Machine learning demonstrates that somatic mutations imprint invariant morphologic features in myelodysplastic syndromes

Clinical and pathological features of clonal cytopenia of undetermined significance presenting with isolated thrombocytopenia (CCUS-IT)

BACKGROUND

Clonal cytopenia of undetermined significance (CCUS) is defined as somatic mutations of myeloid malignancy-associated genes in the blood or bone marrow with one or more persistent unexplained cytopenias that do not meet diagnostic criteria for a defined myeloid neoplasm. CCUS with isolated thrombocytopenia (CCUS-IT) is rare.

METHODS

This is a retrospective case series of patients with prolonged isolated thrombocytopenia, a pathogenic mutation on a myeloid molecular panel, and a bone marrow biopsy with morphologic atypia below the WHO-defined diagnostic threshold for dysplasia.

RESULTS

Five male patients were identified with a median age at CCUS-IT diagnosis of 61 years (56-74). Median duration of thrombocytopenia prior to CCUS-IT diagnosis was 4 years (3-12), and median platelet count at CCUS-IT diagnosis was 41 × 103 /μL (26-80). All patients had megakaryocytic hyperplasia and megakaryocytes with hyperchromasia and high nuclear-cytoplasmic ratio. Pathogenic SRSF2 mutations were identified in all 5 patients with median variant allele frequency of 36% (28%-50%). Three patients were treated with IVIg and/or steroids with no response; one of three responded to thrombopoietin receptor agonists. Three patients progressed to MDS and one to AML.

DISCUSSION

We describe the clinicopathological features of CCUS-IT which can mimic immune thrombocytopenia.

Diagnosis and classification of myelodysplastic syndromes

ABSTRACT

Myelodysplastic syndromes (MDSs) are neoplastic myeloid proliferations characterized by ineffective hematopoiesis resulting in peripheral blood cytopenias. MDS is distinguished from nonneoplastic clonal myeloid proliferations by the presence of morphologic dysplasia and from acute myeloid leukemia by a blast threshold of 20%. The diagnosis of MDS can be challenging because of the myriad other causes of cytopenias: accurate diagnosis requires the integration of clinical features with bone marrow and peripheral blood morphology, immunophenotyping, and genetic testing. MDS has historically been subdivided into several subtypes by classification schemes, the most recent of which are the International Consensus Classification and World Health Organization Classification (fifth edition), both published in 2022. The aim of MDS classification is to identify entities with shared genetic underpinnings and molecular pathogenesis, and the specific subtype can inform clinical decision-making alongside prognostic risk categorization. The current MDS classification schemes incorporate morphologic features (bone marrow and blood blast percentage, degree of dysplasia, ring sideroblasts, bone marrow fibrosis, and bone marrow hypocellularity) and also recognize 3 entities defined by genetics: isolated del(5q) cytogenetic abnormality, SF3B1 mutation, and TP53 mutation. It is anticipated that with advancing understanding of the genetic basis of MDS pathogenesis, future MDS classification will be based increasingly on genetic classes. Nevertheless, morphologic features in MDS reflect the phenotypic expression of the underlying abnormal genetic pathways and will undoubtedly retain importance to inform prognosis and guide treatment.

The consequences of cohesin mutations in myeloid malignancies

Recurrent somatic mutations in the genes encoding the chromatin-regulatory cohesin complex and its modulators occur in a wide range of human malignancies including a high frequency in myeloid neoplasms. The cohesin complex has a ring-like structure which can enclose two strands of DNA. A first function for the complex was described in sister chromatid cohesion during metaphase avoiding defects in chromosome segregation. Later studies identified additional functions of the cohesin complex functions in DNA replication, DNA damage response, 3D genome organisation, and transcriptional regulation through chromatin looping. In this review, we will focus on STAG2 which is the most frequently mutated cohesin subunit in myeloid malignancies. STAG2 loss of function mutations are not associated with chromosomal aneuploidies or genomic instability. We hypothesize that this points to changes in gene expression as disease-promoting mechanism and summarize the current state of knowledge on affected genes and pathways. Finally, we discuss potential strategies for targeting cohesion-deficient disease cells.

A machine learning tool for the diagnosis of SARS-CoV-2 infection from hemogram parameters

Monocytes and neutrophils play key roles in the cytokine storm triggered by SARS-CoV-2 infection, which changes their conformation and function. These changes are detectable at the cellular and molecular level and may be different to what is observed in other respiratory infections. Here, we applied machine learning (ML) to develop and validate an algorithm to diagnose COVID-19 using blood parameters. In this retrospective single-center study, 49 hemogram parameters from 12,321 patients with clinical suspicion of COVID-19 and tested by RT-PCR (4239 positive and 8082 negative) were analysed. The dataset was randomly divided into training and validation sets. Blood cell parameters and patient age were used to construct the predictive model with the support vector machine (SVM) tool. The model constructed from the training set (5936 patients) achieved an accuracy for diagnosis of SARS-CoV-2 infection of 0.952 (95% CI: 0.875-0.892). Test sensitivity and specificity was 0.868 and 0.899, respectively, with a positive (PPV) and negative (NPV) predictive value of 0.896 and 0.872, respectively (prevalence 0.50). The validation set model (4964 patients) achieved an accuracy of 0.894 (95% CI: 0.883-0.903). Test sensitivity and specificity was 0.8922 and 0.8951, respectively, with a positive (PPV) and negative (NPV) predictive value of 0.817 and 0.94, respectively (prevalence 0.34). The area under the receiver operating characteristic curve was 0.952 for the algorithm performance. This algorithm may allow to rule out COVID-19 diagnosis with 94% of probability. This represents a great advance for early diagnostic orientation and guiding clinical decisions.

Integrated Genomic and Transcriptomic Analysis Improves Disease Classification and Risk Stratification of MDS with Ring Sideroblasts

PURPOSE

Ring sideroblasts (RS) define the low-risk myelodysplastic neoplasm (MDS) subgroup with RS but may also reflect erythroid dysplasia in higher risk myeloid neoplasm. The benign behavior of MDS with RS (MDSRS+) is limited to SF3B1-mutated cases without additional high-risk genetic events, but one third of MDSRS+ carry no SF3B1 mutation, suggesting that different molecular mechanisms may underlie RS formation. We integrated genomic and transcriptomic analyses to evaluate whether transcriptome profiles may improve current risk stratification.

EXPERIMENTAL DESIGN

We studied a prospective cohort of MDSRS+ patients irrespective of World Health Organization (WHO) class with regard to somatic mutations, copy-number alterations, and bone marrow CD34+ cell transcriptomes to assess whether transcriptome profiles add to prognostication and provide input on disease classification.

RESULTS

SF3B1, SRSF2, or TP53 multihit mutations were found in 89% of MDSRS+ cases, and each mutation category was associated with distinct clinical outcome, gene expression, and alternative splicing profiles. Unsupervised clustering analysis identified three clusters with distinct hemopoietic stem and progenitor (HSPC) composition, which only partially overlapped with mutation groups. IPSS-M and the transcriptome-defined proportion of megakaryocyte/erythroid progenitors (MEP) independently predicted survival in multivariable analysis.

CONCLUSIONS

These results provide essential input on the molecular basis of SF3B1-unmutated MDSRS+ and propose HSPC quantification as a prognostic marker in myeloid neoplasms with RS.

