Issue Archive

Artificial intelligence (AI) and machine learning algorithms are changing many facets of our lives and have great potential for disease diagnosis. In this Plenary Paper, Matek et al describe training a convolutional neural network model using 171 374 bone marrow cytology images from 945 patients with various hematological diseases before validating its accuracy in classifying single cells in an independent set of 627 images. The system can automatically classify 22 classes of bone marrow leukocytes, and with further improvements, it may bring AI-aided diagnosis of bone marrow biopsy specimens closer to fruition.

Some patients with paroxysmal nocturnal hemoglobinuria (PNH) remain persistently anemic despite treatment with eculizumab and may have significant extravascular hemolysis. Kulasekararaj and colleagues report a phase 2 trial of danicopan, a first-inclass oral complement factor D inhibitor, added to ongoing eculizumab therapy in PNH patients who still needed transfusions. This small study of 12 patients reveals that the combination has acceptable safety while substantially increasing the hemoglobin concentration, reducing extravascular hemolysis, and improving the fatigue score.

Wang et al provide insight into how epigenetic regulation of the transcriptome can rewire cell fate decisions by mapping the RNA-editing landscape during hematopoiesis. They report that adenosine-to-inosine RNA editing of antizyme inhibitor 1 (Azin1) is a novel regulator of hematopoietic cell fate, capable of influencing self-renewal and differentiation of hematopoietic stem cells. This work sets the scene for developing RNA editing–targeted therapeutics for stem cell expansion and modulating AZIN1-induced cancer stem cell generation.

Current improvements in treatment for children with acute lymphoblastic leukemia (ALL) are based on detailed molecular classification that influences prognosis and can guide therapy. Boldrin et al identified a novel biomarker of poor outcome in ALL, and, using cell and in vivo models, they delineated the biological link between microRNA-497/195 repression and more aggressive disease. Their research also points to exploring pharmacological cell cycle inhibition as a potential method for treating these high-risk leukemias.

Activating mutations in MYD88 are highly prevalent in Waldenström macroglobulinemia, primary central nervous system lymphoma, and several other lymphomas. Inhibitors of Bruton tyrosine kinase (BTK) have clinical activity, but over time resistance emerges. Yang and colleagues report the preclinical efficacy of a novel inhibitor that targets both BTK and hematopoietic cell kinase (HCK) and has activity against ibrutinib-resistant tumors.

In this month’s CME article, Bazarbachi et al describe the genomic characterization of immunoglobulin M (IgM) multiple myeloma (MM) and contrast this rare entity with the far more common non-IgM MM and Waldenström macroglobulinemia. Their results help explain why IgM-MM is a distinct clinical and biological entity and highlight specific genetic abnormalities and transcriptomic features. These findings provide a rational basis for the future exploration of additional targeted therapies beyond those typically prescribed for MM in general.

Cold antibody-mediated autoimmune hemolytic anemia (cAIHA) is a rare disease, usually related to an underlying indolent clonal B-cell lymphoproliferative disorder. Jalink and colleagues describe an international retrospective analysis of Bruton tyrosine kinase (BTK) inhibition in patients with cAIHA. The data indicate a rapid and notable improvement in both the hemolytic anemia as well as acrocyanosis, suggesting that this may be an effective approach, regardless of underlying pathology or MYD88 mutational status.