Machine Learning Integrates Genomic Signatures for Subclassification Beyond Primary and Secondary Acute Myeloid Leukemia

Application of m6A regulators to predict transformation from myelodysplastic syndrome to acute myeloid leukemia via machine learning

Myelodysplastic syndrome (MDS) frequently transforms into acute myeloid leukemia (AML). Predicting the risk of its transformation will help to make the treatment plan. Levels of expression of N6-methyladenosine (m6A) regulators is difference in patients with AML, MDS, and MDS transformed into AML. Seven machine learning algorithms were established based on all of 26 m6A or main differentially expressed m6A regulator genes, and attempted to establish a risk assessment method to distinguish AML from MDS and predict the transformation of MDS into AML. In collective of m6A regulators sets, support vector machine (SVM) and neural network (NNK) model best distinguished AML or MDS from control, with area under the ROC curve (AUROC) 0.966 and 0.785 respectively. The SVM model best distinguished MDS from AML, with AUROC 0.943, sensitivity 0.862, specificity 0.864, and accuracy 0.864. In differentially expressed gene sets, SVM and logistic regression (LR) model best distinguished AML or MDS from control, with AUROC 0.945 and 0.801 respectively. The random forest (RF) model best distinguished between MDS and AML, with AUROC 0.928, sensitivity 0.725, specificity 0.898, and accuracy 0.818. For predictive capacity of MDS transformed into AML, SVM model showed the best predicted in collective m6A regulators sets, with AUROC 0.781 and accuracy 0.740. The LR model showed the best predicted in differential expression m6A regulators sets, with AUROC 0.820 and accuracy 0.760. All results suggested that machine learning model established by m6A regulators can be used to distinguished AML or MDS from control, distinguished AML from MDS and predicted the transformation of MDS into AML.

Systematic literature review on reinforcement learning in non-communicable disease interventions

There is evidence that reducing modifiable risk factors and strengthening medical and health interventions can reduce early mortality and economic losses from non-communicable diseases (NCDs). Machine learning (ML) algorithms have been successfully applied to preventing and controlling NCDs. Reinforcement learning (RL) is the most promising of these approaches because of its ability to dynamically adapt interventions to NCD disease progression and its commitment to achieving long-term intervention goals. This paper reviews the preferred algorithms, data sources, design details, and obstacles to clinical application in existing studies to facilitate the early application of RL algorithms in clinical practice research for NCD interventions. We screened 40 relevant papers for quantitative and qualitative analysis using the PRISMA review flow diagram. The results show that researchers tend to use Deep Q-Network (DQN) and Actor-Critic as well as their improved or hybrid algorithms to train and validate RL models on retrospective datasets. Often, the patient's physical condition is the main defining parameter of the state space, while interventions are the main defining parameter of the action space. Mostly, changes in the patient's physical condition are used as a basis for immediate rewards to the agent. Various attempts have been made to address the challenges to clinical application, and several approaches have been proposed from existing research. However, as there is currently no universally accepted solution, the use of RL algorithms in clinical practice for NCD interventions necessitates more comprehensive responses to the issues addressed in this paper, which are safety, interpretability, training efficiency, and the technical aspect of exploitation and exploration in RL algorithms.

Therapy-Related Myeloid Neoplasm: Biology and Mechanistic Aspects of Malignant Progression

Therapy-related myeloid neoplasms (t-MN) arise after a documented history of chemo/radiotherapy as treatment for an unrelated condition and account for 10-20% of myelodysplastic syndromes and acute myeloid leukemia. T-MN are characterized by a specific genetic signature, aggressive features and dismal prognosis. The nomenclature and the subsets of these conditions have changed frequently over time, and despite the fact that, in the last classification, they lost their autonomous entity status and became disease qualifiers, the recognition of this feature remains of major importance. Furthermore, in recent years, extensive studies focusing on clonal hematopoiesis and germline variants shed light on the mechanisms of positive pressure underpinning the rise of driver gene mutations in t-MN. In this manuscript, we aim to review the evolution of defining criteria and characteristics of t-MN from a clinical and biological perspective, the advances in mechanistic aspects of malignant progression and the challenges in prevention and management.

Transplant Eligible and Ineligible Elderly Patients with AML-A Genomic Approach and Next Generation Questions

The management of elderly patients diagnosed with acute myelogenous leukemia (AML) is complicated by high relapse risk and comorbidities that often preclude access to allogeneic hematopoietic cellular transplantation (allo-HCT). In recent years, fast-paced FDA drug approval has reshaped the therapeutic landscape, with modest, albeit promising improvement in survival. Still, AML outcomes in elderly patients remain unacceptably unfavorable highlighting the need for better understanding of disease biology and tailored strategies. In this review, we discuss recent modifications suggested by European Leukemia Network 2022 (ELN-2022) risk stratification and review recent aging cell biology advances with the discussion of four AML cases. While an older age, >60 years, does not constitute an absolute contraindication for allo-HCT, the careful patient selection based on a detailed and multidisciplinary risk stratification cannot be overemphasized.

