Classification Of Acute Leukemia Research Articles

Improving Acute Leukemia Classification through Recursive Feature Elimination and Multilayer Perceptron Analysis of Gene Expression Data

This study presents an approach to improving the classification of acute leukemia subtypes using gene expression data analysis. Leveraging Recursive Feature Elimination (RFE) as a feature selection technique and Multilayer Perceptron (MLP) as the predictive modeling framework, this research aims to identify the most influential genes for distinguishing between Acute Lymphoblastic Leukemia (ALL) and Acute Myeloid Leukemia (AML) cases. RFE systematically ranks and selects the most discriminative gene attributes, while MLP constructs a predictive model based on these attributes. The results demonstrate the effectiveness of this combined approach, achieving precision, accuracy, F1-Score, and recall rates of approximately 99% for leukemia subtype classification. Furthermore, specific genes contributing most to the model's predictive power and shedding light on potential biomarkers for leukemia diagnosis were identified. This research underscores the significance of RFE and MLP in the analysis of gene expression data and their potential impact on clinical decision-making in the field of oncology.

In Silico Identification of Effective Genes for Acute Leukemia Classification Using a Spline Regression-based Framework

Background: Microarray technology enables the examination of gene expression in thousands of genes and can be highly effective in identifying various types of cancers, including leukemia. However, many genes in microarray data are redundant and lack useful information for cancer diagnosis. The main objective of this study is to identify relevant and effective genes in classification of leukemia microarray data using a spline regression-based method, taking into account the correlation between genes.
 Materials and Methods: In this analytical study, leukemia microarray data are used to identify relevant genes in classification of leukemia into Acute Myeloid Leukemia (AML) and Acute Lymphoblastic Leukemia (ALL) using a spline regression-based gene selection method, called SRS3FS based on ℓ2,p-norm (0 < p ≤ 1). Subsequently, the support vector machine (SVM) algorithm is employed to classify leukemia data into AML and ALL.
 Results: In this study, the classification results of SVM algorithm for 5, 10, 15, and 20 genes reveal that the SRS3FS method, employing ℓ2,1/4-norm, ℓ2,1/2-norm and ℓ2,3/4-norm, exhibited the highest accuracy of 97.06% when identifying 10 genes for distinguishing between AML and ALL. Moreover, the leukemia data was classified into AML and ALL with an accuracy of 100%, using a gene identified by the SRS3FS method based on ℓ2,3/4-norm and ℓ2,1-norm. The gene labeled as number 3252, annotated as GLUTATHIONE S-TRANSFERASE, MICROSOMAL, is recognized as the most important gene.
 Conclusion: The experimental results on leukemia microarray data demonstrate that the spline regression-based gene selection method can effectively identify relevant genes in classification and prediction of leukemia.

An efficient decision support system for leukemia identification utilizing nature-inspired deep feature optimization.

In the field of medicine, decision support systems play a crucial role by harnessing cutting-edge technology and data analysis to assist doctors in disease diagnosis and treatment. Leukemia is a malignancy that emerges from the uncontrolled growth of immature white blood cells within the human body. An accurate and prompt diagnosis of leukemia is desired due to its swift progression to distant parts of the body. Acute lymphoblastic leukemia (ALL) is an aggressive type of leukemia that affects both children and adults. Computer vision-based identification of leukemia is challenging due to structural irregularities and morphological similarities of blood entities. Deep neural networks have shown promise in extracting valuable information from image datasets, but they have high computational costs due to their extensive feature sets. This work presents an efficient pipeline for binary and subtype classification of acute lymphoblastic leukemia. The proposed method first unveils a novel neighborhood pixel transformation method using differential evolution to improve the clarity and discriminability of blood cell images for better analysis. Next, a hybrid feature extraction approach is presented leveraging transfer learning from selected deep neural network models, InceptionV3 and DenseNet201, to extract comprehensive feature sets. To optimize feature selection, a customized binary Grey Wolf Algorithm is utilized, achieving an impressive 80% reduction in feature size while preserving key discriminative information. These optimized features subsequently empower multiple classifiers, potentially capturing diverse perspectives and amplifying classification accuracy. The proposed pipeline is validated on publicly available standard datasets of ALL images. For binary classification, the best average accuracy of 98.1% is achieved with 98.1% sensitivity and 98% precision. For ALL subtype classifications, the best accuracy of 98.14% was attained with 78.5% sensitivity and 98% precision. The proposed feature selection method shows a better convergence behavior as compared to classical population-based meta-heuristics. The suggested solution also demonstrates comparable or better performance in comparison to several existing techniques.