The clinicopathologic significance of NPM1 mutation and ability to detect mutated NPM1 by immunohistochemistry in non-AML myeloid neoplasms

NPM1 mutated non-AML myeloid neoplasms (MN; <20% blasts) are characterized by an aggressive clinical course in a few studies. In this retrospective study, we evaluate the clinicopathologic and immunohistochemical features of non-AML MN patients with NPM1 mutations. We assessed NPM1 mutation by targeted next generation sequencing (NGS). Cytoplasmic NPM1 expression was assessed by immunohistochemistry (IHC) on formalin-fixed, formic acid-decalcified bone marrow biopsy specimens. We evaluated 34 non-AML MN patients with NPM1 mutations comprising MDS (22), MPN (3) and MDS/MPN (9). They commonly presented with anemia (88%), thrombocytopenia (58%) and leukopenia (50%). Bone marrow dysplasia was common (79%). The karyotype was often normal (64%). NGS for MN-associated mutations performed in a subset of the patients showed a median of 3 mutations. NPM1 mutations were more often missense (c.859C > T p. L287F; 65%) than frameshift insertion/duplication (35%) with median variant allele frequency (VAF; 9.7%, range 5.1%-49.8%). Mutated NPM1 by IHC showed cytoplasmic positivity in 48% and positivity was associated with higher VAF. The median overall survival (OS) in this cohort was 70 months. Nine patients (26%) progressed to AML. OS in patients who progressed to AML was significantly shorter than the one of patients without progression to AML (OS 20 vs. 128 months, respectively, log rank p = 0.05). NPM1 mutated non-AML MN patients commonly had cytopenias, dysplasia, normal karyotype, mutations in multiple genes, and an unfavorable clinical outcome, including progression to AML. Our data demonstrated that IHC for NPM1 can be a useful supplementary tool to predict NPM1 mutation in some non-AML MN; however, genetic testing cannot be replaced by IHC assessment.

Artificial intelligence to empower diagnosis of myelodysplastic syndromes by multiparametric flow cytometry

The diagnosis of myelodysplastic syndromes (MDS) might be challenging and relies on the convergence of cytological, cytogenetic, and molecular factors. Multiparametric flow cytometry (MFC) helps diagnose MDS, especially when other features do not contribute to the decision-making process, but its usefulness remains underestimated, mostly due to a lack of standardization of cytometers. We present here an innovative model integrating artificial intelligence (AI) with MFC to improve the diagnosis and the classification of MDS. We develop a machine learning model through an elasticnet algorithm directed on a cohort of 191 patients, only based on flow cytometry parameters selected by the Boruta algorithm, to build a simple but reliable prediction score with five parameters. Our AI-assisted MDS prediction score greatly improves the sensitivity of the Ogata score while keeping an excellent specificity validated on an external cohort of 89 patients with an Area Under the Curve of 0.935. This model allows the diagnosis of both high- and low-risk MDS with 91.8% sensitivity and 92.5% specificity. Interestingly, it highlights a progressive evolution of the score from clonal hematopoiesis of indeterminate potential (CHIP) to highrisk MDS, suggesting a linear evolution between these different stages. By significantly decreasing the overall misclassification of 52% for patients with MDS and of 31.3% for those without MDS (P=0.02), our AI-assisted prediction score outperforms the Ogata score and positions itself as a reliable tool to help diagnose MDS.

Computational analysis of peripheral blood smears detects disease-associated cytomorphologies

Many hematological diseases are characterized by altered abundance and morphology of blood cells and their progenitors. Myelodysplastic syndromes (MDS), for example, are a group of blood cancers characterised by cytopenias, dysplasia of hematopoietic cells and blast expansion. Examination of peripheral blood slides (PBS) in MDS often reveals changes such as abnormal granulocyte lobulation or granularity and altered red blood cell (RBC) morphology; however, some of these features are shared with conditions such as haematinic deficiency anemias. Definitive diagnosis of MDS requires expert cytomorphology analysis of bone marrow smears and complementary information such as blood counts, karyotype and molecular genetics testing. Here, we present Haemorasis, a computational method that detects and characterizes white blood cells (WBC) and RBC in PBS. Applied to over 300 individuals with different conditions (SF3B1-mutant and SF3B1-wildtype MDS, megaloblastic anemia, and iron deficiency anemia), Haemorasis detected over half a million WBC and millions of RBC and characterized their morphology. These large sets of cell morphologies can be used in diagnosis and disease subtyping, while identifying novel associations between computational morphotypes and disease. We find that hypolobulated neutrophils and large RBC are characteristic of SF3B1-mutant MDS. Additionally, while prevalent in both iron deficiency and megaloblastic anemia, hyperlobulated neutrophils are larger in the latter. By integrating cytomorphological features using machine learning, Haemorasis was able to distinguish SF3B1-mutant MDS from other MDS using cytomorphology and blood counts alone, with high predictive performance. We validate our findings externally, showing that they generalize to other centers and scanners. Collectively, our work reveals the potential for the large-scale incorporation of automated cytomorphology into routine diagnostic workflows.

Co-occurring mutations in ASXL1, SRSF2, and SETBP1 define a subset of myelodysplastic/ myeloproliferative neoplasm with neutrophilia

Identification of genomic signatures with consistent clinicopathological features in myelodysplastic/myeloproliferative neoplasm (MDS/MPN) is critical for improved diagnosis, elucidation of biology, inclusion in clinical trials, and development of therapies. We describe clinical and pathological features with co-existence of mutations in ASXL1 (missense or nonsense), SRSF2, and SKI homologous region of SETBP1, in 18 patients. Median age was 68 years with a male predominance (83%). Leukocytosis and neutrophilia were common at presentation. Marrow features included hypercellularity, granulocytic hyperplasia with megakaryocytic atypia, while the majority had myeloid hyperplasia and/or erythroid hypoplasia, myeloid dysplasia, and aberrant CD7 expression on blasts. Mutations in growth signaling pathways (RAS or JAK2) were noted at diagnosis or acquired during the disease course in 83% of patients. Two patients progressed upon acquisition of FLT3-TKD (acute myeloid leukemia) or KIT (aggressive systemic mastocytosis) mutations. The prognosis is poor with only two long-term survivors, thus far, who underwent blood or marrow transplantation. We propose that the presence of co-occurring ASXL1, SRSF2, and SETBP1 mutations can be diagnostic of a subtype of MDS/MPN with neutrophilia if clinical and morphological findings align. Our report underscores the association between genotype and phenotype within MDS/MPN and that genomic signatures should guide categorization of these entities.

Leukemia relapse via genetic immune escape after allogeneic hematopoietic cell transplantation

Graft-versus-leukemia (GvL) reactions are responsible for the effectiveness of allogeneic hematopoietic cell transplantation as a treatment modality for myeloid neoplasia, whereby donor T- effector cells recognize leukemia neoantigens. However, a substantial fraction of patients experiences relapses because of the failure of the immunological responses to control leukemic outgrowth. Here, through a broad immunogenetic study, we demonstrate that germline and somatic reduction of human leucocyte antigen (HLA) heterogeneity enhances the risk of leukemic recurrence. We show that preexistent germline-encoded low evolutionary divergence of class II HLA genotypes constitutes an independent factor associated with disease relapse and that acquisition of clonal somatic defects in HLA alleles may lead to escape from GvL control. Both class I and II HLA genes are targeted by somatic mutations as clonal selection factors potentially impairing cellular immune responses and response to immunomodulatory strategies. These findings define key molecular modes of post-transplant leukemia escape contributing to relapse.

