A Chunking-for-Pooling Strategy for Cytometric Representation Learning for Automatic Hematologic Malignancy Classification

Differentiating types of hematologic malignancies is vital to determine therapeutic strategies for the newly diagnosed patients. Flow cytometry (FC) can be used as diagnostic indicator by measuring the multi-parameter fluorescent markers on thousands of antibody-bound cells, but the manual interpretation of large scale flow cytometry data has long been a time-consuming and complicated task for hematologists and laboratory professionals. Past studies have led to the development of representation learning algorithms to perform sample-level automatic classification. In this work, we propose a chunking-for-pooling strategy to include large-scale FC data into a supervised deep representation learning procedure for automatic hematologic malignancy classification. The use of discriminatively-trained representation learning strategy and the fixed-size chunking and pooling design are key components of this framework. It improves the discriminative power of the FC sample-level embedding and simultaneously addresses the robustness issue due to an inevitable use of down-sampling in conventional distribution based approaches for deriving FC representation. We evaluated our framework on two datasets. Our framework outperformed other baseline methods and achieved 92.3% unweighted average recall (UAR) for four-class recognition on the UPMC dataset and 85.0% UAR for five-class recognition on the hema.to dataset. We further compared the robustness of our proposed framework with that of the traditional downsampling approach. Analysis of the effects of the chunk size and the error cases revealed further insights about different hematologic malignancy characteristics in the FC data.

A Machine Learning Approach to the Classification of Acute Leukemias and Distinction From Nonneoplastic Cytopenias Using Flow Cytometry Data

OBJECTIVES

Flow cytometry (FC) is critical for the diagnosis and monitoring of hematologic malignancies. Machine learning (ML) methods rapidly classify multidimensional data and should dramatically improve the efficiency of FC data analysis. We aimed to build a model to classify acute leukemias, including acute promyelocytic leukemia (APL), and distinguish them from nonneoplastic cytopenias. We also sought to illustrate a method to identify key FC parameters that contribute to the model's performance.

METHODS

Using data from 531 patients who underwent evaluation for cytopenias and/or acute leukemia, we developed an ML model to rapidly distinguish among APL, acute myeloid leukemia/not APL, acute lymphoblastic leukemia, and nonneoplastic cytopenias. Unsupervised learning using gaussian mixture model and Fisher kernel methods were applied to FC listmode data, followed by supervised support vector machine classification.

RESULTS

High accuracy (ACC, 94.2%; area under the curve [AUC], 99.5%) was achieved based on the 37-parameter FC panel. Using only 3 parameters, however, yielded similar performance (ACC, 91.7%; AUC, 98.3%) and highlighted the significant contribution of light scatter properties.

CONCLUSIONS

Our findings underscore the potential for ML to automatically identify and prioritize FC specimens that have critical results, including APL and other acute leukemias.

Harmonization of Training, Training Requirements, Board Certification, and Practice of Hematopathology

OBJECTIVES

Hematopathology (HP) is a rapidly changing field with insufficient data to provide guidance to program directors (PDs), the Accreditation Council for Graduate Medical Education, or the American Board of Pathology.

METHODS

Two surveys were performed-one for HP PDs and one, given twice, for HP diplomates doing Maintenance of Certification/Continuing Certification reporting in 2017 to 2018.

RESULTS

Bone marrow (BM), lymph node (LN), and flow cytometry interpretations and peripheral blood/fluid reviews are performed by more than 80% of hematopathologists and are the areas with the greatest amount of training. A smaller proportion of hematopathologists is involved in other HP-related activities. Most PDs believed fellows should perform BM procedures. Interpretation of 400 or more LNs and 500 BMs was PDs' median expectations for fellows. PDs and HP diplomates considered coagulation and benign RBC disorders overemphasized on the certification examination.

CONCLUSIONS

These results highlight how varied the practice of HP is and can provide guidance to HP PDs, those responsible for assessing HP programs, and the American Board of Pathology.