Evaluating Thresholding-Based Segmentation and Humoment Feature Extraction in Acute Lymphoblastic Leukemia Classification using Gaussian Naive Bayes

This study, titled "Evaluating Thresholding-Based Segmentation and HuMoment Feature Extraction in Acute Lymphoblastic Leukemia Classification using Gaussian Naive Bayes," investigates the application of image processing and machine learning techniques in the classification of Acute Lymphoblastic Leukemia (ALL). Utilizing a dataset of microscopic blood smear images, the research focuses on the efficacy of thresholding-based segmentation and Hu moment feature extraction in distinguishing between benign and malignant cases of ALL. Gaussian Naive Bayes, known for its simplicity and effectiveness, is employed as the classification algorithm. The study adopts a 5-fold cross-validation approach to evaluate the model's performance, with particular emphasis on metrics such as accuracy, precision, recall, and F1-score. Results indicate a high precision rate across all folds, averaging approximately 84.13%, while exhibiting variability in accuracy, recall, and F1-scores. These findings suggest that while the model is effective in identifying malignant cases, further refinements are necessary for improving overall accuracy and consistency. This research contributes to the field of medical image analysis by demonstrating the potential of combining simple yet efficient techniques for the automated diagnosis of hepatological diseases. It highlights the importance of integrating image processing with machine learning to enhance diagnostic accuracy in medical applications.

Precise Single-Cell Transcriptomic Mapping of Leukemia Cell States Reveals Unconventional Lineage Priming in Acute Myeloid Leukemia

Initial disease classification within acute leukemia relies on identifying morphological and immunophenotypic features of hematopoietic differentiation retained by leukemic blasts. While existing approaches can classify leukemic cells into broad lineages, they lack precision in discerning between specific cell states, particularly at the level of immature blasts. Single-cell (sc) RNA-sequencing (RNA-seq) provides thousands of new markers to enable precise determination of leukemia cell state and provides an opportunity to refine our classification of acute leukemia. To extend our ability to identify leukemia cell states by scRNA-seq, we developed a comprehensive reference map of human bone marrow hematopoiesis. Specifically, we integrated three unsorted bone marrow datasets and three CD34+ hematopoietic stem and progenitor (HSPC) purified datasets to ensure balanced representation of early HSPCs and terminally differentiated immune cells. This culminated in a curated atlas of 263,159 high-quality sc transcriptomes, representing 55 cellular states spanning 45 healthy donors (Fig 1A). We validated our cell state annotations to ensure concordance between transcriptional states and functional populations through reference mapping of bulk and sc transcriptomes from purified HSPC subsets. For example, mapping of scRNA-seq profiles from Lin-CD34+38-45RA-90+49f+ long-term (LT) HSCs revealed 91% concordance between transcriptional HSC and functional LT-HSC. We next performed scRNA-seq profiling on 12 AML patient samples harbouring either myelodysplasia-related genetic features or classical alterations involving NPM1 or KMT2A. By mapping leukemia cells onto our reference atlas, we identified pronounced involvement of early erythroid-like blasts in a subset of AML samples. To further investigate unconventional lineage priming in AML, we processed and mapped scRNA-seq profiles from eleven studies comprising 166 additional AML samples and 9 mixed phenotype acute leukemia (MPAL) samples (Fig 1B). After quality control and exclusion of mature lymphoid cells, we performed composition analysis utilizing 600,570 leukemia cells mapped to 38 cell states to identify patterns of variation among leukemia cell states. Immature leukemia cells map to a range of cellular states spanning HSC/MPP-like, LMPP-like, MLP-like, GMP-like, MEP-like, and EarlyEry-like, among others. Notably, we identified a subset of AMLs with high MLP-like enrichment that co-cluster with MPALs as well as a subset with high EarlyEry-like involvement that co-cluster with acute erythroid leukemias (AEL), thus highlighting the biological continuum between these diseases. We next identified marker genes from each leukemia cell state and trained sparse regression models to estimate their relative abundance within bulk RNA-seq cohorts. Through bulk analysis, we confirmed that MPAL is highly enriched for MLP-like cells compared to AML (p=2.5e-12) and that M6 AELs are highly enriched for EarlyEry-like cells compared to M0-M5 AMLs (p=0.00067). Among 864 AML patients, we found that high MLP-like abundance was associated with NUP98-NSD1 fusions (p=0.0011) and RUNX1 mutations (p=8.0e-10) while high EarlyEry-like abundance was associated with Complex Cytogenetics (p=8e-15) and TP53 mutations (p=2.3e-25). Cellular states of immature leukemic cells captured inter-patient heterogeneity within 136 AEL samples, wherein high EarlyEry-like abundance was associated with TP53 mutations (p=3.6e-9) and Poor cytogenetic risk (p=1.1e-9) while high GMP-like abundance was associated with KMT2A alterations (p=0.00046) and Good/Intermediate risk (p=0.00032). Finally, through integration of sc transcriptomics and targeted DNA profiling, we found examples of genetic subclones impacting the cellular hierarchies of individual patients through induction of early differentiation blocks, skewing of lineage output from myeloid to erythroid, or potentiation of transcriptional self-renewal programs within mature myeloid cells. Together, our high-resolution mapping approach highlights the cellular state continuum between acute leukemias and enumerates the impact of genetic drivers on leukemia cell hierarchies. Precise definitions of cellular states, coupled with the presence or absence of genetic alterations, may help to refine future disease classification, prognostication, and therapy development.

