Subclassification Of Acute Myeloid Leukemia Research Articles

Machine Learning-Facilitated Cross-Panel Automatic Classification at Both Sample and Cell Level in Acute Myeloid Leukemia

Introduction Flow cytometry (FC) immunophenotyping is essential for accurate and prompt diagnosis of acute myeloid leukemia (AML), with up to 95% of AML patients showing detectable aberrant immunophenotype (IP) by FC. Current manual and subjective analysis of clinical FC data, however, lacks reproducibility and leads to interpretation variation. In previous studies, we showed that supervised machine learning (ML) approaches can detect IP abnormalities and identify the acute leukemia subclassification accurately at the specimen level. We also found that comparable classification performance can be achieved using only a subset of markers of the reagent panel (Ko et al., 2018; Monaghan et al., 2022). To interpret the ML classification results, we have developed data clustering and discriminative learning methods to identify and visualize the diagnostic cell populations in a 2D space for hematopathology review (Ji et al., 2020; Lee et al., 2018). Built upon these approaches, this study reports our findings on developing an interpretable cross-panel ML classification model to support timely and accurate diagnosis and subclassification of AML. Materials and Method The AML diagnosis model at the sample level utilized FC data obtained from 53 bone marrow (BM) aspirate samples from Roswell Park Comprehensive Cancer Center (RPCCC) and 50 BM aspirate samples from UPMC. Data from RPCCC were acquired using a Beckman Coulter Navios EX and measured with the ClearLLab10C panel; while data from UPMC were acquired using a BD FACSCantoII and measured with a UPMC-developed diagnostic panel. We used only the common parameters collected within these 103 FC data sets and applied a previously published Gaussian Mixture Model-Support Vector Machine (GMM-SVM) based approach to train a cross panel sample-level classification model for AML diagnosis (Ko et al., 2018; Monaghan et al., 2022). Three-fold cross-validation was conducted to assess the model performance, using sensitivity, specificity, classification accuracy, and area under the receiver operating characteristic curve (AUC). For the cell-level visualization and analysis, all samples in the training set were pooled together, before a uniform manifold approximation and projection (UMAP) process was applied to transform and visualize the cell populations within the samples. Specifically, 26 non-neoplastic and 27 AML samples from the M2 panel of the ClearLLab10C from the RPCCC cohort were combined for the UMAP visualization. Then the FlowSOM unsupervised clustering method and the DAFI semi-supervised clustering method (Lee et al., 2018) were applied to identify all of the cell populations that have significant differences between the AML samples and the non-neoplastic controls. Results The cross-panel AML versus non-neoplastic classifier, combining FC data from UPMC and RPCCC achieved exceptional performance. The AML detection model using all parameters on the ClearLLab10C panel achieved an AUC of 100%. When using only the overlapping parameters between the ClearLLab10C and the UPMC panels, the classification model achieved an AUC of 99.5%, while the cross-panel classifier achieved an AUC of 99.1% (Table 1). Specimen-level visualization confirmed the separation of AML versus non-neoplastic controls. The UMAP plots, created by pooling non-neoplastic and AML samples, exhibited distinct patterns for different AML subtypes (M1 to M5) and a genetic mutation (e.g., AML IDH1), which were distinct from control samples. Using the automated gating analysis, cell populations identified as different between AML and control samples were characterized and visualized in 2D-by-2D scatter plots for straightforward interpretation and assessment based on their immunophenotype. Conclusion Our study demonstrated the feasibility to develop and assess a ML classification model for AML detection and diagnosis using overlapping markers in FC data across reagent panels and sites. Advanced visualization and automated gating analysis enabled the interpretation of the ML classification by identifying and characterizing the diagnostic cell populations and their phenotypic heterogeneity, providing insights into AML heterogeneity at both the specimen and the single cell level. These findings highlight the potential of ML capability for improving patient care by timely and accurate diagnosis and subclassification of AML.

Clinicopathological Evaluation of Acute Leukemias in a Tertiary Care Hospital: A Cross-Sectional Study.