Molecular patterns identify distinct subclasses of myeloid neoplasia

Genomic mutations drive the pathogenesis of myelodysplastic syndromes and acute myeloid leukemia. While morphological and clinical features have dominated the classical criteria for diagnosis and classification, incorporation of molecular data can illuminate functional pathobiology. Here we show that unsupervised machine learning can identify functional objective molecular clusters, irrespective of anamnestic clinico-morphological features, despite the complexity of the molecular alterations in myeloid neoplasia. Our approach reflects disease evolution, informed classification, prognostication, and molecular interactions. We apply machine learning methods on 3588 patients with myelodysplastic syndromes and secondary acute myeloid leukemia to identify 14 molecularly distinct clusters. Remarkably, our model shows clinical implications in terms of overall survival and response to treatment even after adjusting to the molecular international prognostic scoring system (IPSS-M). In addition, the model is validated on an external cohort of 412 patients. Our subclassification model is available via a web-based open-access resource ( https://drmz.shinyapps.io/mds_latent ).

Differential diagnosis of bone marrow failure syndromes guided by machine learning

The choice to postpone treatment while awaiting genetic testing can result in significant delay in definitive therapies in patients with severe pancytopenia. Conversely, the misdiagnosis of inherited bone marrow failure (BMF) can expose patients to ineffectual and expensive therapies, toxic transplant conditioning regimens, and inappropriate use of an affected family member as a stem cell donor. To predict the likelihood of patients having acquired or inherited BMF, we developed a 2-step data-driven machine-learning model using 25 clinical and laboratory variables typically recorded at the initial clinical encounter. For model development, patients were labeled as having acquired or inherited BMF depending on their genomic data. Data sets were unbiasedly clustered, and an ensemble model was trained with cases from the largest cluster of a training cohort (n = 359) and validated with an independent cohort (n = 127). Cluster A, the largest group, was mostly immune or inherited aplastic anemia, whereas cluster B comprised underrepresented BMF phenotypes and was not included in the next step of data modeling because of a small sample size. The ensemble cluster A-specific model was accurate (89%) to predict BMF etiology, correctly predicting inherited and likely immune BMF in 79% and 92% of cases, respectively. Our model represents a practical guide for BMF diagnosis and highlights the importance of clinical and laboratory variables in the initial evaluation, particularly telomere length. Our tool can be potentially used by general hematologists and health care providers not specialized in BMF, and in under-resourced centers, to prioritize patients for genetic testing or for expeditious treatment.

Leukemia relapse via genetic immune escape after allogeneic hematopoietic cell transplantation

Graft-versus-leukemia (GvL) reactions are responsible for the effectiveness of allogeneic hematopoietic cell transplantation as a treatment modality for myeloid neoplasia, whereby donor T- effector cells recognize leukemia neoantigens. However, a substantial fraction of patients experience relapses because of the failure of the immunological responses to control leukemic outgrowth. Here, through a broad immunogenetic study, we demonstrate that germline and somatic reduction of human leucocyte antigen (HLA) heterogeneity enhances the risk of leukemic recurrence. We show that preexistent germline-encoded low evolutionary divergence of class II HLA genotypes constitutes an independent factor associated with disease relapse and that acquisition of clonal somatic defects in HLA alleles may lead to escape from GvL control. Both class I and II HLA genes are targeted by somatic mutations as clonal selection factors potentially impairing cellular immune reactions and response to immunomodulatory strategies. These findings define key molecular modes of post-transplant leukemia escape contributing to relapse.

A multimodal analysis of genomic and RNA splicing features in myeloid malignancies

RNA splicing dysfunctions are more widespread than what is believed by only estimating the effects resulting by splicing factor mutations (SF^MT) in myeloid neoplasia (MN). The genetic complexity of MN is amenable to machine learning (ML) strategies. We applied an integrative ML approach to identify co-varying features by combining genomic lesions (mutations, deletions, and copy number), exon-inclusion ratio as measure of RNA splicing (percent spliced in, PSI), and gene expression (GE) of 1,258 MN and 63 normal controls. We identified 15 clusters based on mutations, GE, and PSI. Different PSI levels were present at various extents regardless of SF^MT suggesting that changes in RNA splicing were not strictly related to SF^MT. Combination of PSI and GE further distinguished the features and identified PSI similarities and differences, common pathways, and expression signatures across clusters. Thus, multimodal features can resolve the complex architecture of MN and help identifying convergent molecular and transcriptomic pathways amenable to therapies.

Comparison of the International Consensus and 5th WHO edition classifications of adult myelodysplastic syndromes and acute myeloid leukemia

Several editions of the World Health Organization (WHO) classifications of lympho-hemopoietic neoplasms in 2001, 2008, and 2016 served as the international standard for diagnosis. Since the 4th WHO edition, here referred as WHO-HAEM4, significant clinico-pathological, immunophenotypic, and molecular advances have been made in the field of myeloid neoplasms, which have contributed to refine diagnostic criteria, to upgrade entities previously defined as provisional and to identify new entities. This process has resulted in two recent classification proposals of myeloid neoplasms: the International Consensus Classification (ICC) and the 5th edition of the WHO classification (WHO-HAEM5). In this paper, we review and compare the two classifications in terms of diagnostic criteria and entity definition, with a focus on adult myelodysplastic syndromes/neoplasms (MDS) and acute myeloid leukemia (AML). The goal is to provide a tool to facilitate the work of pathologists, hematologists and researchers involved in the diagnosis and treatment of these hematological malignancies.

Artificial intelligence in hematological diagnostics: Game changer or gadget?

The future of clinical diagnosis and treatment of hematologic diseases will inevitably involve the integration of artificial intelligence (AI)-based systems into routine practice to support the hematologists' decision making. Several studies have shown that AI-based models can already be used to automatically differentiate cells, reliably detect malignant cell populations, support chromosome banding analysis, and interpret clinical variants, contributing to early disease detection and prognosis. However, even the best tool can become useless if it is misapplied or the results are misinterpreted. Therefore, in order to comprehensively judge and correctly apply newly developed AI-based systems, the hematologist must have a basic understanding of the general concepts of machine learning. In this review, we provide the hematologist with a comprehensive overview of various machine learning techniques, their current implementations and approaches in different diagnostic subfields (e.g., cytogenetics, molecular genetics), and the limitations and unresolved challenges of the systems.

The Heterogeneous Complexity of Myeloid Neoplasm: Multi-Level Approaches to Study the Disease

Myeloid neoplasms (MNs) include a spectrum of bone marrow malignancies that result from the clonal expansion and arrest of differentiation of myeloid progenitor cells [...].

The Application of Morphogo in the Detection of Megakaryocytes from Bone Marrow Digital Images with Convolutional Neural Networks

The evaluation of megakaryocytes is an important part of the work up on bone marrow smear examination. It has significance in the differential diagnosis, therapeutic efficacy assessment, and predication of prognosis of many hematologic diseases. The process of manual identification of megakaryocytes are tedious and lack of reproducibility; therefore, a reliable method of automated megakaryocytic identification is urgently needed. Three hundred and thirty-three bone marrow aspirate smears were digitized by Morphogo system. Pathologists annotated megakaryocytes on the digital images of marrow smears are applied to construct a large dataset for testing the system's predictive performance. Subsequently, we obtained megakaryocyte count and classification for each sample by different methods (system-automated analysis, system-assisted analysis, and microscopic examination) to study the correlation between different counting and classification methods. Morphogo system localized cells likely to be megakaryocytes on digital smears, which were later annotated by pathologists and the system, respectively. The system showed outstanding performance in identifying megakaryocytes in bone marrow smears with high sensitivity (96.57%) and specificity (89.71%). The overall correlation between the different methods was confirmed the high consistency (r ≥ 0.7218, R² ≥ 0.5211) with microscopic examination in classifying megakaryocytes. Morphogo system was proved as a reliable screen tool for analyzing megakaryocytes. The application of Morphogo system shows promises to advance the automation and standardization of bone marrow smear examination.