Comparison of the 2022 ICC and Who Classification Systems of Myelodysplastic Neoplasms

Introduction: Recently, the World Health Organization (WHO) updated the 5th edition classification of myelodysplastic neoplasms (MDS) and the International Consensus Classification (ICC) of myeloid neoplasms and acute leukemia was published. In this study, we aimed to compare several differences between WHO 2022 and ICC classifications. Methods: Our study consisted of 989 newly-diagnosed primary MDS subjects classified using the WHO 2016 criteria from 08/2016 to 07/2022. CD34 and CD117 immunostaining were used in 30 of 38 subjects with low blasts and BM fibrosis to evaluate a proportion of immature myeloid precursors. All subjects were negative for CD117+ cells. CD34+ BM cells ranged from 0% to 5%. Results: According to WHO 2022 classification, 36 MDS subjects with NPM1 mutation were reclassified as AML and 10 MDS-U subjects were reclassified as clonal cytopenia of undetermined significance (CCUS). Applying to the ICC criteria, 20 MDS subjects with NPM1 mutation were reclassified as AML and 8 MDS-U subjects were reclassified as CCUS. We analyzed the remaining patients. Among the 81 subjects classified as MDS-biTP53 according to WHO 2022, 10 subjects did not meet ICC criteria for MDS or MDS/AML with mutated TP53 because the VAF of TP53 mutation(s) was less than 10%. Subjects not qualifying for MDS or MDS/AML with mutated TP53 had lower Revised International Prognostic Scoring System (IPSS-R) scores, better IPSS-R-defined cytogenetics, a higher frequency of SF3B1 mutations and longer survival compared with subjects with MDS or MDS/AML with mutated TP53 according to ICC (All P-values＜ 0.005, Figure 1 Left). Our data suggested that TP53 mutational burden should also be considered in the diagnostic classification and the cut-off value of 10% seems reasonable. An additional difference between these two classifications is that ≥15% ring sideroblasts may be a surrogate for SF3B1 mutation in WHO 2022. But only those with SF3B1 mutation (VAF≥10%), without 5q-, -7 or complex karyotype and RUNX1 mutation can be diagnosed as MDS with mutated SF3B1(MDS- SF3B1) according to ICC. 23 of 76 subjects (30.3%) with MDS- SF3B1 according to the WHO 2022 classification did not meet the ICC criteria for MDS- SF3B1. Subjects not meeting the ICC for MDS- SF3B1 had lower WBCs ( P = 0.003) ,platelet concentrations ( P = 0.001), higher IPSS-M categories ( P = 0.024) and shorter survival compared with those with MDS- SF3B1 according to ICC( P = 0.003, Figure 1 middle). Therefore, our study shown that SF3B1-unmutated MDS-RS cases have different clinical features and outcome in comparison with SF3B1-mutated subjects and ICC criteria defined a more homogeneous disease entity of MDS- SF3B1. Apart from MDS with defining genetic abnormalities, substantial differences in morphologically defined MDS subtypes between WHO 2022 and ICC also need to be addressed. According to the WHO 2022 classification, only MDS-EB patients were furtherly categorized into MDS-EB with or without BM fibrosis. Here, we explore the impact of BM fibrosis in patients with MDS-LB. Among MDS patients without excess blasts and specific genetic abnormalities, 38 patients had fibrosis. Subjects with low blasts and fibrosis (MDS-LBf) had lower platelet concentrations ( P = 0.002), worse IPSS-R-defined cytogenetics and IPSS-R categories ( P ＜0.05) compared with those without excess blasts and myelofibrosis. MDS-LBf had worse survival compared with those with low blasts and without fibrosis (Figure 1 right, P = 0.007). Conclusion: Both the WHO 2022 and ICC criteria allow more accurate diagnoses and risk stratification. However, the presence of two separate classifications leads to the dilemma in clinical application. When it comes to MDS with defining genetic abnormalities, our analyses suggest ICC defines more homogeneous disease entities of MDS-SF3B1, MDS or MDS/AML with mutated TP53 subjects compared with WHO 2022. Although WHO 2022 places an emphasis on morphologically defined MDS and listed hypoplastic MDS (MDS-h) and MDS with fibrosis (MDS-f) as new entities. Our data shown that fibrosis is a poor prognostic factor in MDS subjects without excess blasts. These subjects tend to have more aggressive clinical manifestations than other MDS subjects with low blasts. Collectively, we propose to integrate these two classifications into one pipeline for diagnosis of MDS (Figure 2), avoiding disadvantages for each one. Bing Liand Zhijian Xiao were co-corresponding authors