Acute Myeloid Leukemia (AML) and Acute Lymphoblastic Leukemia (ALL) are clinically and biologically diverse phenotypic diseases amongst hematological malignancies. The current study objectives were to diagnose and classify cases of AL as per revised 4th edition of WHO 2016 classification of AL's and study their clinicopathological profiles. This cross-sectional, observational study included 68 patients, diagnosed with AL were recruited. Diagnosis was based on peripheral blood smear examination, bone marrow aspiration, flowcytometry, and cytogenetic and molecular studies. Sixty-eight cases of AL were diagnosed in a period of 2 years, where 25 cases were of ALL and 43 cases were of AML. In the subclassification of AML as per WHO 2016, 20 cases were of AML, RGA, 21 cases were of AML, NOS, and 2 cases were of AML, MRC. In AML, RGA, APL with PML-RARA positive cases were 10 out of 20 cases, AML with (8;21) RUNX1-RUNX1T1 were 7/20 cases; there were two cases of AML with mutated NPM1 gene and one case of AML with biallelic mutation of CEBPA. In AML, NOS subcategory AML with maturation was more common with 9/21cases. In subcategory of ALL, B-ALL was more common than T-ALL. B-ALL, NOS was more common than B-ALL, RGA and we had 1 case of NK cell Leukemia. The application of revised 4th edition WHO 2016 classification confers uniformity in reporting acute leukemia cases that aids in the treatment by using targeted therapies and helps in the prediction of prognosis. The WHO classification for acute leukemias is very objective, therapy oriented and the need of the hour.

Scalable Prediction of Acute Myeloid Leukemia Using High-Dimensional Machine Learning and Blood Transcriptomics.

SummaryAcute myeloid leukemia (AML) is a severe, mostly fatal hematopoietic malignancy. We were interested in whether transcriptomic-based machine learning could predict AML status without requiring expert input. Using 12,029 samples from 105 different studies, we present a large-scale study of machine learning-based prediction of AML in which we address key questions relating to the combination of machine learning and transcriptomics and their practical use. We find data-driven, high-dimensional approaches—in which multivariate signatures are learned directly from genome-wide data with no prior knowledge—to be accurate and robust. Importantly, these approaches are highly scalable with low marginal cost, essentially matching human expert annotation in a near-automated workflow. Our results support the notion that transcriptomics combined with machine learning could be used as part of an integrated -omics approach wherein risk prediction, differential diagnosis, and subclassification of AML are achieved by genomics while diagnosis could be assisted by transcriptomic-based machine learning.

The Value of CD64 Expression in Distinguishing Acute Myeloid Leukemia With Monocytic Differentiation From Other Subtypes of Acute Myeloid Leukemia: A Flow Cytometric Analysis of 64 Cases

Abstract Context.—Flow cytometric immunophenotyping is a useful ancillary tool in the diagnosis and subclassification of acute myeloid leukemias (AMLs). A recent study concluded that CD64 is sensitive and specific for distinguishing AMLs with a monocytic component (ie, AML M4 and AML M5) from other AML subtypes. However, in that study, the intensity of CD64 was not well defined and the number of non-M4/non-M5 AMLs was small. Objective.—To evaluate the usefulness of CD64 by flow cytometric immunophenotyping in distinguishing AMLs with monocytic differentiation from other AML subtypes. Design.—Sixty-four AMLs subclassified based on the French-American-British and World Health Organization classifications on pretreatment bone marrows were retrieved from our files (7 M0s, 11 M1s, 17 M2s, 7 M3s, 9 M4s, 7 M5s, 4 M6s, and 2 M7s). A standard panel of markers, including CD2, CD3, CD5, CD7, CD10, CD11b, CD13, CD14, CD15, CD19, CD20, CD33, CD34, CD45, CD56, CD64, CD117, and HLA-DR, were analyzed by flow cytometric i...

The Value of CD64 Expression in Distinguishing Acute Myeloid Leukemia With Monocytic Differentiation From Other Subtypes of Acute Myeloid Leukemia: A Flow Cytometric Analysis of 64 Cases