The International Consensus Classification of myelodysplastic syndromes and related entities

The International Consensus Classification (ICC) of myeloid neoplasms and acute leukemia has updated the classification of myelodysplastic syndromes (MDSs) and placed MDS in a broader group of clonal cytopenias that includes clonal cytopenia of undetermined significance (CCUS) and related entities. Although subject to some interobserver variability and lack of specificity, morphologic dysplasia remains the main feature that distinguishes MDS from other clonal cytopenias and defines MDS as a hematologic malignancy. The ICC has introduced some changes in the definition of MDS whereby some cases categorized as MDS based on cytogenetic abnormalities are now classified as CCUS, while SF3B1 and multi-hit TP53 mutations are now considered to be MDS-defining in a cytopenic patient. The ICC has also recognized several cytogenetic and molecular abnormalities that reclassify some cases of MDS with excess blasts as acute myeloid leukemia (AML) and has introduced a new MDS/AML entity that encompasses cases with 10-19% blasts that lie on the continuum between MDS and AML. Two new genetically defined categories of MDS have been introduced: MDS with mutated SF3B1 and MDS with mutated TP53, the latter requiring bi-allelic aberrations in the TP53 gene. The entity MDS, unclassifiable has been eliminated. These changes have resulted in an overall simplification of the MDS classification scheme from 8 separate entities (including 1 that was genetically defined) in the revised 4th edition WHO classification to 7 separate entities (including 3 that are genetically defined) in the ICC.

Advances in myelodysplastic syndromes: promising novel agents and combination strategies

INTRODUCTION

Myelodysplastic syndromes (MDS) are heterogeneous group of clonal hematopoietic stem cell neoplasms that have limited approved treatment options. Multiple novel agents are currently being tested in a clinical trial setting. From a therapeutic perspective, MDS is generally divided into lower-risk and higher-risk disease. In this review, we summarize some of the most prominent novel agents currently in development.

AREAS COVERED

This review focuses on select clinical trials in both lower- and higher-risk MDS, elucidating the mechanisms of action and rationale for drug combinations and summarizing early safety and efficacy data using novel agents in MDS.

EXPERT OPINION

Advances in understanding the innate immune system, telomere biology, as well as genomic drivers of the disease have led to the development of multiple novel agents that are currently in late stages of clinical development in MDS. Imetelstat is being tested in lower-risk disease and the phase III clinical trial recently completed accrual. Magrolimab, sabatolimab, and venetoclax in addition to novel oral hypomethylating agents (HMA) are being investigated in higher-risk MDS. These advances will hopefully bring better treatment options to patients and lead to a shift in the treatment paradigm. Post HMA therapy remains an area of dire unmet need.

Genetic mutations associated with blood count abnormalities in myeloid neoplasms

INTRODUCTION

Myelodysplastic syndromes (MDS) predominantly present with varying degrees of cytopenia, while myelodysplastic syndromes/myeloproliferative neoplasms (MDS/MPN) exhibit proliferative features. Genetic defects underlying different complete blood count (CBC) alterations remain to be defined.

OBJECTIVE

We aimed to evaluate mutations and impacts on abnormal blood counts in MDS and MDS/MPN.

METHOD

MDS and MDS/MPN patients were recruited and sequenced by targeted next-generation sequencing. Clinical parameters, especially CBC, were evaluated for the association with genetic abnormalities and clinical outcomes.

RESULTS

A total of 168 patients with myeloid neoplasms were recruited (92 cases of low-risk MDS, 57 cases of high-risk MDS and 19 cases of MDS/MPN). Compared to low-risk MDS and MDS/MPN, patients with high-risk MDS were presented with more severe neutropenia with 17.5% showing absolute neutrophil counts (ANC) lower than 0.5 × 10⁹/L. Patients with MDS/MPN more commonly harboured mutations and had a higher number of mutations per case than low-risk MDS (94.7% vs. 56.5%; p < 0.001 and 3 vs. 1; p < 0.001, respectively). Patients with SF3B1 mutations showed lower haemoglobin levels than wild-type (7.9 vs. 8.4 g/dL, p = 0.02), but were associated with normal platelet counts (286 vs. 93 × 10⁹/L; p < 0.001). Patients with U2AF1 mutations were associated with more severe leukopenia than wild-type (3 vs. 4.18 × 10⁹/L; p = 0.02). KRAS mutations were associated with monocytosis (p < 0.001). Multivariate analysis revealed high-risk MDS, MDS/MPN, severe neutropenia (ANC < 0.5 × 10⁹/L), and mutations in ASXL1 and SETBP1 were associated with inferior survival outcomes.

CONCLUSION

Certain mutations were related to more severe anaemia, lower white blood cell count or monocytosis in Asian MDS and MDS/MPN patients.

Precision Medicine in Myeloid Malignancies: Hype or Hope?

PURPOSE OF REVIEW

We review how understanding the fitness and comorbidity burden of patients, and molecular landscape of underlying acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS) at the time of diagnosis is now integral to treatment.

RECENT FINDINGS

The upfront identification of patients' fitness and molecular profile facilitates selection of targeted and novel agents, enables risk stratification, allows consideration of allogeneic hematopoietic cell transplantation in high-risk patients, and provides treatment selection for older (age ≥ 75) or otherwise unfit patients who may not tolerate conventional treatment. The use of measurable residual disease (MRD) assessment improves outcome prediction and can also guide therapeutic strategies such as chemotherapy maintenance and transplant. In recent years, several novel drugs have received FDA approval for treating patients with AML with or without specific mutations. A doublet and triplet combination of molecular targeted and other novel treatments have resulted in high response rates in early trials. Following the initial success in AML, novel drugs are undergoing clinical trials in MDS. Unprecedented advances have been made in precision medicine approaches in AML and MDS. However, lack of durable responses and long-term disease control in many patients still present significant challenges, which can only be met, to some extent, with innovative combination strategies throughout the course of treatment from induction to consolidation and maintenance.

Myelodysplastic syndromes

Myelodysplastic syndromes (MDS) are a family of myeloid cancers with diverse genotypes and phenotypes characterized by ineffective haematopoiesis and risk of transformation to acute myeloid leukaemia (AML). Some epidemiological data indicate that MDS incidence is increasing in resource-rich regions but this is controversial. Most MDS cases are caused by randomly acquired somatic mutations. In some patients, the phenotype and/or genotype of MDS overlaps with that of bone marrow failure disorders such as aplastic anaemia, paroxysmal nocturnal haemoglobinuria (PNH) and AML. Prognostic systems, such as the revised International Prognostic Scoring System (IPSS-R), provide reasonably accurate predictions of survival at the population level. Therapeutic goals in individuals with lower-risk MDS include improving quality of life and minimizing erythrocyte and platelet transfusions. Therapeutic goals in people with higher-risk MDS include decreasing the risk of AML transformation and prolonging survival. Haematopoietic cell transplantation (HCT) can cure MDS, yet fewer than 10% of affected individuals receive this treatment. However, how, when and in which patients with HCT for MDS should be performed remains controversial, with some studies suggesting HCT is preferred in some individuals with higher-risk MDS. Advances in the understanding of MDS biology offer the prospect of new therapeutic approaches.