Analysis of the Impact of Whole Blood and T-Cell Donor Chimerism after Allogeneic Stem Cell Transplant

Background: Allogeneic stem cell transplant (alloSCT) is one of the few currently available therapies to cure acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS). Positive outcomes are dependent on engraftment of donor cells to provide hematopoiesis as well as a graft-versus-tumor effect. Measurement of whole blood (WB) and T-cell (TC) chimerism at routine intervals post alloSCT allows for monitoring of engraftment. Development of mixed donor chimerism (MC) may indicate impending graft failure or disease relapse. The ideal frequency for chimerism measurements as well as the role of WB versus TC chimerism remains unclear. In this study, we analyzed the impact of WB versus TC mixed chimerism on survival outcomes in patients with AML and MDS. Furthermore, we studied baseline factors associated with the development of MC. Methods: In this retrospective analysis, adults with AML or MDS who received an alloSCT between January 1, 2016 and February 1, 2022 that had at least 1 donor chimerism measured on Day 30 (D30), Day 60 (D60), Day 90 (D90) at Juravinski Hospital and Cancer Centre (Hamilton, ON, Canada). Baseline demographic and transplant-related characteristics were collected. AML and MDS diagnoses were based on WHO 2016 classification of myeloid neoplasm and acute leukemia. Conditioning intensities were classified according to CIBMTR definitions. A total of 271 transplants were completed during the study period for AML or MDS. 128 were excluded from the analysis (chimerism testing not performed (n=121), second transplants (n=7)). We analyzed a total of 143 patients with median age of 63 years (19 - 76) who were transplanted for AML (n=105) and MDS (n=38). Donors were HLA-matched siblings (MRD, n=26), unrelated donors (MUD (8/8 or 7/8), n=85) or haploidentical related donor (MMRD, n=32). All patients except one received peripheral blood stem cell grafts. For alloSCT, 20 % of patients received myeloablative conditioning (MAC) and 80% received reduced-intensity conditioning (RIC). Complete chimerism (CC) was defined as ≥95%, whereas MC was defined as ≤95%. All chimerism measurements were measured using a PowerPlex® 16 System (Promega) (Hamilton Regional Laboratory Medicine Program, Hamilton, ON). Results: In our cohort, OS at 2 years was 57% and RFS 49%. Of the baseline factors, only disease-risk based on ELN 2022 for RFS with a hazard ratio (HR) of 1.43, 95% CI [0.83, 2.49] for adverse risk and HR 0.53 95% CI [0.23, 1.03] (p=0.03) for favourable risk. WB MC was detected in 22%, 22%, 23% of patients on D30, D60, and D90, respectively. TC MC was detected in 48%, 46%, and 43% of patients on D30, D60, and D90, respectively. WB MC had a strong negative impact on RFS, with a HR of 2.81, 95% CI [1.69, 4.66; p<0.001] on D30, 3.07, 95% CI [1.67, 5.66; p<0.001] on D60 and 4.77, 95% CI [2.47, 9.22; p<0.001] on D90. The influence of TC MC on RFS was also significant at all time points. Of 101 patients who were alive and did not have a relapse prior to D90 and had D30 chimerism and D90 chimerism evaluated, the patients who had a mixed WB and TC chimerism on D90 had a HR of 6.13, 95% CI [2.23, 16.85] and HR 2.73, 95% CI [1.24, 5.98] for RFS, which was statistically significant even after adjusting for known information of D30 WB/TC chimerism status (Table 1). WB MC had a strong negative impact on OS with a HR of 2.67, 95% CI (1.54, 4.64; p=0.001) on D30, 2.81, 95% CI (1.47, 5.38; p=0.002) on D60 and 4.78, 95% CI (2.47, 9.24; p<0.001) on D90. Again, the later time points remained statistically significant, when adjusting for known D30 chimerism status. The influence of TC MC on OS was also significant at all time points, however only a non-significant trend remained at later time points when adjusting for D30 TC chimerism (Table 2). We analyzed baseline factors predictive for D30 chimerism. For WB chimerism, only donor source remained statistically significant (p=0.038) with MMRD having a lower HR for MC 0.39, 95% CI [0.11, 1.41] and MRD having a higher HR for MC 2.33, 95% CI [0.9 - 5.99]. Conclusions: Our data confirm that a mixed chimerism as early as D30 post-alloSCT can have a detrimental impact on outcomes (RFS and OS) in alloSCT. Patients receiving haploidentical alloSCT had a lower risk to develop mixed chimerism in our cohort. These dismal outcomes of patients with mixed chimerism in our cohort provides a strong rationale to study interventions such as novel maintenance strategies or donor-lymphocyte infusions in this patient group.

