marrow or blood (but fewer than 20% to exclude acute leukemia as a diagnosis), and/or there is evidence for clonality in the eosinophil lineage [15]. A diagnosis of idiopathic HES requires exclusion of all primary and secondary causes of hyerpeosinophilia as well as lymphocyte-variant hypereosinophilia (Table II). The modern definition of HES remains a vestige of the historical criteria outlined by Chusid et al. in 1975: the absolute eosinophil count is > 1,500/mm3 for more than 6 months, and tissue damage is present [16]. The requirement that eosinophilia persist for more than 6 months is less consistently embraced today because of the availability of more sophisticated tools to rapidly evaluate eosinophilia and the need for some patients to receive expedited treatment to minimize organ damage. In contrast to “HES,” “idiopathic hypereosinophilia” is the preferred term when end-organ damage is absent [15]. The pool of classically defined idiopathic HES patients has diminished due to an increasing proportion of cases which have been reassigned as clonal marrow disorders. HES may therefore be considered a provisional diagnosis until a primary or secondary cause of eosinophilia is recognized.

Table I. 2008 World Health Organization (WHO) Classification of Myeloid Malignancies
1. Acute myeloid leukemia and related disorders
2. Myeloproliferative neoplasms (MPN)
 Chronic myelogenous leukemia, BCR-ABL1 positive
 Chronic neutrophilic leukemia
 Polycythemia vera
 Primary myelofibrosis
 Essential thrombocythemia
 Chronic eosinophilic leukemia, not otherwise specified
 Myeloproliferative neoplasms, unclassifiable
3. Myelodysplastic syndromes (MDS)
 Refractory cytopenia with uni-lineage dysplasia
  Refractory anemia
  Refractory neutropenia
  Refractory thrombocytopenia
 Refractory anemia with ring sideroblasts
 Refractory cytopenia with multilineage dysplasia
 Refractory anemia with excess blasts (RAEB)
 Myelodysplastic syndrome with isolated del(5q)
 Myelodysplastic syndrome, unclassifiable
 Chronic myelomonocytic leukemia
 Atypical chronic myeloid leukemia, BCR-ABL1 negative
 Juvenile myelomonocytic leukemia
 MDS/MPN, unclassifiable
  Refractory anemia with ring sideroblasts and thrombocytosis (RARS-T)
5. Myeloid and lymphoid neoplasms associated with eosinophilia and abnormalities of PDGFRA, PDGFRB, or FGFR1
 Myeloid and lymphoid neoplasms associated with PDGFRA rearrangement
 Myeloid neoplasms associated with PDGFRB rearrangement
 Myeloid and lymphoid neoplasms associated with FGFR1 abnormalities
Table II. 2008 World Health Organization Classification of Eosinophilic Disorders
  • a

    Patients presenting with acute myeloid leukemia or lymphoblastic leukemia/lymphoma with eosinophilia and a FIP1L1-PDGFRA fusion gene are also assigned to this category.

  • b

    If appropriate molecular analysis is not available, this diagnosis should be suspected if there is a Ph-negative MPN with the hematological features of chronic eosinophilic leukemia associated with splenomegaly, a marked elevation of serum vitamin B12, elevation of serum tryptase and increased bone marrow mast cells.

  • c

    Because t(5;12)(q31∼q33;p12) does not always lead to an ETV6-PDGFRB fusion gene, molecular confirmation is highly desirable. If molecular analysis is not available, this diagnosis should be suspected if there is a Ph-negative MPN associated with eosinophilia and with a translocation with a 5q31–33 breakpoint.

Myeloid and lymphoid neoplasms with eosinophilia and abnormalities of PDGFRA, PDGFRB, or FGFR1
 Diagnostic criteria of an MPNa with eosinophilia associated with FIP1L1-PDGFRA
  A myeloproliferative neoplasm with prominent eosinophilia
  Presence of a FIP1L1-PDGFRA fusion geneb
 Diagnostic criteria of MPN associated with ETV6-PDGFRB fusion gene or other rearrangement of PDGFRB
  A myeloproliferative neoplasm, often with prominent eosinophilia and sometimes with neutrophilia or monocytosis
  Presence of t(5;12)(q31∼q33;p12) or a variant translocationc or, demonstration of an ETV6-PDGFRB fusion gene or rearrangement of PDGFRB
 Diagnostic criteria of MPN or acute leukemia associated with FGFR1 rearrangement
  A myeloproliferative neoplasm with prominent eosinophilia and sometimes with neutrophilia or monocytosis
   Acute myeloid leukemia or precursor T-cell or precursor B-cell lymphoblastic leukemia/lymphoma (usually associated with peripheral blood or bone marrow eosinophilia)
  Presence of t(8;13)(p11;q12) or a variant translocation leading to FGFR1 rearrangement demonstrated in myeloid cells, lymphoblasts, or both
Chronic Eosinophilic Leukemia, not otherwise specified (NOS)
 1. There is eosinophilia (eosinophil count >1.5x109/L)
 2. There is no Ph chromosome or BCR-ABL fusion gene or other myeloproliferative neoplasms (PV, ET, PMF, systemic mastocytosis) or MDS/MPN (CMML or atypical CML)
 3. There is no t(5;12)(q31∼q35;p13) or other rearrangement of PDGFRB
 4. There is no FIP1L1-PDGFRA fusion gene or other rearrangement of PDGFRA
 5. There is no rearrangement of FGFR1
 6. The blast cell count in the peripheral blood and bone marrow is less than 20% and there is no inv(16)(p13q22) or t(16;16)(p13;q22) or other feature diagnostic of AML
 7. There is a clonal cytogenetic or molecular genetic abnormality, or blast cells are more than 2% in the peripheral blood or more than 5% in the bone marrow.
Idiopathic Hypereosinophilic Syndrome (HES)
Exclusion of the following:
 1. Reactive eosinophilia
 2. Lymphocyte-variant hypereosinophilia (cytokine-producing, immunophenotypically-aberrant T-cell population)
 3. Chronic eosinophilic leukemia, NOS
 4. WHO-defined myeloid malignancies associated eosinophilia (e.g. MDS, MPNs, MDS/MPNs, or AML)
 5. Eosinophilia-associated MPNs or AML/ALL with rearrangements of PDGFRA, PDGFRB, or FGR1.
 6. The absolute eosinophil count of >1,500/mm3 must persist for at least 6 months and tissue damage must be present. If there is no tissue damage, idiopathic hyerpeosinophilia is the preferred diagnosis.