International Consensus Classification of Myeloid Neoplasms and Acute Leukemias: integrating morphologic, clinical, and genomic data

The classification of myeloid neoplasms and acute leukemias was last updated in 2016 within a collaboration between the World Health Organization (WHO), the Society for Hematopathology, and the European Association for Haematopathology. This collaboration was primarily based on input from a clinical advisory committees (CACs) composed of pathologists, hematologists, oncologists, geneticists, and bioinformaticians from around the world. The recent advances in our understanding of the biology of hematologic malignancies, the experience with the use of the 2016 WHO classification in clinical practice, and the results of clinical trials have indicated the need for further revising and updating the classification. As a continuation of this CAC-based process, the authors, a group with expertise in the clinical, pathologic, and genetic aspects of these disorders, developed the International Consensus Classification (ICC) of myeloid neoplasms and acute leukemias. Using a multiparameter approach, the main objective of the consensus process was the definition of real disease entities, including the introduction of new entities and refined criteria for existing diagnostic categories, based on accumulated data. The ICC is aimed at facilitating diagnosis and prognostication of these neoplasms, improving treatment of affected patients, and allowing the design of innovative clinical trials.

A Review of Artificial Intelligence Applications in Hematology Management: Current Practices and Future Prospects

Background

Machine learning (ML) and deep learning (DL) methods have recently garnered a great deal of attention in the field of cancer research by making a noticeable contribution to the growth of predictive medicine and modern oncological practices. Considerable focus has been particularly directed toward hematologic malignancies because of the complexity in detecting early symptoms. Many patients with blood cancer do not get properly diagnosed until their cancer has reached an advanced stage with limited treatment prospects. Hence, the state-of-the-art revolves around the latest artificial intelligence (AI) applications in hematology management.

Objective

This comprehensive review provides an in-depth analysis of the current AI practices in the field of hematology. Our objective is to explore the ML and DL applications in blood cancer research, with a special focus on the type of hematologic malignancies and the patient’s cancer stage to determine future research directions in blood cancer.

Methods

We searched a set of recognized databases (Scopus, Springer, and Web of Science) using a selected number of keywords. We included studies written in English and published between 2015 and 2021. For each study, we identified the ML and DL techniques used and highlighted the performance of each model.

Results

Using the aforementioned inclusion criteria, the search resulted in 567 papers, of which 144 were selected for review.

Conclusions

The current literature suggests that the application of AI in the field of hematology has generated impressive results in the screening, diagnosis, and treatment stages. Nevertheless, optimizing the patient’s pathway to treatment requires a prior prediction of the malignancy based on the patient’s symptoms or blood records, which is an area that has still not been properly investigated.

Classification of acute myeloid leukemia M1 and M2 subtypes using machine learning

BACKGROUND

Classification of acute myeloid leukemia (AML) relies on manual analysis of bone marrow or peripheral blood smear images. We aimed to construct a machine learning model for automatic classification of AML-M1 and M2 subtypes in bone marrow smear images.

METHODS

Bone marrow smear images of AML patients were extracted from the Cancer Imaging Archive (TCIA) open database. Classification criteria of AML subtypes were based on the French-American-British (FAB) classification system. Random forest method and broad learning system (BLS) were used to develop the classification model. Morphological features, radiomics features, and clinical features were extracted. The performance of the classification model was evaluated by calculating accuracy, precision, recall, F1-score, and area under the curve (AUC). A total of 50 bone marrow smear images (AML-M1, 31 cases; AML-M2, 19 cases) with 500 slices were included in this study.

RESULTS

A total of 43 morphological features, 276 radiomics features, and 1 clinical feature were extracted. Finally, 9 variables including 2 morphological features, 6 radiomics features, and 1 clinical feature were selected into the classification model. The best classification performance was observed in the random forest model with 9 variables, with the average accuracy, AUC, F1-score, recall, and precision of the model being 0.998 ± 0.003, 0.998 ± 0.004, 0.998 ± 0.004, 0.996 ± 0.009, and 1 ± 0, respectively.

CONCLUSION

The random forest model performed well for the classification of AML-M1 and M2, which may provide a tool for clinicians to classify AML-M1 and M2.

RUNX1 mutations contribute to the progression of MDS due to disruption of antitumor cellular defense: a study on patients with lower-risk MDS

Patients with lower-risk myelodysplastic syndromes (LR-MDS) have a generally favorable prognosis; however, a small proportion of cases progress rapidly. This study aimed to define molecular biomarkers predictive of LR-MDS progression and to uncover cellular pathways contributing to malignant transformation. The mutational landscape was analyzed in 214 LR-MDS patients, and at least one mutation was detected in 137 patients (64%). Mutated RUNX1 was identified as the main molecular predictor of rapid progression by statistics and machine learning. To study the effect of mutated RUNX1 on pathway regulation, the expression profiles of CD34 + cells from LR-MDS patients with RUNX1 mutations were compared to those from patients without RUNX1 mutations. The data suggest that RUNX1-unmutated LR-MDS cells are protected by DNA damage response (DDR) mechanisms and cellular senescence as an antitumor cellular barrier, while RUNX1 mutations may be one of the triggers of malignant transformation. Dysregulated DDR and cellular senescence were also observed at the functional level by detecting γH2AX expression and β-galactosidase activity. Notably, the expression profiles of RUNX1-mutated LR-MDS resembled those of higher-risk MDS at diagnosis. This study demonstrates that incorporating molecular data improves LR-MDS risk stratification and that mutated RUNX1 is associated with a suppressed defense against LR-MDS progression.

A Phenogenetic Axis that Modulates Clinical Manifestation and Predicts Treatment Outcome in Primary Myeloid Neoplasms

Although the concept of "myeloid neoplasm continuum" has long been proposed, few comparative genomics studies directly tested this hypothesis. Here we report a multi-modal data analysis of 730 consecutive newly diagnosed patients with primary myeloid neoplasm, along with 462 lymphoid neoplasm cases serving as the outgroup. Our study identified a "Pan-Myeloid Axis" along which patients, genes, and phenotypic features were all aligned in sequential order. Utilizing relational information of gene mutations along the Pan-Myeloid Axis improved prognostic accuracy for complete remission and overall survival in adult patients of de novo acute myeloid leukemia and for complete remission in adult patients of myelodysplastic syndromes with excess blasts. We submit that better understanding of the myeloid neoplasm continuum might shed light on how treatment should be tailored to individual diseases.

Significance

The current criteria for disease diagnosis treat myeloid neoplasms as a group of distinct, separate diseases. This work provides genomics evidence for a "myeloid neoplasm continuum" and suggests that boundaries between myeloid neoplastic diseases are much more blurred than previously thought.

Personalized Risk Schemes and Machine Learning to Empower Genomic Prognostication Models in Myelodysplastic Syndromes

Myelodysplastic syndromes (MDS) are characterized by variable clinical manifestations and outcomes. Several prognostic systems relying on clinical factors and cytogenetic abnormalities have been developed to help stratify MDS patients into different risk categories of distinct prognoses and therapeutic implications. The current abundance of molecular information poses the challenges of precisely defining patients' molecular profiles and their incorporation in clinically established diagnostic and prognostic schemes. Perhaps the prognostic power of the current systems can be boosted by incorporating molecular features. Machine learning (ML) algorithms can be helpful in developing more precise prognostication models that integrate complex genomic interactions at a higher dimensional level. These techniques can potentially generate automated diagnostic and prognostic models and assist in advancing personalized therapies. This review highlights the current prognostication models used in MDS while shedding light on the latest achievements in ML-based research.