Bone and joint disease – Not always a rheumatological diagnosis

Abstract A 17-year-old female presented with a 7-day history of joint pains involving the bilateral knee, ankle and small joints of the hands and feet symmetrically, associated with bilateral knee joint swelling and early morning stiffness and bilateral red eyes. General examination was normal except for pallor. Musculoskeletal system examination revealed bilateral swollen and tender knee joints with restriction of movement along with tenderness in the bilateral shoulder, elbow, ankle joints and spine from T3 to T7 thoracic vertebral levels. Examination of other systems was normal. Investigations showed anaemia, elevated urea and creatinine. Skeletal survey showed multiple lytic lesions. Positron emission tomography revealed multiple punched-out lytic lesions in skull bones, scapulae, sternum, clavicles, ribs on both sides, vertebrae at multiple levels and in pelvic girdle bones. Few lesions showed mild 18fluorodeoxyglucose (18FDG) uptake, metaphyseal moth-eaten lysis in long bones with mild FDG uptake. Bone marrow biopsy and immunohistochemistry confirmed the diagnosis of acute B-cell lymphoblastic leukaemia. The terminology is according to 2016 revision of WHO classification of myeloid neoplasms and acute leukemia.

Childhood B-lymphoblastic leukemia: Integrating cytogenetics with other prognostic parameters and correlation with post-induction remission

BackgroundCytogenetics is irrefutably one of the most significant prognostic markers which have imparted ease and feasibility in the classification of acute leukemia. It is an essential prognostic tool in major treatment protocols for all hematological malignancies. This study aimed to note the frequency of different cytogenetic aberrations in children with B-ALL, their association with risk factors, and impact on post-induction remission status. MethodsA retrospective study was conducted in the Hematology department of The Indus Hospital, Karachi. All newly diagnosed patients of B-ALL (aged 1–17 years) from October 2, 2020 to October 13, 2021 were included. They were treated with modified Children's Oncology Group (COG) protocol. Cytogenetic studies including conventional karyotyping and fluorescence in situ hybridization (FISH) for BCR::ABL1, ETV6::RUNX1, and MLL gene rearrangement were performed on semi-automated cell imaging. ResultsAmong 238 patients, cytogenetic abnormalities were identified in 146 (61.3%) cases. A normal karyotype was observed in 92 (38.7%), high hyperdiploidy in 56 (24%), random structural abnormalities in 22 (9.2%), t (1:19) in 21 (8.8%), t (9; 22) in 12 (5.0%), complex karyotype in 10 (4.2%), 11q abnormalities in 5 (2.1%) and monosomy 7 in 1 (0.4%) case(s). FISH positivity for MLL, BCR::ABL1 and ETV6::RUNX1 was observed in 6 (2.5%), 18 (7.5%), and 20 (8.4%), cases, respectively. Post induction morphological remission was observed in 98.3% of cases. ConclusionThe overall frequency of cytogenetic abnormalities in the current cohort is comparable to previously reported studies. Our data also suggest the value of combining the karyotype and FISH techniques to increase the likelihood of detection of recurrent cytogenetic abnormalities. Even though there was a lack of correlation between genetic abnormalities and early outcome, a follow-up may be useful to ascertain the correlation with long-term survival.

Classification of Acute Lymphoblastic Leukemia based on White Blood Cell Images using InceptionV3 Model

Acute lymphoblastic leukemia (ALL) is the most common form of leukemia that occurs in children. Detection of ALL through white blood cell image analysis can help with the prognosis and appropriate treatment. In this study, the author proposes an approach to classifying ALL based on white blood cell images using a convolutional neural network (CNN) model called InceptionV3. The dataset used in this research consists of white blood cell images collected from patients with ALL and healthy individuals. These images were obtained from The Cancer Imaging Archive (TCIA), which is a service that stores large-scale cancer medical images available to the public. During the evaluation phase, the author used training data evaluation metrics such as accuracy and loss to measure the model's performance. The research results show that the InceptionV3 model is capable of classifying white blood cell images with a high level of accuracy. This model achieves an average ALL recognition accuracy of 0.9896 with a loss of 0.031. The use of CNN models such as InceptionV3 in medical image analysis has the potential to improve the efficiency and precision of image-based disease diagnosis.

Phenotypic clues that predict underlying cytogenetic/genetic abnormalities in myeloid malignancies: A contemporary review.