In 2011, the Working Conference on Eosinophil Disorders and Syndromes proposed a new terminology for eosinophilic syndromes [17]. The panel recommended the higher level term “Hypereosinophilia (HE)” for persistent and marked eosinophilia (AEC > 1,500/mm3). In turn, HE subtypes were divided into a hereditary (familial) variant (HEFA), HE of undetermined significance (HEUS), primary (clonal/neoplastic) HE produced by clonal/neoplastic eosinophils (HEN), and secondary (reactive) HE (HER). HEUS was introduced as a novel term in lieu of “idiopathic hypereosinophilia.” Any HE (not just idiopathic) associated with organ damage is referred to as “HES” with specific variants designated by subscripts (e.g., HESUS, HESN, and HESR). Additional recommendations advanced by the consensus panel are summarized in their report.

Clinical presentation and diagnosis

The varied clinical presentations of primary eosinophilias/HES reflect their heterogeneous pathophysiology. In two retrospective series published in 1982 and 2009, eosinophilia was an incidental finding in 12 and 6% of patients, respectively [18, 19]. The most common presenting signs and symptoms were weakness and fatigue (26%), cough (24%), dyspnea (16%), myalgias or angioedema (14%), rash or fever (12%), and rhinitis (10%) [17]. In HES, leukocytosis (e.g., 20,000–30,000/mm3 or higher) with peripheral eosinophilia in the range of 30–70% is a common finding [16, 18–20]; the aforementioned retrospective analysis of 188 patients from 2009 observed a mean peak eosinophil count of 6,600/mm3 with a range of 1,500–400,000/mm3 [18]. Other hematologic findings include peripheral blood or bone marrow neutrophilia, basophilia, myeloid immaturity, and both mature and immature eosinophils with varying degrees of dysplasia [20–22]. In one series, anemia was present in 53% of patients, thrombocytopenia was more common than thrombocytosis (31% vs. 16%), and bone marrow eosinophilia ranged from 7 to 57% (mean 33%) [22]. Marrow findings of Charcot-Leyden crystals, and sometimes increased blasts and marrow fibrosis, are also observed [22].

Essentially all organ systems may be susceptible to the effects of sustained eosinophilia [reviewed in Ref.23]. During follow-up of patients with hypereosinophilia, dermatologic involvement was also the most common clinical manifestation reported in 69% of patients, followed by pulmonary (44%), and gastrointestinal (38 %) manifestations. Cardiac disease unrelated to hypertension, atherosclerosis, or rheumatic disease was eventually identified in 20% of patients (only 6% at the time of initial presentation) [19]. Progressive heart failure is a proto-typical example of eosinophil-mediated organ injury. It involves a multistep pathophysiological process involving eosinophil infiltration of cardiac tissue and release of toxic mediators from eosinophils [reviewed in Refs.18 and24]. Endocardial damage with resulting platelet thrombus can lead to mural thrombi and increased embolic risk. In the later fibrotic stage, fibrous thickening of the endocardial lining can evolve to a restrictive cardiomyopathy [18, 24]. Valvular insufficiency results from mural endocardial thrombosis and fibrosis involving leaflets of the mitral or tricuspid valves [25–27].


Step 1: Exclude secondary (reactive) causes of eosinophilia

Secondary eosinophilia has numerous causes which may require diagnostic evaluation by a cadré of different sub-specialty consultants. In developing countries, eosinophilia most commonly derives from infections, particularly tissue-invasive parasites [13]. Allergy/atopy and hypersensitivity conditions, drug reaction, collagen-vascular disease (e.g., Churg-Strauss Syndrome, granulomatosis with polyangiitis [Wegener's], systemic lupus erythematosus), pulmonary eosinophilic diseases (e.g., idiopathic acute or chronic eosinophilia pneumonia, tropical pulmonary eosinophilia, allergic bronchopulmonary aspergillosis, etc.), allergic gastroenteritis (with associated peripheral eosinophilia), and metabolic conditions such as adrenal insufficiency are diagnostic considerations in the appropriate clinical context [28–30]. Nonmyeloid malignancies may be associated with secondary eosinophilia which results from the production of cytokines such as IL-3, IL-5, and GM-CSF which promote eosinophil differentiation and survival. For example, these cytokines may be elaborated from malignant cells in T-cell lymphomas [31], Hodgkin's disease [32], and acute lymphoblastic leukemias [33]. Rare conditions associated with eosinophilia include familial eosinophilia whose genetic basis remains unknown, hyper IgE Syndrome, Omenn Syndrome, episodic angioedema and eosinophilia (Gleich's syndrome), and eosinophilia-myalgia syndrome (e.g., possibly related to tryptophan ingestion, or of historical interest, the epidemic of toxic-oil syndrome) [17]. Repeated ova and parasite testing, stool culture, and antibody testing for specific parasites (e.g., strongyloides) is paramount for identifying infectious etiologies in the appropriate clinical context. Additional laboratory and imaging tests (e.g., chest-x-ray, electrocardiogram, and echocardiography, CT scan of the chest, abdomen/pelvis) are guided by the patient's travel history, presenting symptoms, and findings on physical examination. For eosinophilic lung diseases, pulmonary function testing, bronchoscopy, serologic tests (e.g., aspergillus IgE to evaluate for allergic bronchopulmonary aspergillosis [ABPA]) may be obtained to further characterize lung involvement.