Immunomodulatory drugs downregulate IKZF1 leading to expansion of hematopoietic progenitors with concomitant block of megakaryocytic maturation

The immunomodulatory drugs, lenalidomide and pomalidomide yield high response rates in multiple myeloma patients, but are associated with a high rate of thrombocytopenia and increased risk of secondary hematologic malignancies. Here, we demonstrate that the immunomodulatory drugs induce self-renewal of hematopoietic progenitors and upregulate megakaryocytic colonies by inhibiting apoptosis and increasing proliferation of early megakaryocytic progenitors via down-regulation of IKZF1. In this process, the immunomodulatory drugs degrade IKZF1 and subsequently down-regulate its binding partner, GATA1. This results in the decrease of GATA1 targets such as ZFPM1 and NFE2, leading to expansion of megakaryocytic progenitors with concomitant inhibition of maturation of megakaryocytes. The down-regulation of GATA1 further decreases CCND1 and increases CDKN2A expression. Overexpression of GATA1 abrogated the effects of the immunomodulatory drugs and restored maturation of megakaryocytic progenitors. Our data not only provide the mechanism for the immunomodulatory drugs induced thrombocytopenia but also help to explain the higher risk of secondary malignancies and long-term cytopenia induced by enhanced cell cycling and subsequent exhaustion of the stem cell pool.

Activating JAK2 mutants reveal cytokine receptor coupling differences that impact outcomes in myeloproliferative neoplasm

Janus tyrosine kinase 2 (JAK2) mediates downstream signaling of cytokine receptors in all hematological lineages, yet constitutively active JAK2 mutants are able to drive selective expansion of particular lineage(s) in myeloproliferative neoplasm (MPN). The molecular basis of lineage specificity is unclear. Here, we show that three activating JAK2 mutants with similar kinase activities in vitro elicit distinctive MPN phenotypes in mice by differentially expanding erythroid vs granulocytic precursors. Molecularly, this reflects the differential binding of JAK2 mutants to cytokine receptors EpoR and GCSFR in the erythroid vs granulocytic lineage and the creation of unique receptor/JAK2 complexes that generate qualitatively distinct downstream signals. Our results demonstrate that activating JAK2 mutants can differentially couple to selective cytokine receptors and change the signaling repertoire, revealing the molecular basis for phenotypic differences elicited by JAK2 (V617F) or mutations in exon 12. On the basis of these findings, receptor-JAK2 interactions could represent new targets of lineage-specific therapeutic approaches against MPN, which may be applicable to other cancers with aberrant JAK-STAT signaling.

Does Taking the Fellowship In-Service Hematopathology Examination and Performance Relate to Success on the American Board of Pathology Hematology Examination?

OBJECTIVES

The biannual Fellow In-Service Hematopathology Examination (FISHE) assesses knowledge in five content areas. We examined the relationship between taking the FISHE and performance on it with outcomes on the first attempted American Board of Pathology Hematology subspecialty certifying examination (ABP-HE).

METHODS

The pass rate between the ABP-HE candidates who took the spring FISHE and those who did not were compared. The likelihood of fellows passing the ABP-HE based on their percentiles on the FISHE was also assessed.

RESULTS

ABP-HE candidates who took the spring FISHE had a higher pass rate (96.4%) than those who did not (76.1%, P < .001). Spring FISHE performance, including total percentile and percentiles in four of five FISHE content areas, was only a weak predictor of passing the ABP-HE.

CONCLUSIONS

Candidates who take the spring FISHE do better on the ABP-HE than those who do not. Most fellows passed the first attempted ABP-HE regardless of FISHE performance. Whether this is due to fellows making use of the FISHE as a self-evaluation tool to help identify and then correct their knowledge deficiencies remains to be determined.

Longitudinal bone marrow evaluations for myelodysplasia in patients with myeloma before and after treatment with lenalidomide

Lenalidomide (LEN) treatment in multiple myeloma (MM) results in a superior outcome. However, there is concern for increased myelodysplastic syndrome/acute myeloid leukemia (MDS/AML) associated with LEN. Thus, bone marrow morphology and cytogenetics studies from 40 patients were evaluated for early signs of MDS prior to therapy, during therapy and at follow-up. Newly diagnosed patients with MM treated with LEN and dexamethasone (LD) alone or followed by autologous stem cell transplant (LD/ASCT), or patients with relapsed/refractory MM treated with LEN, bendamustine and dexamethasone (BLD) were included. One patient developed MDS. Baseline prevalence of mild morphologic myelodysplasia was highest in pretreated patients with MM (BLD, 71%), but was also seen in newly diagnosed patients (LD and LD/ASCT, 17%). The prevalence of myelodysplasia did not increase over time. Thus, this study did not reveal rapidly emerging MDS in 39 of 40 patients with MM treated with LEN. The development of MDS in one patient suggests that longer follow-up is needed for all.