Validation of the 2022 European LeukemiaNet risk stratification for acute myeloid leukemia

This study aimed to validate the 2022 European LeukemiaNet (ELN) risk stratification for acute myeloid leukemia (AML). A total of 624 newly diagnosed AML patients from 1998 to 2014 were included in the analysis. Genetic profiling was conducted using targeted deep sequencing of 45 genes based on recurrent driver mutations. In total, 134 (21.5%) patients had their risk classification reassessed according to the 2022 ELN risk stratification. Among those initially classified as having a favorable risk in 2017 (n = 218), 31 and 3 patients were reclassified as having intermediate risk or adverse risk, respectively. Among the three subgroups, the 2022 ELN favorable-risk group showed significantly longer survival outcomes than the other groups. Within the 2017 ELN intermediate-risk group (n = 298), 21 and 46 patients were reclassified as having favorable risk or adverse risk, respectively, and each group showed significant stratifications in survival outcomes. Some patients initially classified as having adverse risk in 2017 were reclassified into the intermediate-risk group (33 of 108 patients), but no prognostic improvements were observed in this group. A multivariable analysis identified the 2022 ELN risk stratification, age, and receiving allogeneic hematopoietic cell transplantation as significant prognostic factors for survival. The 2022 ELN risk stratification enables more precise decisions for proceeding with allogeneic hematopoietic cell transplantation for AML patients.

Identification of hub genes and potential molecular mechanisms related to drug sensitivity in acute myeloid leukemia based on machine learning

Background: Acute myeloid leukemia (AML) is the most common form of leukemia among adults and is characterized by uncontrolled proliferation and clonal expansion of hematopoietic cells. There has been a significant improvement in the treatment of younger patients, however, prognosis in the elderly AML patients remains poor. Methods: We used computational methods and machine learning (ML) techniques to identify and explore the differential high-risk genes (DHRGs) in AML. The DHRGs were explored through multiple in silico approaches including genomic and functional analysis, survival analysis, immune infiltration, miRNA co-expression and stemness features analyses to reveal their prognostic importance in AML. Furthermore, using different ML algorithms, prognostic models were constructed and validated using the DHRGs. At the end molecular docking studies were performed to identify potential drug candidates targeting the selected DHRGs. Results: We identified a total of 80 DHRGs by comparing the differentially expressed genes derived between AML patients and normal controls and high-risk AML genes identified by Cox regression. Genetic and epigenetic alteration analyses of the DHRGs revealed a significant association of their copy number variations and methylation status with overall survival (OS) of AML patients. Out of the 137 models constructed using different ML algorithms, the combination of Ridge and plsRcox maintained the highest mean C-index and was used to build the final model. When AML patients were classified into low- and high-risk groups based on DHRGs, the low-risk group had significantly longer OS in the AML training and validation cohorts. Furthermore, immune infiltration, miRNA coexpression, stemness feature and hallmark pathway analyses revealed significant differences in the prognosis of the low- and high-risk AML groups. Drug sensitivity and molecular docking studies revealed top 5 drugs, including carboplatin and austocystin-D that may significantly affect the DHRGs in AML. Conclusion: The findings from the current study identified a set of high-risk genes that may be used as prognostic and therapeutic markers for AML patients. In addition, significant use of the ML algorithms in constructing and validating the prognostic models in AML was demonstrated. Although our study used extensive bioinformatics and machine learning methods to identify the hub genes in AML, their experimental validations using knock-out/-in methods would strengthen our findings.

Using machine learning to unravel the intricacy of acute myeloid leukemia

Not available.

Prognostic value of European LeukemiaNet 2022 criteria and genomic clusters using machine learning in older adults with acute myeloid leukemia

This study aimed to validate the new European Leukemia Net (ELN) 2022 criteria for genetic risk stratification in older adults with acute myeloid leukemia (AML) and to determine the most likely set of clusters of similar cytogenetic and mutation properties correlated with survival outcomes in three treatment groups: intensive chemotherapy (IC), hypomethylating agents (HMA) alone, and HMA plus venetoclax (HMA/VEN). The study included 279 patients (aged ≥60 years) who received IC (N=131), HMA (N=76), and HMA/VEN (N=72) between July 2017 and October 2021. No significant differences were observed in survival among the groups according to ELN 2022 risk stratification. Unsupervised hierarchical clustering analysis identified nine genomic clusters (C1-9) with varying survival outcomes depending on treatment type. For example, C4 (predominant for core binding factor-AML) displayed a favorable prognosis in the IC group, but not in the HMA or HMA/VEN groups. The HMA/VEN group had better outcomes than the HMA group in many clusters (C1, 2, 3, and 5); however, the addition of VEN to HMA or IC did not improve the survival outcomes compared with those of HMA alone in C7 and C9 (predominant for -5, del(5q), -7, -17/abn(17p), complex karyotypes, and mutated TP53). The study highlights the limitations of ELN genetic risk stratification in older adults with AML. It emphasizes the need for a more comprehensive approach that considers co-occurring somatic mutations to guide treatment selection in older adults with AML.

Preliminary Evidence of the Use of Generative AI in Health Care Clinical Services: Systematic Narrative Review

BACKGROUND

Generative artificial intelligence tools and applications (GenAI) are being increasingly used in health care. Physicians, specialists, and other providers have started primarily using GenAI as an aid or tool to gather knowledge, provide information, train, or generate suggestive dialogue between physicians and patients or between physicians and patients' families or friends. However, unless the use of GenAI is oriented to be helpful in clinical service encounters that can improve the accuracy of diagnosis, treatment, and patient outcomes, the expected potential will not be achieved. As adoption continues, it is essential to validate the effectiveness of the infusion of GenAI as an intelligent technology in service encounters to understand the gap in actual clinical service use of GenAI.

OBJECTIVE

This study synthesizes preliminary evidence on how GenAI assists, guides, and automates clinical service rendering and encounters in health care The review scope was limited to articles published in peer-reviewed medical journals.