Step 2: Evaluate for primary (clonal) eosinophilia

If secondary causes of eosinophilia are excluded, the work-up should proceed to the evaluation of a primary bone marrow disorder. Examination of the blood smear and blood tests (e.g., circulating blasts, dysplastic cells, monocytosis, elevated serum B12, or tryptase level) in conjunction with bone marrow morphologic, cytogenetic, and immunophenoytpic analysis will help ascertain whether the differential diagnosis of eosinophilia includes a well-defined WHO myeloid neoplasm such as systemic mastocytosis, chronic myelogenous leukemia, acute myelogenous leukemia (especially the historically defined M2 and M4 Eo French-American-British subtypes), myelodysplastic syndrome (MDS), or MDS/MPN overlap disorder (e.g., CMML). Although not formally included in the WHO monograph, the term “myeloproliferative variant of hypereosinophilia” has been used to refer to some of these marrow-derived eosinophilic myeloid malignancies because of clinicopathologic similarity to CML and the BCR-ABL-negative MPNs [9, 23].

Laboratory evaluation of primary eosinophilia should begin with screening of the peripheral blood for the FIP1L1-PDGFRA gene fusion (by RT-PCR or interphase/metaphase FISH) (Fig. 1). FISH probes that hybridize to the region between the FIP1L1 and PDGFRA genes are used to detect the presence of the cytogenetically occult 800-kb deletion on 4q12 that results in FIP1L1-PDGFRA [9, 34]. Since the CHIC2 gene is located in this deleted genetic segment, this widely available clinical test is referred to as “FISH for the CHIC2 deletion” [34]. In instances where FIP1L1-PDGFRA screening is not available, evaluation of the serum tryptase can be a useful surrogate marker for FIP1L1-PDGFRA-positive disease since increased levels segregate with this molecular abnormality and myeloproliferative variants of hypereosinophilia [35]. FIP1L1-PDGFRA has also been also identified histopathologically defined cases of systemic mastocytosis with associated eosinophilia [4]. The bone marrows of such patients typically exhibit less dense clusters of mast cells by tryptase immunostaining than are observed in SM with the highly prevalent KIT D816V mutation [4]. The FIP1L1-PDGFRA fusion has also been found in cases of AML and T-cell lymphoblastic lymphoma associated with eosinophilia [36]. In addition to dysregulation of PDGFRA by fusion to FIP1L1 or other partner genes, activating point mutations have been identified in PDGFRA in patients with hypereosinophilia [37]. Although there was variability in their transforming ability, injection of cells harboring these mutants into mice induced a leukemia-like disease. Imatinib treatment significantly decreased leukemic growth and prolonged survival [37].

Figure 1.

Diagnostic and treatment algorithm based on 2008 WHO classification of eosinophilic disorders.

Absence of the FIP1L1-PDGFRA fusion should prompt evaluation for other primary eosinophilias associated with recurrent molecular abnormalities. Molecular evidence for a PDGFRA, PDGFRB, or FGFR1 fusion gene is often accompanied by its abnormal karyotype equivalent: rearrangement of 4q12 (PDGFRA fusion partners besides FIP1L1), 5q31-33 (PDGFRB) or 8p11-13 (FGFR1) [14]. Despite the rare frequency (<1%) of PDGFRB-rearrangements in cytogenetically-defined cases of CMML and other myeloid neoplasms (e.g. atypical CML, juvenile myelomonocytic leukemia, MDS/MPN overlap disorders), their identification is critical given their responsiveness to imatinib [38]. Over 20 gene fusion partners of PDGFRB have been described. Eosinophilic myeloid neoplasms related to fusions involving the FGFR1 gene are similarly rare [14]. In these cases, the association of t(8p11-13) breakpoint with lymphoblastic lymphoma with eosinophilia and myeloid hyperplasia was first described in 1995, and was previously referred to as “8p11 myeloproliferative syndrome” or “stem cell leukemia/lymphoma.” Following the discovery of the ZNF198-FGFR1 fusion gene in 1998 by several groups [39–42], more than 10 fusion partners of FGFR1 have been reported [reviewed in Ref.2].

A negative screen for PDGFRA/B- or FGFR1-rearranged eosinophilias should lend consideration to a diagnosis of CEL-NOS when there is cytogenetic and/or morphologic evidence of a eosinophilic myeloid malignancy that is otherwise not classifiable [15]. CEL- NOS may be distinguished from HES by the presence of a nonspecific clonal cytogenetic abnormality or increased blast cells (>2% in the peripheral blood or >5% in the bone marrow, but <20% blasts in both compartments). Lymphocyte-variant hypereosinophilia is a more obscure diagnostic entity characterized by an abnormal T-cell population (demonstrated by peripheral blood lymphocyte immunophenotyping or T cell receptor gene rearrangement studies) which may be associated with excessive eosinophilopoietic cytokine production in vitro (e.g., serum interleukin-5) [15, 43, 44]. If none of the aforementioned conditions are identified, a diagnosis HES is made if organ damage is present.