Flow cytometric immunophenotyping is a useful ancillary tool in the diagnosis and subclassification of acute myeloid leukemias (AMLs). A recent study concluded that CD64 is sensitive and specific for distinguishing AMLs with a monocytic component (ie, AML M4 and AML M5) from other AML subtypes. However, in that study, the intensity of CD64 was not well defined and the number of non-M4/non-M5 AMLs was small. To evaluate the usefulness of CD64 by flow cytometric immunophenotyping in distinguishing AMLs with monocytic differentiation from other AML subtypes. Sixty-four AMLs subclassified based on the French-American-British and World Health Organization classifications on pretreatment bone marrows were retrieved from our files (7 M0s, 11 M1s, 17 M2s, 7 M3s, 9 M4s, 7 M5s, 4 M6s, and 2 M7s). A standard panel of markers, including CD2, CD3, CD5, CD7, CD10, CD11b, CD13, CD14, CD15, CD19, CD20, CD33, CD34, CD45, CD56, CD64, CD117, and HLA-DR, were analyzed by flow cytometric immunophenotyping in all AMLs (52 bone marrow samples; 12 peripheral blood samples). CD64 was expressed in AML subtypes M0 to M5 in varying intensities: heterogeneously expressed in 1 of 7 M0s; dimly expressed in 3 of 11 M1s; dimly and moderately expressed in 6 and 2 of 17 M2s, respectively; dimly and moderately expressed in 5 and 1 of 7 M3s, respectively; dimly expressed in 4 of 9 M4s; and heterogeneously, moderately, and strongly expressed in 1, 3, and 3 of 7 M5s, respectively. Strong CD64 expression distinguishes AML M5; however, heterogeneous, dim, or moderate expression in itself does not distinguish M0 through M4 subtypes from M5 with dim to moderate CD64 expression. However, any CD64 expression associated with strong CD15 expression distinguishes AML M4 or M5, from other AML subtypes.

Proposed changes in the definitions of acute myeloid leukemia and myelodysplastic syndrome: are they helpful?

For most of the 20th century, subclassification of acute myeloid leukemia (AML) was based on the resemblance of blasts to normal hematopoiesis. This approach was standardized by the French-American-British (FAB) group. Because of limited clinical relevance, clinicians resorted to other patient characteristics to determine treatment and predict outcome in AML. A different approach based on the relationship of a case to myelodysplastic syndrome (MDS) has been proposed. The new World Health Organization (WHO) subclassification of AML includes elements of this new proposal but retains as a major category the historical subclassification. The WHO group has also proposed modifications of the FAB subclassification of MDS. These MDS proposals have generated discussion of diagnostic criteria for MDS and a philosophical discussion of whether MDS should still be considered a syndrome, or rather a specific set of diseases characterized by genetic instability and poor outcome.

The usefulness of CD64, other monocyte-associated antigens, and CD45 gating in the subclassification of acute myeloid leukemias with monocytic differentiation.

Immunophenotyping by flow cytometry is widely used in the diagnosis and subclassification of acute myeloid leukemia (AML). CD14 is the monocyte-associated antigen most widely used to identify AML with monocytic differentiation (French-American-British classes M4 and M5); however, we observed that CD14 expression is frequently diminished or absent in such cases. To identify monocyte-associated antigens that might improve recognition of AML M4 and M5, we used 3-color flow cytometry and a panel of antibodies reported to distinguish cells of monocytic lineage in 44 cases of AML. In addition, CD45 vs logarithmic side scatter plots were analyzed in all cases. As expected, CD14 was highly specific but was only moderately sensitive for monocytic differentiation. CD64 had the best-combined sensitivity and specificity for AML M4 and M5. CD45 vs logarithmic side scatter analysis showed a higher percentage of monocytes in AML M4 and M5 compared with nonmonocytic AML. CD64 was expressed in 5 of 5 cases of acute promyelocytic leukemia (AML M3), but the intensity of staining was significantly less in AML M3 than in AML M4 and M5. Our findings show that addition of CD64 and CD45 vs logarithmic side scatter analysis to CD14 greatly improves flow cytometric detection of AML with monocytic differentiation and that CD64, also expressed in AML M3, may help distinguish AML M3 from other subtypes.

Role of blast cell immunophenotyping for the diagnosis and prognosis of acute myeloid leukemia.