METHODS

We screened and selected 0.38% (161/42,459) of articles published between January 1, 2020, and May 31, 2023, identified from PubMed. We followed the protocols outlined in the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines to select highly relevant studies with at least 1 element on clinical use, evaluation, and validation to provide evidence of GenAI use in clinical services. The articles were classified based on their relevance to clinical service functions or activities using the descriptive and analytical information presented in the articles.

RESULTS

Of 161 articles, 141 (87.6%) reported using GenAI to assist services through knowledge access, collation, and filtering. GenAI was used for disease detection (19/161, 11.8%), diagnosis (14/161, 8.7%), and screening processes (12/161, 7.5%) in the areas of radiology (17/161, 10.6%), cardiology (12/161, 7.5%), gastrointestinal medicine (4/161, 2.5%), and diabetes (6/161, 3.7%). The literature synthesis in this study suggests that GenAI is mainly used for diagnostic processes, improvement of diagnosis accuracy, and screening and diagnostic purposes using knowledge access. Although this solves the problem of knowledge access and may improve diagnostic accuracy, it is oriented toward higher value creation in health care.

CONCLUSIONS

GenAI informs rather than assisting or automating clinical service functions in health care. There is potential in clinical service, but it has yet to be actualized for GenAI. More clinical service-level evidence that GenAI is used to streamline some functions or provides more automated help than only information retrieval is needed. To transform health care as purported, more studies related to GenAI applications must automate and guide human-performed services and keep up with the optimism that forward-thinking health care organizations will take advantage of GenAI.

Molecular and clinical analyses of PHF6 mutant myeloid neoplasia provide their pathogenesis and therapeutic targeting

PHF6 mutations (PHF6MT) are identified in various myeloid neoplasms (MN). However, little is known about the precise function and consequences of PHF6 in MN. Here we show three main findings in our comprehensive genomic and proteomic study. Firstly, we show a different pattern of genes correlating with PHF6MT in male and female cases. When analyzing male and female cases separately, in only male cases, RUNX1 and U2AF1 are co-mutated with PHF6. In contrast, female cases reveal co-occurrence of ASXL1 mutations and X-chromosome deletions with PHF6MT. Next, proteomics analysis reveals a direct interaction between PHF6 and RUNX1. Both proteins co-localize in active enhancer regions that define the context of lineage differentiation. Finally, we demonstrate a negative prognostic role of PHF6MT, especially in association with RUNX1. The negative effects on survival are additive as PHF6MT cases with RUNX1 mutations have worse outcomes when compared to cases carrying single mutation or wild-type.

Artificial intelligence-based prediction models for acute myeloid leukemia using real-life data: A DATAML registry study

We designed artificial intelligence-based prediction models (AIPM) using 52 diagnostic variables from 3687 patients included in the DATAML registry treated with intensive chemotherapy (IC, N = 3030) or azacitidine (AZA, N = 657) for an acute myeloid leukemia (AML). A neural network called multilayer perceptron (MLP) achieved a prediction accuracy for overall survival (OS) of 68.5% and 62.1% in the IC and AZA cohorts, respectively. The Boruta algorithm could select the most important variables for prediction without decreasing accuracy. Thirteen features were retained with this algorithm in the IC cohort: age, cytogenetic risk, white blood cells count, LDH, platelet count, albumin, MPO expression, mean corpuscular volume, CD117 expression, NPM1 mutation, AML status (de novo or secondary), multilineage dysplasia and ASXL1 mutation; and 7 variables in the AZA cohort: blood blasts, serum ferritin, CD56, LDH, hemoglobin, CD13 and disseminated intravascular coagulation (DIC). We believe that AIPM could help hematologists to deal with the huge amount of data available at diagnosis, enabling them to have an OS estimation and guide their treatment choice. Our registry-based AIPM could offer a large real-life dataset with original and exhaustive features and select a low number of diagnostic features with an equivalent accuracy of prediction, more appropriate to routine practice.

A scoring system based on fusion genes to predict treatment outcomes of the non-acute promyelocytic leukemia pediatric acute myeloid leukemia

Background

Fusion genes are considered to be one of the major drivers behind cancer initiation and progression. Meanwhile, non-acute promyelocytic leukemia (APL) pediatric patients with acute myeloid leukemia (AML) in children had limited treatment efficacy. Hence, we developed and validated a simple clinical scoring system for predicting outcomes in non-APL pediatric patients with AML.

Method

A total of 184 non-APL pediatric patients with AML who were admitted to our hospital and an independent dataset (318 patients) from the TARGET database were included. Least absolute shrinkage and selection operation (LASSO) and Cox regression analysis were used to identify prognostic factors. Then, a nomogram score was developed to predict the 1, 3, and 5 years overall survival (OS) based on their clinical characteristics and fusion genes. The accuracy of the nomogram score was determined by calibration curves and receiver operating characteristic (ROC) curves. Additionally, an internal verification cohort was used to assess its applicability.

Results

Based on Cox and LASSO regression analyses, a nomogram score was constructed using clinical characteristics and OS-related fusion genes (CBFβ::MYH11, RUNX1::RUNX1T1, KMT2A::ELL, and KMT2A::MLLT10), yielded good calibration and concordance for predicting OS of non-APL pediatric patients with AML. Furthermore, patients with higher scores exhibited worse outcomes. The nomogram score also demonstrated good discrimination and calibration in the whole cohort and internal validation. Furthermore, artificial neural networks demonstrated that this nomogram score exhibits good predictive performance.

Conclusion

Our model based on the fusion gene is a prognostic biomarker for non-APL pediatric patients with AML. The nomogram score can provide personalized prognosis prediction, thereby benefiting clinical decision-making.