Precise subclassification of myeloid malignancies per the World Health Organization (WHO) classification system and the International Consensus Classification of Myeloid Neoplasms and Acute Leukaemias (ICC) requires investigation and documentation of the presence of cytogenetic and/or molecular genetic changes. These ancillary studies not only help in diagnosis, but also the prognosis of disease; however, they take time to be completed. In contrast, morphological evaluation of material from the blood and bone marrow specimens of cases where myeloid malignancies are suspected is usually completed quickly. Cytomorphological assessment may predict genetic changes and can be helpful in triaging acuity. This is especially true in haematological emergencies such as acute promyelocytic leukaemia (APL), where prompt APL-specific therapy can be life changing. Similarly, some morphological clues may help identify core binding factor leukaemias where a diagnosis of acute myeloid leukaemia (AML) could be rendered without reaching the 20% blast cutoff with immediate treatment-decision implications, or even a subset of cases of AML with FLT3 ITD/NPM1 mutation(s) which show characteristic features. Even though FISH/cytogenetics and/or PCR are still required for establishing the final diagnosis, evaluation for the presence of specific cytomorphological features that help predict genetic changes can be a useful tool to help guide early therapy.

DeepLeukNet—A CNN based microscopy adaptation model for acute lymphoblastic leukemia classification

DeepLeukNet—A CNN based microscopy adaptation model for acute lymphoblastic leukemia classification

Classification and genetics of pediatric B-other acute lymphoblastic leukemia by targeted RNA sequencing.

Acute lymphoblastic leukemia (ALL) can be classified into different subgroups based on recurrent genetic alterations. Here, targeted RNA sequencing was used to identify the novel subgroups of ALL in 144 B-other and 40 "classical" ALL samples. The classical TCF3-PBX1, ETV6-RUNX1, KMT2A-rearranged, and BCR-ABL1, and novel P2RY8-CRLF2, ABL-, JAK2-, ZNF384-, MEF2D-, and NUTM1-fusions were easily identified by fusion transcript analysis. IGH-CRLF2 and IGH-EPOR were found by abnormally high levels of expression of CRLF2 or EPOR. DUX4-rearranged was identified by the unusual expression of DUX4 genes and an alternative exon of ERG, or by clustering analysis of gene expression. PAX5-driven ALL, including fusions, intragenic amplifications, and mutations were identified by single-nucleotide variant analysis and manual inspection using the IGV software. Exon junction analysis allowed detection of some intragenic ERG and IKZF1 deletions. CRLF2-high associated with initial white blood cell (WBC) counts of ≥50× 103/μL and GATA3 risk alleles (rs3781093 and rs3824662), whereas ABL/JAK2/EPOR-fusions associated with high WBC counts, National Cancer Institute's high-risk classification, and IKZF1del. ZNF384-fusions associated with CALLA-negativity and NUTM1-fusions in infants. In conclusion, targeted RNA sequencing further classified 66.7% (96 of 144) B-other ALL cases. All BCP-ALL subgroups, except for iAMP21, hyperdiploid and hypodiploid cases, were identified. Curiously, we observed higher frequencies of females within B-rest ALLs and males in PAX5-driven cases.

Management of Patients with Lower-Risk Myelodysplastic Neoplasms (MDS).

Myelodysplastic neoplasms (MDS) are a heterogenous group of clonal hematologic disorders characterized by morphologic dysplasia, ineffective hematopoiesis, and cytopenia. In the past year, the classification of MDS has been updated in the 5th edition of the World Health Organization (WHO) Classification of Haematolymphoid Tumours and the International Consensus Classification (ICC) of Myeloid Neoplasms and Acute Leukemia with incorporation of morphologic, clinical, and genomic data. Furthermore, the more comprehensive International Prognostic Scoring System-Molecular (IPSS-M) allows for improved risk stratification and prognostication. These three developments allow for more tailored therapeutic decision-making in view of the expanding treatment options in MDS. For patients with lower risk MDS, treatment is aimed at improving cytopenias, usually anemia. The recent approval of luspatercept and decitabine/cedazuridine have added on to the current armamentarium of erythropoietic stimulating agents and lenalidomide (for MDS with isolated deletion 5q). Several newer agents are being evaluated in phase 3 clinical trials for this group of patients, such as imetelstat and oral azacitidine. This review provides a summary of the classification systems, the prognostic scores and clinical management of patients with lower risk MDS.

ResRandSVM: Hybrid Approach for Acute Lymphocytic Leukemia Classification in Blood Smear Images.

Acute Lymphocytic Leukemia is a type of cancer that occurs when abnormal white blood cells are produced in the bone marrow which do not function properly, crowding out healthy cells and weakening the immunity of the body and thus its ability to resist infections. It spreads quickly in children's bodies, and if not treated promptly it may lead to death. The manual detection of this disease is a tedious and slow task. Machine learning and deep learning techniques are faster than manual detection and more accurate. In this paper, a deep feature selection-based approach ResRandSVM is proposed for the detection of Acute Lymphocytic Leukemia in blood smear images. The proposed approach uses seven deep-learning models: ResNet152, VGG16, DenseNet121, MobileNetV2, InceptionV3, EfficientNetB0 and ResNet50 for deep feature extraction from blood smear images. After that, three feature selection methods are used to extract valuable and important features: analysis of variance (ANOVA), principal component analysis (PCA), and Random Forest. Then the selected feature map is fed to four different classifiers, Adaboost, Support Vector Machine, Artificial Neural Network and Naïve Bayes models, to classify the images into leukemia and normal images. The model performs best with a combination of ResNet50 as a feature extractor, Random Forest as feature selection and Support Vector Machine as a classifier with an accuracy of 0.900, precision of 0.902, recall of 0.957 and F1-score of 0.929.