Lymphocyte-variant hypereosinophilia

Some patients may exhibit expansion of a cytokine-producing, immunophenotypically aberrant T-cell population [15, 44]. The condition is a mixture of clonal and reactive processes: it is clonal with regard to the production of abnormal T-cell lymphocytes; however, the eosinophilia is reactive to the eosinophilopoietic growth factors elaborated by the T-cells. These patients typically have cutaneous signs and symptoms as the primary disease manifestation. The immunophenotype of these lymphocytes include double-negative, immature T-cells (e.g., CD3+CD4CD8) or absence of CD3 (e.g., CD3CD4+), a normal component of the T-cell receptor complex [45–47]. Additional immunophenotypic abnormalities include elevated CD5 expression on CD3CD4+ cells, and loss of surface CD7 and/or expression of CD27 [44]. In patients with T-cell mediated hypereosinophilia with elevated IgE levels, lymphocyte production of IL-5, and in some cases IL-4 and IL-13, suggests that these T-cells have a helper type 2 (Th2) cytokine profile [44, 45, 47–49]. In a study of 60 patients primarily from dermatology clinics, 16 demonstrated circulating T-cells with an abnormal immunophenotype [43]. Clonal rearrangement of T-cell receptor genes was demonstrated in half of these individuals (8/60 total patients). The abnormal T-cells secreted high levels of interleukin-5 in vitro, and displayed an activated immunophenotype (e.g., CD25 and/or HLA-DR expression).

Consensus criteria for the diagnosis of lymphocyte-variant hypereosinophilia have not been established. The finding of isolated T-cell clonality by PCR without T-cell immunophenotypic abnormalities or demonstration of Th2 cytokine production is not felt to be sufficient to make a diagnosis of this eosinophilia variant [50]. Despite a recent study demonstrating that a high proportion of idiopathic HES patients exhibit a clonal T-cell receptor gene rearrangement by PCR (18/42 patients, 43%), it is unclear whether such clonal T-cell populations are always relevant to the disease process [51]. Detection of elevated serum levels of TARC, a chemokine implicated in Th2-mediated diseases, in addition to the finding of increased in vitro production of cytokines from cultured peripheral blood mononuclear cells and/or T-cells (research-based assays), may provide additional support for a diagnosis of lymphocyte-variant hypereosinophilia [19, 50, 52].

Risk Stratification

Older case series indicate that lives of patients with HES were overshadowed by early cardiac death. A review of 57 HES cases published through 1973 reported a median survival of 9 months and the 3-year survival was only 12% [16]. Patients usually presented with advanced disease, with congestive heart failure accounting for 65% of deaths at autopsy. In addition to cardiac disease, peripheral blood blasts or a WBC count greater than 100,000/mm3 were poor prognostic factors [16]. A later report of 40 HES patients cited a 5-year survival rate of 80%, decreasing to 42% at 15 years [21]. Factors predictive of a worse outcome included the presence of a concurrent myeloproliferative syndrome, corticosteroid-refractory hypereosinophilia, cardiac disease, male sex, and the height of eosinophilia [21]. These data are less relevant to the current era of molecularly defined eosinophilias where availability of targeted therapy such as imatinib, improved diagnostic testing, and better medical treatment and surgery for cardiovascular sequelae have contributed to improved survival.

The prognosis of WHO-defined CEL-NOS is poor. In a recently reported cohort of 10 patients, the median survival was 22.2 months, and 5 of the 10 patients developed acute transformation after median of 20 months from diagnosis [53]. Three of 5 patients who did not develop AML died with active disease; one patient underwent an allogeneic stem-cell transplant and maintained a long-term remission, and the remaining patient achieved a complete remission on imatinib and hydroxyurea [53].

In the lymphocyte-variant of hypereosinophilia, an indolent disease course is usually observed. However, patients may infrequently develop either T-cell lymphoma or Sézary syndrome, indicating this condition has malignant potential [43]. Accumulation of cytogenetic changes (e.g., partial 6q and 10p deletions, trisomy 7) in T-cells, and proliferation of lymphocytes with the CD3CD4+ phenotype have been observed with progression to lymphoma [49, 54–56].

In WHO-defined myeloid malignancies, the prognostic importance of associated eosinophilia has been only been studied in few diseases. In a series of 123 patients with systemic mastocytosis, eosinophilia was prevalent in 34% of cases, but was prognostically neutral and not affected by exclusion of FIP1L1-PDGFRA-positive cases [57]. In a study of 1008 patients with de novo MDS, eosinophilia (and basophilia) predicted a significantly reduced survival without having a significant impact on leukemia-free survival [58]. A retrospective of 288 individuals with newly diagnosed MDS, revealed that significantly higher numbers of patients with eosinophilia or basophilia (compared to patients with neither) had chromosomal abnormalities carrying an intermediate or poor prognosis [59]. In addition, the overall survival rate was significantly lower, and evolution to AML occurred more frequently.