Bone marrow blast cell antigen expression from 86 patients with de novo acute myeloid leukemias (AML) was studied and correlated with FAB classification and clinical outcome. Among a panel of 14 monoclonal antibodies routinely used for the diagnosis of acute leukemias we studied the expression of six antibodies (CD13, CD15, VIM2, CD33, CD14, CD34) of the granulomonocytic lineage and found that some of them were useful for diagnosis and/or prognosis. For FAB subclassification of AML, the CD13 or VIM2 antigen expression was of no benefit. Monocytic leukemias (M4 + M5PD + M5WD) more frequently expressed CD34 antigen (28/31) than granulocytic (M1 + M2 + M3) subtypes (33/53) (P < 0.01). Finally, the most striking differences were found with CD14 antigen expression: CD14 antigen was more frequently expressed in M4 + M5 leukemias (21/31) than in M1 + M2 + M3 subtypes (12/33) (P < 0.01). The mean percentage of CD14 positive blast cells was accordingly higher in monocytic leukemias than in granulocytic leukemias and the difference was highly significant (P < 0.0001). The CD15 antigen was more frequently expressed in differentiated leukemias (M2 + M3 + M4 + M5WD) (35/44) than in poorly differentiated forms (M1 + M5PD) (17/37) (P < 0.001). The statistical difference was higher when the mean percentage of CD15 positive blast cells were compared (P < 0.0003). Moreover these latter percentages were different in M1 and M2 subtypes (P < 0.003). The blast cell expression of CD13, CD14, CD15 or CD33 was not predictive of the length of CR or survival. Moreover, our results support previously published findings suggesting a longer overall survival duration for patients whose leukemic cells do not express the CD34 antigen (P < 0.01). We also confirm that patients with the more differentiated subtypes of AML (CD13-, CD34+) tend to survive longer than patients with the less differentiated subtypes of AML (CD13-, CD34+) (P < 0.001).

Electrophoretic and cytochemical characterization of alpha-naphthyl acetate esterases in acute myeloid leukemia: relationships with membrane receptor and monocyte-specific antigen expression

Alpha-naphthyl acetate esterases (ANAE) were examined by cytochemical and isoelectric focusing (IEF) techniques in 48 cases of acute myeloid leukemia that were classified by conventional morphological criteria. Four main types of ANAE isoenzyme patterns were found by IEF, and comparisons with the expression of membrane receptors (Fc-IgG and C3b) and monocyte-specific antigens (UCHM1, UCHALF, and E11) suggest relationships between ANAE isoenzyme synthesis and distinct myeloid maturational stages. The results further indicate that the blast cells of acute myelomonocytic leukemia (AMML) may represent an immature variant of monocytic leukemia (AMoL) and that morphological examination alone is inadequate in the assessment of monocytic differentiation in acute myeloid leukemias. Inhibition studies of cytochemical ANAE activity with sodium fluoride (NaF) show that the presence of NaF- sensitive or NaF-resistant ANAE enzymes is often unrelated to the diagnostic category of acute leukemia. The results of this study are examined in relation to current concepts of myeloid differentiation, and the application of these findings to the subclassification of acute myeloid leukemias is discussed.

Electrophoretic and Cytochemical Characterization of Alpha-Naphthyl Acetate Esterases in Acute Myeloid Leukemia: Relationships With Membrane Receptor and Monocyte-Specific Antigen Expression

Alpha-naphthyl acetate esterases (ANAE) were examined by cytochemical and isoelectric focusing (IEF) techniques in 48 cases of acute myeloid leukemia that were classified by conventional morphological criteria. Four main types of ANAE isoenzyme patterns were found by IEF, and comparisons with the expression of membrane receptors (Fc-IgG and C3b) and monocyte-specific antigens (UCHM1, UCHALF, and E11) suggest relationships between ANAE isoenzyme synthesis and distinct myeloid maturational stages. The results further indicate that the blast cells of acute myelomonocytic leukemia (AMML) may represent an immature variant of monocytic leukemia (AMoL) and that morphological examination alone is inadequate in the assessment of monocytic differentiation in acute myeloid leukemias. Inhibition studies of cytochemical ANAE activity with sodium fluoride (NaF) show that the presence of NaF-sensitive or NaF-resistant ANAE enzymes is often unrelated to the diagnostic category of acute leukemia. The results of this study are examined in relation to current concepts of myeloid differentiation, and the application of these findings to the subclassification of acute myeloid leukemias is discussed.

Morphology, cytochemistry, and ultrastructure of leukemic cells with regard to the classification of leukemias.

Morphology and cytochemistry now play a major part in the diagnosis and subclassification of acute myeloid leukemias (AML). In acute lymphatic leukemias (ALL), however, these methods have largely been replaced by immunological membrane marker investigations.