Genomics of deletion 7 and 7q in myeloid neoplasm: from pathogenic culprits to potential synthetic lethal therapeutic targets

Complete or partial deletions of chromosome 7 (-7/del7q) belong to the most frequent chromosomal abnormalities in myeloid neoplasm (MN) and are associated with a poor prognosis. The disease biology of -7/del7q and the genes responsible for the leukemogenic properties have not been completely elucidated. Chromosomal deletions may create clonal vulnerabilities due to haploinsufficient (HI) genes contained in the deleted regions. Therefore, HI genes are potential targets of synthetic lethal strategies. Through the most comprehensive multimodal analysis of more than 600 -7/del7q MN samples, we elucidated the disease biology and qualified a list of most consistently deleted and HI genes. Among them, 27 potentially synthetic lethal target genes were identified with the following properties: (i) unaffected genes by hemizygous/homozygous LOF mutations; (ii) prenatal lethality in knockout mice; and (iii) vulnerability of leukemia cells by CRISPR and shRNA knockout screens. In -7/del7q cells, we also identified 26 up or down-regulated genes mapping on other chromosomes as downstream pathways or compensation mechanisms. Our findings shed light on the pathogenesis of -7/del7q MNs, while 27 potential synthetic lethal target genes and 26 differential expressed genes allow for a therapeutic window of -7/del7q.

Diagnosis of acute myeloid leukaemia on microarray gene expression data using categorical gradient boosted trees

We define an iterative method for dimensionality reduction using categorical gradient boosted trees and Shapley values and created four machine learning models which potentially could be used as diagnostic tests for acute myeloid leukaemia (AML). For the final Catboost model we use a dataset of 2177 individuals using as features 16 probe sets and the age in order to classify if someone has AML or is healthy. The dataset is multicentric and consists of data from 27 organizations, 25 cities, 15 countries and 4 continents. The performance of our last model is specificity: 0.9909, sensitivity: 0.9985, F1-score: 0.9976 and its ROC-AUC: 0.9962 using ten fold cross validation. On an inference dataset the perormance is: specificity: 0.9909, sensitivity: 0.9969, F1-score: 0.9969 and its ROC-AUC: 0.9939. To the best of our knowledge the performance of our model is the best one in the literature, as regards the diagnosis of AML using similar or not data. Moreover, there has not been any bibliographic reference which associates AML or any other type of cancer with the 16 probe sets we used as features in our final model.

Machine Learning Predicts 30-Day Outcome among Acute Myeloid Leukemia Patients: A Single-Center, Retrospective, Cohort Study

Acute myeloid leukemia (AML) is a clinical emergency requiring treatment and results in high 30-day (D30) mortality. In this study, the prediction of D30 survival was studied using a machine learning (ML) method. The total cohort consisted of 1700 survivors and 130 non-survivors at D30. Eight clinical and 42 laboratory variables were collected at the time of diagnosis by pathology. Among them, six variables were selected by a feature selection method: induction chemotherapy (CTx), hemorrhage, infection, C-reactive protein, blood urea nitrogen, and lactate dehydrogenase. Clinical and laboratory data were entered into the training model for D30 survival prediction, followed by testing. Among the tested ML algorithms, the decision tree (DT) algorithm showed higher accuracy, the highest sensitivity, and specificity values (95% CI) of 90.6% (0.918-0.951), 70.4% (0.885-0.924), and 92.1% (0.885-0.924), respectively. DT classified patients into eight specific groups with distinct features. Group 1 with CTx showed a favorable outcome with a survival rate of 97.8% (1469/1502). Group 6, with hemorrhage and the lowest fibrinogen level at diagnosis, showed the worst survival rate of 45.5% (25/55) and 20.5 days. Prediction of D30 survival among AML patients by classification of patients with DT showed distinct features that might support clinical decision-making.

Applicability of 2022 classifications of acute myeloid leukemia in the real-world setting

The increasing knowledge of molecular genetics of acute myeloid leukemia (AML) necessitated the update of previous diagnostic and prognostic schemes, which resulted in the development of the World Health Organization (WHO), the International Consensus Classification (ICC), and the new European LeukemiaNet (ELN) recommendations in 2022. We aimed to provide a real-world application of the new models, unravel differences and similarities, and test their implementation in clinical AML diagnosis. A total of 1001 patients diagnosed with AML were reclassified based on the new schemes. The overall diagnostic changes between the WHO 2016 and the WHO 2022 and ICC classifications were 22.8% and 23.7%, respectively, with a 13.1% difference in patients' distribution between ICC and WHO 2022. The 2022 ICC "not otherwise specified" and WHO "defined by differentiation" AML category sizes shrank when compared with that in WHO 2016 (24.1% and 26.8% respectively, vs 38.7%), particularly because of an expansion of the myelodysplasia (MDS)-related group. Of 397 patients with a MDS-related AML according to the ICC, 55.9% were defined by the presence of a MDS-related karyotype. The overall restratification between ELN 2017 and ELN 2022 was 12.9%. The 2022 AML classifications led to a significant improvement of diagnostic schemes. In the real-world setting, conventional cytogenetics, usually rapidly available and less expensive than molecular characterization, stratified 56% of secondary AML, still maintaining a powerful diagnostic role. Considering the similarities between WHO and ICC diagnostic schemes, a tentative scheme to generate a unified model is desirable.