Detection and Classification of Acute Lymphocytic Leukemia

Acute Leukemia is a life - threatening disease common both in children and adults that can lead to death if left un treated. Acute Lymphoblastic Leukemia (ALL) spreads out in children’s bodies rapidly and takes the life within a few weeks. To diagnose ALL, the hematologists perform blood and bone marrow examination. Manual blood testing techniques that have been used since long time are often slow and come out with the less accurate diagnosis. This work improves the diagnosis of ALL with a computer –aided system, which yields accurate result by using image processing and deep learning techniques. This research proposed a method for the classification of ALL into its sub types and reactive bone marrow (normal) in stained bone marrow images. Classification of ALL into its subtypes and reactive bone marrow (normal) in stained bone marrow images. A robust segmentation and deep learning techniques with the convolutional neural network are used to train the model on the bone marrow images to achieve accurate classification results. Experimental results reveal that the proposed method achieved more than 96% accuracy having four different classes. The obtained results exhibit that the proposed approach could be used as a tool to diagnose. Acute Lymphoblastic Leukemia and its sub-types that will definitely assist pathologists.

An Efficient Detection and Classification of Acute Leukemia Using Transfer Learning and Orthogonal Softmax Layer-Based Model.

For the early diagnosis of hematological disorders like blood cancer, microscopic analysis of blood cells is very important. Traditional deep CNNs lead to overfitting when it receives small medical image datasets such as ALLIDB1, ALLIDB2, and ASH. This paper proposes a new and effective model for classifying and detecting Acute Lymphoblastic Leukemia (ALL) or Acute Myelogenous Leukemia (AML) that delivers excellent performance in small medical datasets. Here, we have proposed a novel Orthogonal SoftMax Layer (OSL)-based Acute Leukemia detection model that consists of ResNet 18-based deep feature extraction followed by efficient OSL-based classification. Here, OSL is integrated with the ResNet18 to improve the classification performance by making the weight vectors orthogonal to each other. Hence, it integrates ResNet benefits (residual learning and identity mapping) with the benefits of OSL-based classification (improvement of feature discrimination capability and computational efficiency). Furthermore, we have introduced extra dropout and ReLu layers in the architecture to achieve a faster network with enhanced performance. The performance verification is performed on standard ALLIDB1, ALLIDB2, and C_NMC_2019 datasets for efficient ALL detection and ASH dataset for effective AML detection. The experimental performance demonstrates the superiority of the proposed model over other compairing models.

Emerging molecular subtypes and therapies in acute lymphoblastic leukemia.

Tremendous strides have been made in the molecular and cytogenetic classification of acute lymphoblastic leukemia based on gene expression profiling data, leading to an expansion of entities in the recent International Consensus Classification (ICC) of myeloid neoplasms and acute leukemias and 2022 WHO Classification of Tumours: Haematolymphoid Tumors, 5th edition. This increased diagnostic and therapeutic complexity can be overwhelming, and this review compares nomenclature differences between the ICC and WHO 5th edition publications, compiles key features of each entity, and provides a diagnostic algorithmic approach. In covering B-lymphoblastic leukemia (B-ALL), we divided the entities into established (those present in the revised 4th edition WHO) and novel (those added to either the ICC or WHO 5th edition) groups. The established B-ALL entities include B-ALL with BCR::ABL1 fusion, BCR::ABL1-like features, KMT2A rearrangement, ETV6::RUNX1 rearrangement, high hyperdiploidy, hypodiploidy (focusing on near haploid and low hypodiploid), IGH::IL3 rearrangement, TCF3::PBX1 rearrangement, and iAMP21. The novel B-ALL entities include B-ALL with MYC rearrangement; DUX4 rearrangement; MEF2D rearrangement; ZNF384 or ZNF362 rearrangement, NUTM1 rearrangement; HLF rearrangement; UBTF::ATXN7L3/PAN3,CDX2; mutated IKZF1 N159Y; mutated PAX5 P80R; ETV6::RUNX1-like features; PAX5 alteration; mutated ZEB2 (p.H1038R)/IGH::CEBPE; ZNF384 rearranged-like; KMT2A-rearranged-like; and CRLF2 rearrangement (non-Ph-like). Classification of T-ALL is complex with some variability in how the subtypes are defined in recent literature. It was classified as early T-precursor lymphoblastic leukemia/lymphoma and T-ALL, NOS in the WHO revised 4th edition and WHO 5th edition. The ICC added an entity into early T-cell precursor ALL, BCL11B-activated, and also added provisional entities subclassified based on transcription factor families that are aberrantly activated.