Risk-Adapted Therapy

General considerations

It is difficult to predict what duration and severity of eosinophilia will precipitate tissue damage in individual patients. Inadequate data exists to support initiation of therapy based on a specific eosinophil count in the absence of organ disease, although an absolute eosinophil count of 1,500–2000/mm3 has been recommended by some as a threshold for starting treatment [60]. Treatment algorithms have incorporated serial monitoring of eosinophil counts, bone marrow aspiration and biopsy with cytogenetics, evaluation of clonality (e.g., T-cell receptor gene rearrangement, immunophenotyping), and directed organ assessment (e.g., echocardiography, pulmonary function testing) in order to identify occult organ disease and defined causes of eosinophilia which may emerge after an initial diagnosis of HES [24, 61].

Given the historically poor-prognosis of chronic eosinophilic leukemias and HES, and the exquisite sensitivity to imatinib in patients with rearranged PDGFRA/B, consensus has emerged that these individuals be treated in the absence of organ dysfunction. Proactive treatment has the potential to not only forestall tissue damage, but also to achieve complete molecular remissions.

In patients with eosinophilia-related organ damage (e.g., heart, lungs, gastrointestinal, central nervous system, skin), risk-adapted therapy rests on the premise of identifying the specific WHO-defined eosinophilic disorder and individualizing treatment accordingly. For patients with an eosinophilia-associated WHO defined myeloid malignancy (e.g., AML, MDS, systemic mastocytosis, CML, and other MPNs, MDS/MPN), therapy is dictated by disease-specific algorithms and guidelines. As a multikinase inhibitor, imatinib has not only demonstrated remarkable benefit in CML, but is now definitive first line therapy in patients with FIP1L1-PDGFRA-positive disease, and the rare patients with alternate PDGFRA fusions or rearranged PDGFRB. The discussion of treatment options below will focus on the experience with imatinib in PDGFRA/B-rearranged neoplasms and separately the therapeutic options available for patients with CEL, NOS, HES, and lymphocyte-variant hypereosinophilia, which is based primarily on small case series and retrospective studies. The use of other chemotherapeutics for HES, and recent investigational approaches with the anti-IL-5 antibody mepolizumab and anti-CD52 antibody alemtuzumab, will also be addressed.

PDGFRA/B-rearranged neoplasms: The imatinib experience

The success of imatinib in chronic myelogenous leukemia led to its empiric use in patients with hypereosinophilia who exhibited signs suggestive of a myeloproliferative disorder. Several case reports and small case series of HES patients were published in 2001–2002 highlighting rapid and complete hematologic responses to imatinib 100–400 mg daily [61–64]. FIP1L1-PDGFRα was ultimately identified as the therapeutic target of imatinib [9, 65]. The identification of the clonal marker FIP1L1-PDGFRA in these cases operationally redefined them as a form of chronic eosinophilic leukemia, and now comprise the WHO major category of myeloid and lymphoid neoplasms with eosinophilia and abnormalities of PDGFRA, PDGFRB, or FGFR1 [14]. The hematologic benefit of imatinib in FIP1L1-PDGFRA-positive myeloid neoplasms has been confirmed in numerous studies. Molecular remissions were first reported by the NIH group by PCR testing of the peripheral blood in 5 of 6 FIP1L1-PDGFRA-positive patients after 1–12 months of imatinib therapy [66]. Several reports have now described rapid induction of molecular remission in imatinib-treated FIP1L1-PDGFRA positive patients or with bone marrow transplantation. Although 100 mg daily may be sufficient to achieve a molecular remission in some patients, others may require higher maintenance doses in the range of 300–400 mg daily. Maintenance dosing of 100–200 mg weekly may be sufficient to achieve a molecular remission in some patients [67]. The optimal imatinib dose which sustains a molecular remission has not been defined.

The natural history of imatinib-treated FIP1L1-PDGFRA-positive myeloid neoplasms was evaluated in an Italian prospective cohort of 27 patients with a median follow-up period of 25 months (range 15-60 months) [10]. Patients were dose escalated from an initial dose of 100 mg daily to a final dose of 400 mg daily. Complete hematologic remission was achieved in all patients within 1 month, and all patients became PCR negative for FIP1L1-PDGFRA after a median of 3 months of treatment (range 1 to 10 months). Patients continuing imatinib remained PCR-negative during a median follow-up period of 19 months (range 6-56+ months). Another European study prospectively assessed the natural history of molecular responses to imatinib doses of 100-400 mg daily [3]. Among 11 patients with high pre-treatment transcript levels, all achieved a 3-log reduction in transcript levels by one year of therapy, and 9 of 11 patients achieved a molecular remission.

Although in-depth and durable molecular responses occur with imatinib, discontinuation of the drug can lead to relapse [3, 10]. In a dose de-escalation trial of imatinib in 5 patients who had achieved a stable hematologic and molecular remission at 300-400 mg daily for at least one year, molecular relapse was observed in all patients after 2-5 months of either dose imatinib reduction or discontinuation [68]. Molecular remissions could be re-established with re-induction of imatinib in all cases at a dose range of 100-400 mg daily. Hematologic relapse was noted only several weeks after discontinuation of imatinib in 4 patients in a Mayo series [5]. These data indicate that imatinib can effectively suppress, but not eliminate the FIP1L1-PDGFRA clone and therefore ongoing treatment is recommended similar to guidelines in CML.

