Eosinophils in peripheral blood account for 0.3-5% of leukocytes, which is equivalent to 0.05-0.5 x 109/l. A count equal or above 0.5 x 109/l is considered eosinophilia, while a count equal or above 1.5 x 109/l is defined as hypereosinophilia. In bone marrow, eosinophilia is considered when eosinophils make up more than 6% of the total nuclear cells. In daily clinical practice, the most common causes of reactive eosinophilia are non- hematologic, whether non-neoplastic (allergic diseases, drugs, infections or immunological diseases) or neoplastic (solid tumors). Eosinophilia associated with a haematological malignancy may be reactive or secondary to the production of eosinophilopoietic cytokines, and this is mainly seen in lymphoid neoplasms (Hodgkin lymphoma, mature T-cell neoplasms, lymphoid variant of hypereosinophilic syndrome and B-acute lymphoblastic leukemia/lymphoma). Eosinophilia associated with a haematological malignancy may also be neoplastic or primary, derived from the malignant clone, usually in myeloid neoplasms or with origin in stem cell (myeloid/lymphoid neoplasms with eosinophilia and tyrosine kinase gene fusions, acute myeloid leukemia with Core Binding Factor translocations, mastocytosis, myeloproliferative neoplasms, myelodysplastic/myeloproliferative neoplasms and myelodysplastic neoplasms). There are no concrete data in standardized cytological and cytometric procedures that could predict whether eosinophilia is reactive o clonal. The verification is usually indirect, based on the categorization of the accompanying hematologic malignancy. This review focuses on the broad differential diagnosis of haematological malignancies with eosinophilia.

Artificial intelligence methods may help in unveliling information hidden in high-dimensional oncological data. Flow cytometry studies of haematological malignancies provide quantitative data with the potential to be used for the construction of response biomarkers. Many computational methods from the bioinformatics toolbox can be applied to these data but have not been exploited in their full potential in leukaemias, specifically for the case of childhood B-cell acute lymphoblastic leukemia. In this paper we analysed flow cytometry data obtained on diagnosis from 54 paediatric B-cell acute lymphoblastic leukemia patients from two local institutions. We constructed classifiers based on the Fisher’s Ratio to quantify differences in expression levels of immunophenotypical markers between patients with relapsing and non-relapsing disease. The distribution of the marker CD38 was found and validated to have a strong discriminating power between both patient cohorts, thus providing a classifier.