Interaction between myelodysplasia-related gene mutations and ontogeny in acute myeloid leukemia

Accurate classification and risk stratification are critical for clinical decision making in patients with acute myeloid leukemia (AML). In the newly proposed World Health Organization and International Consensus classifications of hematolymphoid neoplasms, the presence of myelodysplasia-related (MR) gene mutations is included as 1 of the diagnostic criteria for AML, AML-MR, based largely on the assumption that these mutations are specific for AML with an antecedent myelodysplastic syndrome. ICC also prioritizes MR gene mutations over ontogeny (as defined in the clinical history). Furthermore, European LeukemiaNet (ELN) 2022 stratifies these MR gene mutations into the adverse-risk group. By thoroughly annotating a cohort of 344 newly diagnosed patients with AML treated at the Memorial Sloan Kettering Cancer Center, we show that ontogeny assignments based on the database registry lack accuracy. MR gene mutations are frequently observed in de novo AML. Among the MR gene mutations, only EZH2 and SF3B1 were associated with an inferior outcome in the univariate analysis. In a multivariate analysis, AML ontogeny had independent prognostic values even after adjusting for age, treatment, allo-transplant and genomic classes or ELN risks. Ontogeny also helped stratify the outcome of AML with MR gene mutations. Finally, de novo AML with MR gene mutations did not show an adverse outcome. In summary, our study emphasizes the importance of accurate ontogeny designation in clinical studies, demonstrates the independent prognostic value of AML ontogeny, and questions the current classification and risk stratification of AML with MR gene mutations.

SETBP1 is dispensable for normal and malignant hematopoiesis

SETBP1 is a potential epigenetic regulator whose hotspot mutations preventing proteasomal degradation are recurrently detected in myeloid malignancies with poor prognosis. It is believed that the mutant SETBP1 exerts amplified effects of wild-type SETBP1 rather than neomorphic functions. This indicates that dysregulated quantitative control of SETBP1 would result in the transformation of hematopoietic cells. However, little is known about the roles of endogenous SETBP1 in malignant and normal hematopoiesis. Thus, we integrated the analyses of primary AML and healthy samples, cancer cell lines, and a newly generated murine model, Vav1-iCre;Setbp1fl/fl. Despite the expression in long-term hematopoietic stem cells, SETBP1 depletion in normal hematopoiesis minimally alters self-renewal, differentiation, or reconstitution in vivo. Indeed, its loss does not profoundly alter transcription or chromatin accessibilities. Furthermore, although AML with high SETBP1 mRNA is associated with genetic and clinical characteristics for dismal outcomes, SETBP1 is dispensable for the development or maintenance of AML. Contrary to the evidence that SETBP1 mutations are restricted to myeloid malignancies, dependency on SETBP1 mRNA expression is not observed in AML. These unexpected results shed light on the unrecognized idea that a physiologically nonessential gene can act as an oncogene when the machinery of protein degradation is damaged.

Trends in the approval of cancer therapies by the FDA in the twenty-first century

The cancer treatment landscape has changed dramatically since the turn of the century, resulting in substantial improvements in outcomes for patients. This Review summarizes trends in the approval of oncology therapeutic products by the United States Food and Drug Administration (FDA) from January 2000 to October 2022, based on a categorization of these products by their mechanism of action and primary target. Notably, the rate of oncology indication approvals has increased in this time, driven by approvals for targeted therapies, as has the rate of introduction of new therapeutic approaches. Kinase inhibitors are the dominant product class by number of approved products and indications, yet immune checkpoint inhibitors have the second most approvals despite not entering the market until 2011. Other trends include a slight increase in the share of approvals for biomarker-defined populations and the emergence of tumour-site-agnostic approvals. Finally, we consider the implications of the trends for the future of oncology therapeutic product development, including the impact of novel therapeutic approaches and technologies.

Artificial Intelligence-Assisted Diagnostic Cytology and Genomic Testing for Hematologic Disorders

Artificial intelligence (AI) is a rapidly evolving field of computer science that involves the development of computational programs that can mimic human intelligence. In particular, machine learning and deep learning models have enabled the identification and grouping of patterns within data, leading to the development of AI systems that have been applied in various areas of hematology, including digital pathology, alpha thalassemia patient screening, cytogenetics, immunophenotyping, and sequencing. These AI-assisted methods have shown promise in improving diagnostic accuracy and efficiency, identifying novel biomarkers, and predicting treatment outcomes. However, limitations such as limited databases, lack of validation and standardization, systematic errors, and bias prevent AI from completely replacing manual diagnosis in hematology. In addition, the processing of large amounts of patient data and personal information by AI poses potential data privacy issues, necessitating the development of regulations to evaluate AI systems and address ethical concerns in clinical AI systems. Nonetheless, with continued research and development, AI has the potential to revolutionize the field of hematology and improve patient outcomes. To fully realize this potential, however, the challenges facing AI in hematology must be addressed and overcome.

Impact of TP53 on outcome of patients with myelofibrosis undergoing hematopoietic stem cell transplantation