Machine learning identifies the independent role of dysplasia in the prediction of response to chemotherapy in AML

The independent prognostic impact of specific dysplastic features in acute myeloid leukemia (AML) remains controversial and may vary between genomic subtypes. We apply a machine learning framework to dissect the relative contribution of centrally reviewed dysplastic features and oncogenetics in 190 patients with de novo AML treated in ALFA clinical trials. One hundred and thirty-five (71%) patients achieved complete response after the first induction course (CR). Dysgranulopoiesis, dyserythropoiesis and dysmegakaryopoiesis were assessable in 84%, 83% and 63% patients, respectively. Multi-lineage dysplasia was present in 27% of assessable patients. Micromegakaryocytes (q = 0.01), hypolobulated megakaryocytes (q = 0.08) and hyposegmented granulocytes (q = 0.08) were associated with higher ELN-2017 risk. Using a supervised learning algorithm, the relative importance of morphological variables (34%) for the prediction of CR was higher than demographic (5%), clinical (2%), cytogenetic (25%), molecular (29%), and treatment (5%) variables. Though dysplasias had limited predictive impact on survival, a multivariate logistic regression identified the presence of hypolobulated megakaryocytes (p = 0.014) and micromegakaryocytes (p = 0.035) as predicting lower CR rates, independently of monosomy 7 (p = 0.013), TP53 (p = 0.004), and NPM1 mutations (p = 0.025). Assessment of these specific dysmegakarypoiesis traits, for which we identify a transcriptomic signature, may thus guide treatment allocation in AML.

Is There a Need for a More Precise Description of Biomolecule Interactions to Understand Cell Function?

An important goal of biological research is to explain and hopefully predict cell behavior from the molecular properties of cellular components. Accordingly, much work was done to build extensive “omic” datasets and develop theoretical methods, including computer simulation and network analysis to process as quantitatively as possible the parameters contained in these resources. Furthermore, substantial effort was made to standardize data presentation and make experimental results accessible to data scientists. However, the power and complexity of current experimental and theoretical tools make it more and more difficult to assess the capacity of gathered parameters to support optimal progress in our understanding of cell function. The purpose of this review is to focus on biomolecule interactions, the interactome, as a specific and important example, and examine the limitations of the explanatory and predictive power of parameters that are considered as suitable descriptors of molecular interactions. Recent experimental studies on important cell functions, such as adhesion and processing of environmental cues for decision-making, support the suggestion that it should be rewarding to complement standard binding properties such as affinity and kinetic constants, or even force dependence, with less frequently used parameters such as conformational flexibility or size of binding molecules.

Prognostic scoring systems and risk stratification in myelodysplastic syndrome: focus on integration of molecular profile

Myelodysplastic syndromes (MDS) form a clinically and molecularly heterogeneous disease group. Precise risk stratification remains crucial for choosing optimal management strategies. Several conventional prognostic scoring systems have been developed and validated in the MDS population. These risk models divide patients into prognostic subgroups based on clinical and cytogenetic characteristics. Lack of dynamicity, variable risk estimate across models, and heterogeneity within intermediate-risk group are the limitations of traditional models like IPSS-R, with questionable relevance of these scoring systems in treated MDS patients. Recent progress in next-generation sequencing techniques has improved understanding of the distribution and prognostic importance of recurrent genetic mutations in MDS. Early studies have suggested that incorporating mutations in risk stratification could supplement IPSS-R in further refining the model's performance in predicting overall survival and risk of transformation to acute myeloid leukemia and should translate into a molecularly driven prognostication approach in the near future.

Have we reached a molecular era in myelodysplastic syndromes?

Myelodysplastic syndromes (MDS) are characterized by heterogeneous biological and clinical characteristics, leading to variable outcomes. The availability of sophisticated platforms of genome sequencing allowed the discovery of recurrently mutated genes, which have led to a new era in MDS. This is reflected by the 2016 update of the World Health Organization classification, in which the criteria to define MDS with ringed sideroblasts include the presence of SF3B1 mutations. Further, the detection of somatic mutations in myeloid genes at high variant allele frequency guides the diagnostic algorithm in cases with cytopenias, unclear dysplastic changes, and normal karyotypes, supporting MDS over alternative diagnoses. SF3B1 mutations have been shown to play a positive prognostic role, while mutations in ASXL1, EZH2, RUNX1, and TP53 have been associated with a dismal prognosis. This is particularly relevant in lower- and intermediate-risk disease, in which a higher number of mutations and/or the presence of "unfavorable" somatic mutations may support the use of disease-modifying treatments. In the near future, the incorporation of mutation profiles in currently used prognostication systems, also taking into consideration the classical patient clinical variables (including age and comorbidities), will support a more precise disease stratification, eg, the assignment to targeted treatment approaches or to allogeneic stem cell transplantation in younger patients.

TET2 mutations as a part of DNA dioxygenase deficiency in myelodysplastic syndromes

5-hydroxymethylcytosine and TET2 messenger RNA (mRNA) downregulation are common in myelodysplastic syndromes irrespective of TET2 mutations.

TET3 mRNA expression levels are associated with distinct clinical outcomes in myelodysplastic syndromes with and without TET2 mutations.

Decrease in DNA dioxygenase activity generated by TET2 gene family is crucial in myelodysplastic syndromes (MDS). The general downregulation of 5-hydroxymethylcytosine (5-hmC) argues for a role of DNA demethylation in MDS beyond TET2 mutations, which albeit frequent, do not convey any prognostic significance. We investigated TETs expression to identify factors which can modulate the impact of mutations and thus 5-hmC levels on clinical phenotypes and prognosis of MDS patients. DNA/RNA-sequencing and 5-hmC data were collected from 1665 patients with MDS and 91 controls. Irrespective of mutations, a significant fraction of MDS patients exhibited lower TET2 expression, whereas 5-hmC levels were not uniformly decreased. In searching for factors explaining compensatory mechanisms, we discovered that TET3 was upregulated in MDS and inversely correlated with TET2 expression in wild-type cases. Although TET2 was reduced across all age groups, TET3 levels were increased in a likely feedback mechanism induced by TET2 dysfunction. This inverse relationship of TET2 and TET3 expression also corresponded to the expression of L-2-hydroxyglutarate dehydrogenase, involved in agonist/antagonist substrate metabolism. Importantly, elevated TET3 levels influenced the clinical phenotype of TET2 deficiency whereby the lack of compensation by TET3 (low TET3 expression) was associated with poor outcomes of TET2 mutant carriers.