Zinc-induced copper deficiency: a report of three cases initially recognized on bone marrow examination

Copper deficiency is a rare cause of sideroblastic anemia and neutropenia that often is not suspected clinically. The morphologic findings in bone marrow, while not pathognomonic, are sufficiently characteristic to suggest the diagnosis, leading to further testing to establish the correct diagnosis. Excess zinc ingestion is among the causes of copper deficiency. We present 3 cases of zinc-induced copper deficiency in which the diagnosis first was suggested on the basis of bone marrow examination. The first patient was a 47-year-old man with a debilitating peripheral neuropathy that had progressed during the previous 18 months, mild anemia, and severe neutropenia. The second was a 21-year-old man receiving zinc supplementation for acrodermatitis enteropathica in whom moderate normocytic anemia and neutropenia developed. The third patient was a 42-year-old man with anemia, severe neutropenia, and a peripheral neuropathy that had progressed during 8 months. The bone marrow findings in all cases suggested copper deficiency, which was confirmed by further laboratory testing and determined to be due to zinc excess. The morphologic features, clinical manifestations, differential diagnosis, and pathogenetic mechanisms are discussed.

Zinc-induced copper deficiency: a report of three cases initially recognized on bone marrow examination

Copper deficiency is a rare cause of sideroblastic anemia and neutropenia that often is not suspected clinically. The morphologic findings in bone marrow, while not pathognomonic, are sufficiently characteristic to suggest the diagnosis, leading to further testing to establish the correct diagnosis. Excess zinc ingestion is among the causes of copper deficiency. We present 3 cases of zinc-induced copper deficiency in which the diagnosis first was suggested on the basis of bone marrow examination. The first patient was a 47-year-old man with a debilitating peripheral neuropathy that had progressed during the previous 18 months, mild anemia, and severe neutropenia. The second was a 21-year-old man receiving zinc supplementation for acrodermatitis enteropathica in whom moderate normocytic anemia and neutropenia developed. The third patient was a 42-year-old man with anemia, severe neutropenia, and a peripheral neuropathy that had progressed during 8 months. The bone marrow findings in all cases suggested copper deficiency, which was confirmed by further laboratory testing and determined to be due to zinc excess. The morphologic features, clinical manifestations, differential diagnosis, and pathogenetic mechanisms are discussed.

Pan B-cell markers are not redundant in analysis of chronic lymphocytic leukemia (CLL)

BACKGROUND

The classic immunophenotype for chronic lymphocytic leukemia (CLL) is CD19(+), restricted dim surface expression of kappa or lambda light chain, CD5(+), CD23(+), dim CD20(+), negative FMC7, and negative CD79b. However, the necessity of assaying for all 3 pan B-cell markers (CD20, FMC7, and CD79b) by flow cytometry has not been definitively documented for CLL.

METHODS

Qualitative patterns and semi-quantitative assessment of staining intensity for CD20, FMC7 and CD79b were performed in 70 cases with a current or prior diagnosis of CLL or CLL with increased prolymphocytes leukemia (CLL/PL). The concurrent morphology in 66 of 70 specimens was classified as typical CLL in 53 cases, CLL/PL in 10 cases, and large cell lymphoma in 3 cases.

RESULTS

Forty percent of the cases varied from the characteristic immunophenotype by having moderate or bright staining of CD20 (36%), FMC7 (7%), and/or CD79b (18%). Discrepant qualitative staining patterns were found between FMC7 and CD20 (21%), CD20 and CD79b (15%), and CD79b and FMC7 (10%). Semiquantitative measurement of staining intensity showed little correlation between CD79b and CD20 or FMC7. Moderate correlation was seen between CD20 and FMC7. No correlation was observed between morphology and intensity of marker expression.

CONCLUSIONS

Variable patterns and intensity of staining were seen for FMC7, CD20, and CD79b in this cohort of CLL samples. Dim or negative staining was most consistently seen for FMC7 (93% of specimens). Although FMC7 staining intensity was moderately correlated with CD20, CD79b intensity was poorly correlated with either CD20 or FMC7, and thus, may provide some independent information.