In contrast to CML, very few cases of acquired imatinib resistance have been reported with almost 10 years experience in treating FIP1L1-PDGFRA-positive disease [9, 69–71]. Most of the cases have involve the T674I mutation within the ATP-binding domain of PDGFRα and have occurred during the blast of disease. The T674I mutation is analogous to the T315I BCR-ABL mutation in CML which confers resistance to the tyrosine kinase inhibitors imatinib, dasatinib, and nilotinib [72]. One patient with the FIP1L1-PDGFRα T674I mutation in blast crisis responded briefly to sorafenib, but was followed by rapid emergence of a pan-resistant FIP1L1-PDGFRα D842V mutant [71]. Despite in vitro nanomolar activity of sorafenib, midostaurin (PKC412), or nilotinib against the T674I mutant [73–76], these drugs have limited clinical activity [77]. Thus far, the only report of primary resistance to imatinib relates to the tandem PDGFRA mutations, S601P and L629P, identified in a FIP1L1-PDGFRA-positive patient in the chronic phase of the disease [78]. In patients with rearrangements of PDGFRB or PDGFRA variants other than FIP1L1-PDGFRA, case reports and series indicate that imatinib, usually given at doses of 400 mg daily, can elicit durable hematologic and cytogenetic remissions [reviewed in 2,79]. Similar to FIP1L1-PDGFRA, FISH can be used to assess response to imatinib in PDGFRB-rearranged cases (Fig. 2 ). The natural history of patients with myeloid/lymphoid disease with rearranged FGFR1 follows an aggressive course usually terminating in AML in 1-2 years [14]. Therefore, intensive chemotherapy with regimens such as hyper-CVAD (directed to treatment of T- or B-cell lymphoma), followed by early allogeneic transplantation, is recommended. The data for use of small molecule inhibitors for patients with FGFR1-rearranged disease is sparse. The small molecule midostaurin (PKC412) inhibited the constitutively activated ZNF198-FGFR1 fusion in vitro, as well as in a patient who exhibited a hematologic and cytogenetic response [80]. Although not yet formally included in the WHO category of myeloid/lymphoid neoplasms with rearranged PDGFRA/B and FGFR1, other recurrent genetic abnormalities have been identified in patients with hypereosinophilia, including rearranged JAK2 (PCM1-JAK2 fusion) [81], and rearranged FLT3 (ETV6-FLT3 fusion) [82]. Involvement of these genes can be surmised by reciprocal translocations on standard karyotyping (e.g. 9p24 breakpoint for JAK2 and 13q12 for FLT3). The use of small molecule inhibitors (e.g. against JAK2 and FLT3) should be entertained for such patients.

Figure 2.

Interphase FISH results using a break-apart FISH probe for the PDGFRB gene region in a patient with MDS/MPN with eosinophilia and an abnormal karyotype, t(5;12)(q31-33;p13). A: Representative interphase nucleus demonstrating a typical abnormal signal pattern of a split red and green signal (corresponding to the PDGFRB gene disruption; see arrows) and one intact red/green fusion signal (corresponding to the normal/intact PDGFRB allele). 85% of nuclei were abnormal. B: After 3 months of treatment with imatinib and normalization of blood counts, a representative interphase nucleus demonstrates a normal FISH signal pattern with two intact Red/Green fusion signals, indicating two intact copies of the PDGFRB gene region. All cells demonstrated a normal FISH pattern. Courtesy of Dr. Rhett P. Ketterling, Mayo Clinic, Rochester. [Color figure can be viewed in the online issue, which is available at]

Imatinib's safety profile in eosinophilic disorders parallels the drug's good tolerability in CML. A few cases of cardiogenic shock have been reported in FIP1L1-PDGFRA-positive patients after initiation of imatinib [83, 84]. Currently, prophylactic use of steroids during the first 7–10 days of imatinib treatment is recommended for patients with known cardiac disease and/or elevated serum troponin levels which may be related to eosinophil-mediated heart damage or other cardiac comorbidities [84].


Imatinib is considered definitive treatment for PDGFRA/B-rearranged neoplasms with eosinophilia. The FDA-recommended starting dose for patients with the FIP1L1-PDGFRA rearrangement is 100 mg daily. Cumulative data with long-term follow-up indicate that this dose is sufficient to elicit complete and durable hematologic and complete molecular remissions. For patients with myeloid neoplasms (usually MDS/MPNs) with eosinophilia and rearranged PDGFRB, the recommended dose is 400 mg daily which reflects the dose consistently used in several case series with excellent outcomes.

Treatment of HES and CEL, NOS: Corticosteroids, hydroxyurea, and interferon-alpha

For patients with strictly defined HES (e.g., exclusion of all other possible causes of hypereosinophilia), corticosteroids (e.g., prednisone 1 mg/kg) are the mainstay of therapy and are effective in producing rapid reductions in the eosinophil count. However, therapy can be complicated by side effects in those patients requiring long-term treatment to suppress eosinophilia and organ damage. In a retrospective analysis, 141/188 (75%) HES patients received corticosteroids as initial monotherapy with 85% of these individuals achieving a complete or partial response after one month of treatment [19]. In this series, the median maximal dose was 40 mg (5–625 mg), the median maintenance dose was 10 mg daily (range, 1–40 mg daily) and the duration of therapy ranged from 2 months to 20 years. With symptom control and reduction of the eosinophil count to below 1,500/mm3, corticosteroids can be tapered. Recrudescence of symptoms, signs of organ damage, and/or significant increase of the eosinophil count with a prednisone dose > 10 mg daily is an indication for addition of other agents.

Hydroxyurea is an effective first-line agent for HES which may be used in conjunction with corticosteroids or in steroid nonresponders [18, 24, 85]. A typical starting dose is 500–1,000 mg daily. Hydroxyurea was used in 64/188 patients (34%) in the retrospective study; among 18 patients receiving hydroxuyrea as monotherapy, 13 patients (72%) achieved a complete or partial response [18]. When hydroxyurea was combined with corticosteroids, the overall response rate was 69%.