Relationship between clone metrics and clinical outcome in clonal cytopenia

Clonal cytopenia of undetermined significance (CCUS) is associated with an increased risk of developing a myeloid neoplasm with myelodysplasia (MN). To identify the features of the mutant clone(s) that is associated with clinical phenotype and progression, we studied the following cohorts of individuals: 311 patients with idiopathic cytopenia of undetermined significance (ICUS), 532 community-dwelling individuals without hematologic phenotype (n = 355) or with unexplained anemia (n = 177), and 592 patients with overt MN. Ninety-two of 311 (30%) patients with ICUS carried a somatic genetic lesion that signaled CCUS. Clonal hematopoiesis (CH) was detected in 19.7% and 27.7% of nonanemic and anemic community-dwelling individuals, respectively. Different mutation patterns and variant allele frequencies (VAFs) (clone metrics parameters) were observed in the conditions studied. Recurrent mutation patterns exhibited different VAFs associated with marrow dysplasia (0.17-0.48), indicating variable clinical expressivity of mutant clones. Unsupervised clustering analysis based on mutation profiles identified 2 major clusters, characterized by isolated DNMT3A mutations (CH-like cluster) or combinatorial mutation patterns (MN-like cluster), and showing different overall survival (HR, 1.8). In patients with CCUS, the 2 clusters had different risk of progression to MN (HR, 2.7). Within the MN-like cluster, distinct subsets with different risk of progression to MN were identified based on clone metrics. These findings unveil marked variability in the clinical expressivity of myeloid driver genes and underline the limitations of morphologic dysplasia for clinical staging of mutant hematopoietic clones. Clone metrics appears to be critical for informing clinical decision-making in patients with clonal cytopenia.

Deep learning detects acute myeloid leukemia and predicts NPM1 mutation status from bone marrow smears

The evaluation of bone marrow morphology by experienced hematopathologists is essential in the diagnosis of acute myeloid leukemia (AML); however, it suffers from a lack of standardization and inter-observer variability. Deep learning (DL) can process medical image data and provides data-driven class predictions. Here, we apply a multi-step DL approach to automatically segment cells from bone marrow images, distinguish between AML samples and healthy controls with an area under the receiver operating characteristic (AUROC) of 0.9699, and predict the mutation status of Nucleophosmin 1 (NPM1)—one of the most common mutations in AML—with an AUROC of 0.92 using only image data from bone marrow smears. Utilizing occlusion sensitivity maps, we observed so far unreported morphologic cell features such as a pattern of condensed chromatin and perinuclear lightening zones in myeloblasts of NPM1-mutated AML and prominent nucleoli in wild-type NPM1 AML enabling the DL model to provide accurate class predictions.

A predictive algorithm using clinical and laboratory parameters may assist in ruling out and in diagnosing MDS

We present a noninvasive Web-based app to help exclude or diagnose myelodysplastic syndrome (MDS), a bone marrow (BM) disorder with cytopenias and leukemic risk, diagnosed by BM examination. A sample of 502 MDS patients from the European MDS (EUMDS) registry (n > 2600) was combined with 502 controls (all BM proven). Gradient-boosted models (GBMs) were used to predict/exclude MDS using demographic, clinical, and laboratory variables. Area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were used to evaluate the models, and performance was validated using 100 times fivefold cross-validation. Model stability was assessed by repeating its fit using different randomly chosen groups of 502 EUMDS cases. AUC was 0.96 (95% confidence interval, 0.95-0.97). MDS is predicted/excluded accurately in 86% of patients with unexplained anemia. A GBM score (range, 0-1) of less than 0.68 (GBM < 0.68) resulted in a negative predictive value of 0.94, that is, MDS was excluded. GBM ≥ 0.82 provided a positive predictive value of 0.88, that is, MDS. The diagnosis of the remaining patients (0.68 ≤ GBM < 0.82) is indeterminate. The discriminating variables: age, sex, hemoglobin, white blood cells, platelets, mean corpuscular volume, neutrophils, monocytes, glucose, and creatinine. A Web-based app was developed; physicians could use it to exclude or predict MDS noninvasively in most patients without a BM examination. Future work will add peripheral blood cytogenetics/genetics, EUMDS-based prospective validation, and prognostication.

Myelodysplastic Syndromes in the Postgenomic Era and Future Perspectives for Precision Medicine

Simple Summary

With demographic ageing, improved cancer survivorship and increased diagnostic sensitivity, incident cases of patients with Myelodysplastic Syndromes (MDS) are continuously rising, leading to a relevant impact on health care resources. Disease heterogeneity and various comorbidities are challenges for the management of the generally elderly patients. Therefore, experienced physicians and multidisciplinary teams should be involved in the establishment of the correct diagnosis, risk-assessment and personalized treatment plan. Next-generation sequencing allows for early detection of clonal hematopoiesis and monitoring of clonal evolution, but also poses new challenges for its appropriate use. At present, allogeneic hematopoietic stem cell transplantation remains the only curative treatment option for a minority of fit MDS patients. All others receive palliative treatment and will eventually progress, having an unmet need for novel therapies. Targeting compounds are in prospect for precision medicine, however, abrogation of clonal evolution to acute myeloid leukemia remains actually out of reach.

Abstract

Myelodysplastic syndromes (MDS) represent a heterogeneous group of clonal disorders caused by sequential accumulation of somatic driver mutations in hematopoietic stem and progenitor cells (HSPCs). MDS is characterized by ineffective hematopoiesis with cytopenia, dysplasia, inflammation, and a variable risk of transformation into secondary acute myeloid leukemia. The advent of next-generation sequencing has revolutionized our understanding of the genetic basis of the disease. Nevertheless, the biology of clonal evolution remains poorly understood, and the stochastic genetic drift with sequential accumulation of genetic hits in HSPCs is individual, highly dynamic and hardly predictable. These continuously moving genetic targets pose substantial challenges for the implementation of precision medicine, which aims to maximize efficacy with minimal toxicity of treatments. In the current postgenomic era, allogeneic hematopoietic stem cell transplantation remains the only curative option for younger and fit MDS patients. For all unfit patients, regeneration of HSPCs stays out of reach and all available therapies remain palliative, which will eventually lead to refractoriness and progression. In this review, we summarize the recent advances in our understanding of MDS pathophysiology and its impact on diagnosis, risk-assessment and disease monitoring. Moreover, we present ongoing clinical trials with targeting compounds and highlight future perspectives for precision medicine.

Deep learning for diagnosis of acute promyelocytic leukemia via recognition of genomically imprinted morphologic features

Acute promyelocytic leukemia (APL) is a subtype of acute myeloid leukemia (AML), classified by a translocation between chromosomes 15 and 17 [t(15;17)], that is considered a true oncologic emergency though appropriate therapy is considered curative. Therapy is often initiated on clinical suspicion, informed by both clinical presentation as well as direct visualization of the peripheral smear. We hypothesized that genomic imprinting of morphologic features learned by deep learning pattern recognition would have greater discriminatory power and consistency compared to humans, thereby facilitating identification of t(15;17) positive APL. By applying both cell-level and patient-level classification linked to t(15;17) PML/RARA ground-truth, we demonstrate that deep learning is capable of distinguishing APL in both discovery and prospective independent cohort of patients. Furthermore, we extract learned information from the trained network to identify previously undescribed morphological features of APL. The deep learning method we describe herein potentially allows a rapid, explainable, and accurate physician-aid for diagnosing APL at the time of presentation in any resource-poor or -rich medical setting given the universally available peripheral smear.

Machine Learning of Bone Marrow Histopathology Identifies Genetic and Clinical Determinants in Patients with MDS

In myelodysplastic syndrome (MDS) and myeloproliferative neoplasm (MPN), bone marrow (BM) histopathology is assessed to identify dysplastic cellular morphology, cellularity, and blast excess. Yet, other morphologic findings may elude the human eye. We used convolutional neural networks to extract morphologic features from 236 MDS, 87 MDS/MPN, and 11 control BM biopsies. These features predicted genetic and cytogenetic aberrations, prognosis, age, and gender in multivariate regression models. Highest prediction accuracy was found for TET2 [area under the receiver operating curve (AUROC) = 0.94] and spliceosome mutations (0.89) and chromosome 7 monosomy (0.89). Mutation prediction probability correlated with variant allele frequency and number of affected genes per pathway, demonstrating the algorithms' ability to identify relevant morphologic patterns. By converting regression models to texture and cellular composition, we reproduced the classical del(5q) MDS morphology consisting of hypolobulated megakaryocytes. In summary, this study highlights the potential of linking deep BM histopathology with genetics and clinical variables.