TP53 mutations (TP53MTs) have been associated with poor outcomes in various hematologic malignancies, but no data exist regarding its role in patients with myelofibrosis undergoing hematopoietic stem cell transplantation (HSCT). Here, we took advantage of a large international multicenter cohort to evaluate the role of TP53MT in this setting. Among 349 included patients, 49 (13%) had detectable TP53MT, of whom 30 showed a multihit configuration. Median variant allele frequency was 20.3%. Cytogenetic risk was favorable (71%), unfavorable (23%), and very high (6%), with complex karyotype present in 36 patients (10%). Median survival of patients with TP53MT was 1.5 vs 13.5 years for those with wild-type TP53 (TP53WT; P < .001). Outcome was driven by multihit TP53MT constellation (P < .001), showing 6-year survival of 56% for individuals with single-hit vs 25% for those with multihit TP53MT vs 64% for those with TP53WT. Outcome was independent of current transplantation-specific risk factors and conditioning intensity. Similarly, cumulative incidence of relapse was 17% for single-hit vs 52% for multihit vs 21% for TP53WT. Ten patients with TP53MT (20%) presented as leukemic transformation vs only 7 (2%) in the TP53WT group (P < .001). Out of the 10 patients with TP53MT, 8 showed multihit constellation. Median time to leukemic transformation was shorter for multihit and single-hit TP53MT (0.7 and 0.5 years, respectively) vs 2.5 years for TP53WT. In summary, multihit TP53MT represents a very high-risk group in patients with myelofibrosis who are undergoing HSCT, whereas single-hit TP53MT alone showed similar outcome to patients with nonmutated TP53, informing prognostication for survival and relapse together with current transplantation-specific tools.

Unsupervised meta-clustering identifies risk clusters in acute myeloid leukemia based on clinical and genetic profiles

BACKGROUND

Increasingly large and complex biomedical data sets challenge conventional hypothesis-driven analytical approaches, however, data-driven unsupervised learning can detect inherent patterns in such data sets.

METHODS

While unsupervised analysis in the medical literature commonly only utilizes a single clustering algorithm for a given data set, we developed a large-scale model with 605 different combinations of target dimensionalities as well as transformation and clustering algorithms and subsequent meta-clustering of individual results. With this model, we investigated a large cohort of 1383 patients from 59 centers in Germany with newly diagnosed acute myeloid leukemia for whom 212 clinical, laboratory, cytogenetic and molecular genetic parameters were available.

RESULTS

Unsupervised learning identifies four distinct patient clusters, and statistical analysis shows significant differences in rate of complete remissions, event-free, relapse-free and overall survival between the four clusters. In comparison to the standard-of-care hypothesis-driven European Leukemia Net (ELN2017) risk stratification model, we find all three ELN2017 risk categories being represented in all four clusters in varying proportions indicating unappreciated complexity of AML biology in current established risk stratification models. Further, by using assigned clusters as labels we subsequently train a supervised model to validate cluster assignments on a large external multicenter cohort of 664 intensively treated AML patients.

CONCLUSIONS

Dynamic data-driven models are likely more suitable for risk stratification in the context of increasingly complex medical data than rigid hypothesis-driven models to allow for a more personalized treatment allocation and gain novel insights into disease biology.

Older Adults With Newly Diagnosed AML: Hot Topics for the Practicing Clinician

Over the past decade, our understanding of AML pathogenesis and pathophysiology has improved significantly with mutational profiling. This has led to translational advances in therapeutic options, as there have been 10 new US Food and Drug Administration (FDA) approvals for AML therapies since 2017, half of which target specific driver mutations in FLT3, IDH1, or IDH2. These new agents have expanded the therapeutic armamentarium for AML, particularly for patients who are considered ineligible for intensive chemotherapy with anthracycline- and cytarabine-containing regimens. These new treatment options are relevant because the median age at diagnosis is 68 years, and outcomes for patients older than 60 years have historically been dismal. However, the optimal approach to incorporating novel agents into frontline regimens remains a clinical challenge, particularly with regard to sequencing of therapies, considering the role of allogeneic hematopoietic stem cell transplantation and managing toxicities.

Recent advances in targeted therapies in acute myeloid leukemia

Acute myeloid leukemia (AML) is the most common acute leukemia in adults. While survival for younger patients over the last several decades has improved nearly sixfold with the optimization of intensive induction chemotherapy and allogeneic stem cell transplantation (alloHSCT), this effect has been largely mitigated in older and less fit patients as well as those with adverse-risk disease characteristics. However, the last 10 years has been marked by major advances in the molecular profiling of AML characterized by a deeper understanding of disease pathobiology and therapeutic vulnerabilities. In this regard, the classification of AML subtypes has recently evolved from a morphologic to a molecular and genetic basis, reflected by recent updates from the World Health Organization and the new International Consensus Classification system. After years of stagnation in new drug approvals for AML, there has been a rapid expansion of the armamentarium against this disease since 2017. Low-intensity induction therapy with hypomethylating agents and venetoclax has substantially improved outcomes, including in those previously considered to have a poor prognosis. Furthermore, targeted oral therapies against driver mutations in AML have been added to the repertoire. But with an accelerated increase in treatment options, several questions arise such as how to best sequence therapy, how to combine therapies, and if there is a role for maintenance therapy in those who achieve remission and cannot undergo alloHSCT. Moreover, certain subtypes of AML, such as those with TP53 mutations, still have dismal outcomes despite these recent advances, underscoring an ongoing unmet need and opportunity for translational advances. In this review, we will discuss recent updates in the classification and risk stratification of AML, explore the literature regarding low-intensity and novel oral combination therapies, and briefly highlight investigative agents currently in early clinical development for high-risk disease subtypes.

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.

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 [...].

CD86 Is Associated with Immune Infiltration and Immunotherapy Signatures in AML and Promotes Its Progression

Background

Cluster of differentiation 86 (CD86), also known as B7-2, is a molecule expressed on antigen-presenting cells that provides the costimulatory signals required for T cell activation and survival. CD86 binds to two ligands on the surface of T cells: the antigen CD28 and cytotoxic T lymphocyte-associated protein 4 (CTLA-4). By binding to CD28, CD86-together with CD80-promotes the participation of T cells in the antigen presentation process. However, the interrelationships among CD86, immunotherapy, and immune infiltration in acute myeloid leukemia (AML) are unclear.