Interferon-α (IFN-α) can produce hematologic and cytogenetic remissions in HES and CEL patients refractory to other therapies including prednisone and/or hydroxyurea [86–92], or can be used in conjunction with corticosteroids as a steroid-sparing agent for individuals requiring higher doses of prednisone. Some have advocated its use as initial therapy for these disorders [91]. In the retrospective study, 46/188 patients were treated with IFN-α (mostly in combination with glucocorticoids) with response rates of 50 and 75% as monotherapy or combination therapy, respectively [19]. The optimal starting or maintenance dose of IFN-α has not been well-defined, but the initial dose required to control eosinophil counts often exceeds the doses needed to maintain a remission [93]. Initiation of therapy at 1 million units by subcutaneous injection three times weekly (tiw) and gradual escalation of the dose to 3–4 million units tiw or higher may be required to control hypereosinophilia in some patients. Remissions have been associated with improvement in clinical symptoms and organ disease, including hepatosplenomegaly [87, 91], cardiac, and thromboembolic complications [86, 88], mucosal ulcers [89], and skin involvement [92]. Treatment of four HES patients with PEG-IFN-α-2b among a larger cohort of BCR-ABL-negative MPN cohort resulted in 1 complete and 1 partial response, but side effects required that the initial study dose be reduced from 3 to 2 mcg/kg/week [94]. A lower starting dose of 90 mcg/kg weekly (e.g., 1–1.5 mcg/kg weekly) is better tolerated based on the experience of PEG-IFN-α-2a in PV and ET [95, 96]. Side effects of short- and longer-acting formulations of IFN-α are usually dose-dependent and include can include fatigue and flu-like symptoms, transaminitis, cytopenias, depression, hypothyroidism, and peripheral neuropathy. Unlike hydroxyurea which is a teratogen, interferon-alpha is considered safe for use in pregnancy.

Hematologic benefit has been observed with second- and third-line agents, including vincristine [97–99], cyclophosphamide [100], and etoposide [101, 102]. Responses to 2-chlorodeoxyadenosine alone [103] or in combination with cytarabine [104], and cyclosporin-A [105, 106] have also been reported in HES, with a discontinuation rate of 82% with CSA in one series due to poor tolerance [19].

In selected cases, patients with CEL, NOS, or HES may benefit from imatinib, usually administered at higher doses (>400 mg daily) [107]. However, hematologic responses in this group are more often partial, short-lived, and may reflect drug-related myelosuppression [9, 10]. Rare complete responses may represent diagnostically occult PDGFRA or PDGFRB mutations or other unknown pathogenetic targets [108]. Clinical trials with novel agents should always be considered.


Corticosteroids are potent anti-eosinophil agents with established efficacy in HES and should be considered first-line treatment. Similar to other MPNs, hydroxyurea can serve as effective palliative chemotherapy to control leukocytosis and eosinophilia, but with no proven role in favorably altering the natural history of HES or CEL, NOS. On the basis of a limited published literature, IFN-α has demonstrated hematologic responses and reversion of organ injury in patients with HES and CEL. The logic of using IFN-α in CEL is partly extrapolated from its efficacy in other MPNs such as CML, as well as PV and ET, and evidence for cytogenetic remitting activity. Although typically used as a second line-agent in HES steroid-failures, IFN-α could be used as initial therapy in patients with contra-indications or intolerance to steroid therapy. The optimal dose and duration of IFN-α therapy in HES is unknown and is tailored to individual response and tolerability.

Treatment of lymphocyte-variant hypereosinophilia

Patients with clonal population(s) of T-cells with an aberrant immunophenotype and/or cytokine production should initially be treated with corticosteroids. Patients who are refractory to therapy or exhibit relapse may be considered for treatment with IFN-α or steroid-sparing immunosuppressive agents. Hydroxyurea and imatinib are less likely to demonstrate efficacy in this lymphocyte-variant of hypereosinophilia compared with myeloproliferative forms of the disease which can be very responsive to these drugs as discussed above. Elevated serum IgE and TARC levels were associated with responsiveness to steroids in the lymphocyte-variant of hypereosinophilia [19].


Corticosteroids are first-line therapy in patients with hypereosinophilia in whom a clonal population of T-cells with an abnormal immunophenotype are identified and other causes of an elevated eosinophil count are excluded.

Antibody Approaches for HES


Anti-IL-5 antibody approaches have been studied in HES based on the cytokine's role as a differentiation, activation, and survival factor for eosinophils. Mepolizumab is a fully humanized monoclonal IgG antibody that inhibits binding of IL-5 to the α chain of the IL-5 receptor expressed on eosinophils [109]. In HES patients, regression of constitutional symptoms, eosinophilic dermatologic lesions, and improvements FEV1 measurements in individuals with pulmonary disease have been observed with anti-IL-5 antibody therapy [110–112]. Among the few patients studied, response has not been predicted by pre-treatment serum IL-5 levels or presence of FIP1L1-PDGFRA. Rebound eosinophilia, accompanied by increases in serum IL-5 levels, has been noted in some cases, and tachyphylaxis has been observed with repeated doses without development of neutralizing antibodies [108]. In the largest study of HES patients to date, the safety and steroid-sparing effects of mepolizumab was evaluated in a randomized, double-blind, placebo-controlled trial of 85 FIP1L1-PDGFRA-negative patients [113]. Blood eosinophil levels were stabilized at <1,000 cells/mm3 on 20–60 mg/day prednisone during a run-in period of up to 6 weeks. Patients were subsequently randomized to intravenous mepolizumab 750 mg or placebo every 4 weeks for 36 weeks. No adverse events were significantly more frequent with mepolizumab compared to placebo. A significantly higher proportion of mepolizumab-treated HES patients versus placebo were able to achieve the primary efficacy endpoint of a daily prednisone dose of ≤10 mg daily for at least 8 consecutive weeks. Mepolizumab has not yet been approved by the FDA, but is currently available on a compassionate use basis ( Identifier NCT00244686) for individuals with life-threatening HES who have failed prior therapies.