Neurofibromatosis Symptom-Lacking B-Cell Lineage Acute Lymphoblastic Leukemia with Only an NF1 Gene Pathogenic Variant

Next-generation sequencing technology has improved molecular genetic analysis, and many molecular genetic studies have been utilized for diagnostic classification, risk stratification, and prognosis prediction of acute lymphoblastic leukemia (ALL). Inactivation of neurofibromin or Nf1, a protein derived from the NF1 gene, causes Ras pathway regulation failure, which is related to leukemogenesis. Pathogenic variants of the NF1 gene in B-cell lineage ALL are uncommon, and in this study, we reported a pathogenic variant that is not registered in any public database. The patient diagnosed with B-cell lineage ALL had no clinical symptoms of neurofibromatosis. Studies on the biology, diagnosis, and treatment of this uncommon disease, as well as other related hematologic neoplasms, such as acute myeloid leukemia and juvenile myelomonocytic leukemia, were reviewed. Biological studies included epidemiological differences among age intervals and pathways for leukemia, such as the Ras pathway. Diagnostic studies included cytogenetic, FISH, and molecular tests for leukemia-related genes and ALL classification, such as Ph-like ALL or BCR-ABL1-like ALL. Treatment studies included pathway inhibitors and chimeric antigen cell receptor T-cells. Resistance mechanisms related to leukemia drugs were also investigated. We believe that these literature reviews will enhance medical care for the uncommon diagnosis of B-cell lineage ALL.

Abstract 2053: Structural variants detected by optical genome mapping in acute lymphoblastic leukemia patient-derived xenografts models

Abstract Background: The established classification of acute lymphoblastic leukemia (ALL) does not cover all high-risk patients because of the difficulty in detecting novel or rare structural variants (SVs). Furthermore, many SVs and formed gene fusions have adverse prognostic effects on hematological malignancies. This study aims to investigate SVs in 22 patient-derived xenografts (PDX) of ALL and to assess rare SVs and chromosomal aberrations, particularly hematological malignancy genes and gene fusions in different individuals detected by optical genome mapping (OGM). Methods: Leukemia cells from 16 patients with B-cell precursor acute lymphoblastic leukemia (B-ALL) and 6 patients with T-cell acute lymphoblastic leukemia (T-ALL) were inoculated into immuno-deficient mice, growth curves were monitored, and leukemia cells were collected after human CD45 was more than 80%. OGM (Bionano Genomics), RNA sequencing (RNA-seq) and whole genome sequencing (WGS) techniques were used to detect SVs and gene fusions. In validation, polymerase chain reaction (PCR) and fluorescence in situ hybridization (FISH) tests were performed to identify specific SVs. Results: On average, each model detected 42 rare SVs with known overlapping genes in different types by OGM: 12 insertions, 25 deletions, 1 inversion, 3 duplications, 2 inter-chromosomal translocations and 1 intra-chromosomal translocation. 5 rare SVs were found in multiple models (>=5), and two of them were never reported in B-ALL and T-ALL. Gene fusions were detected in all ALL models by OGM, and 17% were novel fusions. At the mRNA level, gene fusions were detected only in 19 models, and 24% of the predicted gene fusions by OGM were identical to the results of RNA-seq (high confidence results including pseudogenes). These two technologies could detect different fusion partners. For example, MLLT1 was seen at the mRNA level, and ACER1 was detected at the DNA level in the same KMT2A rearrangement model. Of the 16 B-ALL models, 6 were p190 BCR-ABL1 (a fusion of BCR exon 1 and ABL1 exon 2), 1 was p210 BCR-ABL1 (a fusion of BCR exon 13 and ABL1 exon 2) and 5 of 7 had IKZF1 gene deletions. 2 p190 BCR-ABL1 models were insensitive to imatinib and 2 p190 BCR-ABL1 models as well as the model with IKZF1 exons 4-7 deletion were sensitive to imatinib when treatment was initiated early in the disease. A novel ARL15-IKZF1 fusion was identified in one of the sensitive models. Deletion of the PRKAR2B gene was another rare SV in both models sensitive to imatinib besides the IKZF1 deletion. Conclusion: OGM assays were more comprehensive, and novel rare SVs were identified in ALL. Combining the results of gene fusions between different technologies will provide more accurate predictions for ALL classifications. The combination of aberrant SVs may synergize to influence the efficacy of imatinib, which warrants further investigations. Citation Format: Yueying Wang, Mengting Qin, JinPing Liu, Xiaobo Chen, Jia Xue, Sheng Guo. Structural variants detected by optical genome mapping in acute lymphoblastic leukemia patient-derived xenografts models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2053.