Methods

The immunological effects of CD86 in various cancers (including on chemokines, immunostimulators, MHC, and receptors) were evaluated through a pan-cancer analysis using TCGA and GEO databases. The relationship between CD86 expression and mononucleotide variation, gene copy number variation, methylation, immune checkpoint blockers (ICBs), and T-cell inflammation score in AML was subsequently examined. ESTIMATE and limma packages were used to identify genes at the intersection of CD86 with StromalScore and ImmuneScore. Subsequently, GO/KEGG and PPI network analyses were performed. The immune risk score (IRS) model was constructed, and the validation set was used for verification. The predictive value was compared with the TIDE score.

Results

CD86 was overexpressed in many cancers, and its overexpression was associated with a poor prognosis. CD86 expression was positively correlated with the expression of CTLA4, PDCD1LG2, IDO1, HAVCR2, and other genes and negatively correlated with CD86 methylation. The expression of CD86 in AML cell lines was detected by QRT-PCR and Western blot, and the results showed that CD86 was overexpressed in AML cell lines. Immune infiltration assays showed that CD86 expression was positively correlated with CD8 T cell, Dendritic cell, macrophage, NK cell, and Th1_cell and also with immune examination site, immune regulation, immunotherapy response, and TIICs. ssGSEA showed that CD86 was enriched in immune-related pathways, and CD86 expression was correlated with mutations in the genes RB1, ERBB2, and FANCC, which are associated with responses to radiotherapy and chemotherapy. The IRS score performed better than the TIDE website score.

Conclusion

CD86 appears to participate in immune invasion in AML and is an important player in the tumor microenvironment in this malignancy. At the same time, the IRS score developed by us has a good effect and may provide some support for the diagnosis of AML. Thus, CD86 may serve as a potential target for AML immunotherapy.

Digging into the HLA pockets: A new association with acute leukaemias

Acute leukaemias represent a highly heterogeneous group of clonal proliferations of myeloid or lymphoid blasts. In the last decade, the contribution of immunogenetics to cancer biology has elicited a renewed interest in the structures of immune adaptive responses, due to the growing body of evidence concerning their involvement into disease pathogenesis and treatment. The report by Boukouaci and colleagues suggests new associations between patterns of distribution of specific human leukocyte antigen motifs and leukaemogenesis. Commentary on Boukouaci Wahid et al. Comparative analysis of the variability of the HLA peptide-binding pockets in patients with acute leukemias" Br J Haematol 2023;200:203-215.

Genomic profiling for clinical decision making in myeloid neoplasms and acute leukemia

Myeloid neoplasms and acute leukemias derive from the clonal expansion of hematopoietic cells driven by somatic gene mutations. Although assessment of morphology plays a crucial role in the diagnostic evaluation of patients with these malignancies, genomic characterization has become increasingly important for accurate diagnosis, risk assessment, and therapeutic decision making. Conventional cytogenetics, a comprehensive and unbiased method for assessing chromosomal abnormalities, has been the mainstay of genomic testing over the past several decades and remains relevant today. However, more recent advances in sequencing technology have increased our ability to detect somatic mutations through the use of targeted gene panels, whole-exome sequencing, whole-genome sequencing, and whole-transcriptome sequencing or RNA sequencing. In patients with myeloid neoplasms, whole-genome sequencing represents a potential replacement for both conventional cytogenetic and sequencing approaches, providing rapid and accurate comprehensive genomic profiling. DNA sequencing methods are used not only for detecting somatically acquired gene mutations but also for identifying germline gene mutations associated with inherited predisposition to hematologic neoplasms. The 2022 International Consensus Classification of myeloid neoplasms and acute leukemias makes extensive use of genomic data. The aim of this report is to help physicians and laboratorians implement genomic testing for diagnosis, risk stratification, and clinical decision making and illustrates the potential of genomic profiling for enabling personalized medicine in patients with hematologic neoplasms.

The role of next-generation sequencing in acute myeloid leukemia

PURPOSE OF REVIEW

The development of high-throughput techniques like next-generation sequencing (NGS) has unraveled the genetic profile of cancer. In this review, we discuss the role of NGS on the diagnostic, risk stratification, and follow-up of patients with acute myeloid leukemia (AML).

RECENT FINDINGS

NGS has become an essential tool in clinical practice for AML management. Therefore, efforts are being made to improve its applications, automation, and turnaround time. Other high-throughput techniques, such as whole genome sequencing or RNA-sequencing, can be also used to this end. However, not all institutions may be able to implement these approaches. NGS is being investigated for measurable residual disease (MRD) assessment, especially with the development of error-correction NGS. New data analysis approaches like machine learning are being investigated in order to integrate genomic and clinical data and develop comprehensive classifications and risk scores.

SUMMARY

NGS has proven to be a useful approach for the analysis of genomic alterations in patients with AML, which aids patient management. Current research is being directed at reducing turnaround time and simplifying processes so that these techniques can be universally integrated into clinical practice.

Diagnosis and management of AML in adults: 2022 recommendations from an international expert panel on behalf of the ELN

The 2010 and 2017 editions of the European LeukemiaNet (ELN) recommendations for diagnosis and management of acute myeloid leukemia (AML) in adults are widely recognized among physicians and investigators. There have been major advances in our understanding of AML, including new knowledge about the molecular pathogenesis of AML, leading to an update of the disease classification, technological progress in genomic diagnostics and assessment of measurable residual disease, and the successful development of new therapeutic agents, such as FLT3, IDH1, IDH2, and BCL2 inhibitors. These advances have prompted this update that includes a revised ELN genetic risk classification, revised response criteria, and treatment recommendations.