Reslizumab is a humanized anti-IL5 IgG4 mAb currently in clinical trials for pediatric subjects with eosinophilic esophagitis and for patients with eosinophilic asthma, but has not yet been evaluated extensively in HES [114]. In a double-blind, placebo-controlled, randomized trial, reslizumab significantly reduced intraepithelial esophageal eosinophil counts in children and adolescents with eosinophilic esophagitis, but symptom improvement was observed in both treatment groups [115].

It is unknown whether mepolizumab or other anti-IL5 antibody approaches have a role in WHO-defined eosinophilic myeloid disorders. However, preclinical models suggest a pathobiologic rationale for their use. Mice expressing FIP1L1-PDGFRA in their bone marrow cells only develop a general myeloproliferative disease [73]. In contrast, expression of FIP1L1-PDGFRA together with overexpression of IL-5 mimics eosinophilic disease much better in mice with typical features of HES such as tissue infiltration of eosinophils [116].


Alemtuzumab is an anti-CD52 monoclonal antibody that has been evaluated in idiopathic HES based on expression of the CD52 antigen on eosinophils [117, 118]. Similar to mepolizumab, it has not been formally evaluated in myeloid-related eosinophilias. In patients with refractory HES, alemtuzumab administered intravenously at a dose of 5–30 mg once to thrice weekly, elicited a hematologic remission in 10/11 subjects (91%), but responses were not sustained off therapy [119].


Use of anti-IL-5 and anti-CD52 antibody approaches in the treatment of HES remain investigational. Potential benefits include resolution of eosinophilia and disease-related symptomology, and a steroid-sparing effect (mepolizumab). However, with discontinuation of therapy, benefits appear to be short-lived and the potential for rebound eosinophilia exists.


Bone marrow/peripheral blood stem cell allogeneic transplantation has been attempted in patients with aggressive disease. Disease-free survival ranging from 8 months to 5 years has been reported [120–124] with one patient relapsing at 40 months [125]. Allogeneic transplantation using nonmyeloablative conditioning regimens have been reported in three patients, with remission duration of 3–12 months at the time of last reported follow-up [126, 127]. In one patient who underwent an allogeneic stem cell transplantation from an HLA-matched sibling, the patient was disease free at 3 years, and there was no evidence of the FIP1L1-PDGFRA fusion which was present at diagnosis [128]. Despite success in selected cases, the role of transplantation in HES is not well established.

Supportive care and surgery

Leukapheresis can elicit transient reductions in high leukocyte and eosinophil counts, but is not an effective maintenance therapy [129–131]. Similar to other myeloproliferative neoplasms, splenectomy has been performed for hypersplenism-related abdominal pain and splenic infarction, but is not considered standard treatment [20, 132]. Anticoagulants and antiplatelet agents have demonstrated variable success in preventing recurrent thromboembolism [20, 133, 134].

Advanced cardiac disease is less common today in patients with eosinophilic disease. Cardiac surgery can extend the life of patients with late-stage heart disease manifested by endomyocardial fibrosis, mural thrombosis, and valvular insufficiency [18, 24], Mitral and/or tricuspid valve repair or replacement [135–139] and endomyocardectomy for late-stage fibrotic heart disease [136, 140] can improve cardiac function. Bioprosthetic devices are generally preferred over their mechanical counterparts because of the reduced frequency of valve thrombosis.

Concluding Remarks

A more sophisticated understanding of the cellular and molecular basis of eosinophilic disorders has translated into more biologically oriented classification schemes which carry therapeutic implications. In this regard, imatinib for has dramatically reversed the poor-prognosis of patients previously diagnosed as “HES” and who likely represented a significant portion of older series of patients who exhibited a poor prognosis. Corticosteroids, interferon-alpha, hydroxyurea/other chemotherapy, and targeted antibodies can elicit clinical benefit in the primary eosinophilias, but response durability is variable, and treatment is often complicated by both short- and long-term side effects. Regarding future directions, the recently approved JAK1/2 inhibitor ruxolitinib (and other JAK inhibitors currently being evaluated in phase I-III trials of myelofibrosis) may have a role in patients with eosinophilic disease. For example, rare patients with hypereosinophilia have been found to carry the JAK2 V617F activating mutation [141, 142]. More germane to eosinophil biology is the finding that the JAK2 pathway mediates anti-apoptosis signals in eosinophils in response to GM-CSF and IL-5 [143, 144] in addition to FIP1L1-PDGFRα [145]. Inhibition of this signaling cascade may be a useful therapeutic approach across eosinophilic disorders regardless of their subtype. In addition, the finding that PDGF receptor fusion oncogenes skew proliferation and differentiation towards the eosinophil lineage in a process that requires NF-κB suggests the possibility for new treatments that target this pathway [146].