SIGNIFICANCE

Histopathology is elementary in the diagnostics of patients with MDS, but its high-dimensional data are underused. By elucidating the association of morphologic features with clinical variables and molecular genetics, this study highlights the vast potential of convolutional neural networks in understanding MDS pathology and how genetics is reflected in BM morphology. See related commentary by Elemento, p. 195.

Towards artificial intelligence-driven pathology assessment for hematological malignancies

In this issue of Blood Cancer Discovery, Brück et al applied unsupervised and supervised machine learning to bone marrow histopathology images from Myelodysplastic Syndrome (MDS) patients. Their study provides new insights into the pathobiology of MDS and paves the way for increased use of artificial intelligence for the assessment and diagnosis of hematological malignancies.

Genetic Aspects of Myelodysplastic/Myeloproliferative Neoplasms

Simple Summary

Myelodysplastic/myeloproliferative neoplasms (MDS/MPN) are clonal myeloid neoplasms characterized, at the time of their presentation, by the simultaneous presence of both myelodysplastic and myeloproliferative features. In MDS/MPN, the karyotype is often normal but mutations in genes that are common across myeloid neoplasms can be detected in a high proportion of cases by targeted sequencing. In this review, we intend to summarize the main genetic findings across all MDS/MPN overlap syndromes and discuss their relevance in the management of patients.

Abstract

Myelodysplastic/myeloproliferative neoplasms (MDS/MPN) are myeloid neoplasms characterized by the presentation of overlapping features from both myelodysplastic syndromes and myeloproliferative neoplasms. Although the classification of MDS/MPN relies largely on clinical features and peripheral blood and bone marrow morphology, studies have demonstrated that a large proportion of patients (~90%) with this disease harbor somatic mutations in a group of genes that are common across myeloid neoplasms. These mutations play a role in the clinical heterogeneity of these diseases and their clinical evolution. Nevertheless, none of them is specific to MDS/MPN and current diagnostic criteria do not include molecular data. Even when such alterations can be helpful for differential diagnosis, they should not be used alone as proof of neoplasia because some of these mutations may also occur in healthy older people. Here, we intend to review the main genetic findings across all MDS/MPN overlap syndromes and discuss their relevance in the management of the patients.

Application of Single-Cell Approaches to Study Myeloproliferative Neoplasm Biology

Philadelphia-negative myeloproliferative neoplasms (MPNs) are an excellent tractable disease model of a number of aspects of human cancer biology, including genetic evolution, tissue-associated fibrosis, and cancer stem cells. In this review, we discuss recent insights into MPN biology gained from the application of a number of new single-cell technologies to study human disease, with a specific focus on single-cell genomics, single-cell transcriptomics, and digital pathology.

Management of patients with higher-risk myelodysplastic syndromes after failure of hypomethylating agents: What is on the horizon?

The hypomethylating agents (HMA) azacitidine (AZA) and decitabine (DAC) are the standard of care for frontline treatment of patients with higher-risk myelodysplastic syndromes (MDS). As complete responses to HMAs are rare and typically not durable, HMA failure is a common clinical dilemma and associated with very short survival in most patients. Salvage therapies with various agents such as novel HMAs (guadecitabine, CC-486), and CTLA-4/PD1-type immune checkpoint inhibitors (ICPIs) have yielded mixed and only modest results at best in MDS patients with HMA failure. Thanks to advances in the understanding of the molecular and biologic pathogenesis of MDS, several novel targeted agents such as the BCL-2 inhibitor venetoclax, TP-53 refolding agent APR-246, IDH1/2 inhibitors, and novel ICPIs such as magrolimab and sabatolimab have been developed and demonstrated activity in combination with HMA in the frontline setting. However, clinical testing of these agents post HMA failure has been limited to date. Furthermore, the biology of HMA failure remains poorly defined which significantly limits rationale drug development. This highlights the importance of optimization of frontline therapy to avoid/delay HMA failure in addition to development of more effective salvage therapies.

Navigating Myelodysplastic and Myelodysplastic/Myeloproliferative Overlap Syndromes

Myelodysplastic syndromes (MDS) and MDS/myeloproliferative neoplasms (MPNs) are clonal diseases that differ in morphologic diagnostic criteria but share some common disease phenotypes that include cytopenias, propensity to acute myeloid leukemia evolution, and a substantially shortened patient survival. MDS/MPNs share many clinical and molecular features with MDS, including frequent mutations involving epigenetic modifier and/or spliceosome genes. Although the current 2016 World Health Organization classification incorporates some genetic features in its diagnostic criteria for MDS and MDS/MPNs, recent accumulation of data has underscored the importance of the mutation profiles on both disease classification and prognosis. Machine-learning algorithms have identified distinct molecular genetic signatures that help refine prognosis and notable associations of these genetic signatures with morphologic and clinical features. Combined geno-clinical models that incorporate mutation data seem to surpass the current prognostic schemes. Future MDS classification and prognostication schema will be based on the portfolio of genetic aberrations and traditional features, such as blast count and clinical factors. Arriving at these systems will require studies on large patient cohorts that incorporate advanced computational analysis. The current treatment algorithm in MDS is based on patient risk as derived from existing prognostic and disease classes. Luspatercept is newly approved for patients with MDS and ring sideroblasts who are transfusion dependent after erythropoietic-stimulating agent failure. Other agents that address red blood cell transfusion dependence in patients with lower-risk MDS and the failure of hypomethylating agents in higher-risk disease are in advanced testing. Finally, a plethora of novel targeted agents and immune checkpoint inhibitors are being evaluated in combination with a hypomethylating agent backbone to augment the depth and duration of response and, we hope, improve overall survival.

Prognostic mutation constellations in acute myeloid leukaemia and myelodysplastic syndrome

PURPOSE OF REVIEW

In the past decade, numerous studies analysing the genome and transcriptome of large cohorts of acute myeloid leukaemia (AML) and myelodysplastic syndrome (MDS) patients have substantially improved our knowledge of the genetic landscape of these diseases with the identification of heterogeneous constellations of germline and somatic mutations with prognostic and therapeutic relevance. However, inclusion of integrated genetic data into classification schema is still far from a reality. The purpose of this review is to summarize recent insights into the prevalence, pathogenic role, clonal architecture, prognostic impact and therapeutic management of genetic alterations across the spectrum of myeloid malignancies.

RECENT FINDINGS

Recent multiomic-studies, including analysis of genetic alterations at the single-cell resolution, have revealed a high heterogeneity of lesions in over 200 recurrently mutated genes affecting disease initiation, clonal evolution and clinical outcome. Artificial intelligence and specifically machine learning approaches have been applied to large cohorts of AML and MDS patients to define in an unbiased manner clinically meaningful disease patterns including, disease classification, prognostication and therapeutic vulnerability, paving the way for future use in clinical practice.

SUMMARY

Integration of genomic, transcriptomic, epigenomic and clinical data coupled to conventional and machine learning approaches will allow refined leukaemia classification and risk prognostication and will identify novel therapeutic targets for these still high-risk leukaemia subtypes.