A Focus on Intermediate-Risk Acute Myeloid Leukemia: Sub-Classification Updates and Therapeutic Challenges

Acute myeloid leukemia (AML) represents a heterogeneous group of hematopoietic neoplasms deriving from the abnormal proliferation of myeloid progenitors in the bone marrow. Patients with AML may have highly variable outcomes, which are generally dictated by individual clinical and genomic characteristics. As such, the European LeukemiaNet 2017 and 2022 guidelines categorize newly diagnosed AML into favorable-, intermediate-, and adverse-risk groups, based on their molecular and cytogenetic profiles. Nevertheless, the intermediate-risk category remains poorly defined, as many patients fall into this group as a result of their exclusion from the other two. Moreover, further genomic data with potential prognostic and therapeutic influences continue to emerge, though they are yet to be integrated into the diagnostic and prognostic models of AML. This review highlights the latest therapeutic advances and challenges that warrant refining the prognostic classification of intermediate-risk AML.

The Anti-Leukemia Effect of Ascorbic Acid: From the Pro-Oxidant Potential to the Epigenetic Role in Acute Myeloid Leukemia

Data derived from high-throughput sequencing technologies have allowed a deeper understanding of the molecular landscape of Acute Myeloid Leukemia (AML), paving the way for the development of novel therapeutic options, with a higher efficacy and a lower toxicity than conventional chemotherapy. In the antileukemia drug development scenario, ascorbic acid, a natural compound also known as Vitamin C, has emerged for its potential anti-proliferative and pro-apoptotic activities on leukemic cells. However, the role of ascorbic acid (vitamin C) in the treatment of AML has been debated for decades. Mechanistic insight into its role in many biological processes and, especially, in epigenetic regulation has provided the rationale for the use of this agent as a novel anti-leukemia therapy in AML. Acting as a co-factor for 2-oxoglutarate-dependent dioxygenases (2-OGDDs), ascorbic acid is involved in the epigenetic regulations through the control of TET (ten-eleven translocation) enzymes, epigenetic master regulators with a critical role in aberrant hematopoiesis and leukemogenesis. In line with this discovery, great interest has been emerging for the clinical testing of this drug targeting leukemia epigenome. Besides its role in epigenetics, ascorbic acid is also a pivotal regulator of many physiological processes in human, particularly in the antioxidant cellular response, being able to scavenge reactive oxygen species (ROS) to prevent DNA damage and other effects involved in cancer transformation. Thus, for this wide spectrum of biological activities, ascorbic acid possesses some pharmacologic properties attractive for anti-leukemia therapy. The present review outlines the evidence and mechanism of ascorbic acid in leukemogenesis and its therapeutic potential in AML. With the growing evidence derived from the literature on situations in which the use of ascorbate may be beneficial in vitro and in vivo, we will finally discuss how these insights could be included into the rational design of future clinical trials.

Reinforcement Learning for Precision Oncology

Precision oncology is grounded in the increasing understanding of genetic and molecular mechanisms that underly malignant disease and offer different treatment pathways for the individual patient. The growing complexity of medical data has led to the implementation of machine learning techniques that are vastly applied for risk assessment and outcome prediction using either supervised or unsupervised learning. Still largely overlooked is reinforcement learning (RL) that addresses sequential tasks by exploring the underlying dynamics of an environment and shaping it by taking actions in order to maximize cumulative rewards over time, thereby achieving optimal long-term outcomes. Recent breakthroughs in RL demonstrated remarkable results in gameplay and autonomous driving, often achieving human-like or even superhuman performance. While this type of machine learning holds the potential to become a helpful decision support tool, it comes with a set of distinctive challenges that need to be addressed to ensure applicability, validity and safety. In this review, we highlight recent advances of RL focusing on studies in oncology and point out current challenges and pitfalls that need to be accounted for in future studies in order to successfully develop RL-based decision support systems for precision oncology.

How to Improve Prognostication in Acute Myeloid Leukemia with CBFB-MYH11 Fusion Transcript: Focus on the Role of Molecular Measurable Residual Disease (MRD) Monitoring

Acute myeloid leukemia (AML) carrying inv(16)/t(16;16), resulting in fusion transcript CBFB-MYH11, belongs to the favorable-risk category. However, even if most patients obtain morphological complete remission after induction, approximately 30% of cases eventually relapse. While well-established clinical features and concomitant cytogenetic/molecular lesions have been recognized to be relevant to predict prognosis at disease onset, the independent prognostic impact of measurable residual disease (MRD) monitoring by quantitative real-time reverse transcriptase polymerase chain reaction (qRT-PCR), mainly in predicting relapse, actually supersedes other prognostic factors. Although the ELN Working Party recently indicated that patients affected with CBFB-MYH11 AML should have MRD assessment at informative clinical timepoints, at least after two cycles of intensive chemotherapy and after the end of treatment, several controversies could be raised, especially on the frequency of subsequent serial monitoring, the most significant MRD thresholds (most commonly 0.1%) and on the best source to be analyzed, namely, bone marrow or peripheral blood samples. Moreover, persisting low-level MRD positivity at the end of treatment is relatively common and not predictive of relapse, provided that transcript levels remain stably below specific thresholds. Rising MRD levels suggestive of molecular relapse/progression should thus be confirmed in subsequent samples. Further prospective studies would be required to optimize post-remission monitoring and to define effective MRD-based therapeutic strategies.