Evaluation of a machine-learning model based on laboratory parameters for the prediction of acute leukaemia subtypes: a multicentre model development and validation study in France.

BACKGROUND: Acute leukaemias are life-threatening haematological cancers characterised by the infiltration of transformed immature haematopoietic cells in the blood and bone marrow. Prompt and accurate diagnosis of the three main acute leukaemia subtypes (ie acute lymphocytic leukaemia [ALL], acute myeloid leukaemia [AML], and acute promyelocytic leukaemia [APL]) is of utmost importance to guide initial treatment and prevent early mortality but requires cytological expertise that is not always available. We aimed to benchmark different machine-learning strategies using a custom variable selection algorithm to propose an extreme gradient boosting model to predict leukaemia subtypes on the basis of routine laboratory parameters. METHODS: This multicentre model development and validation study was conducted with data from six independent French university hospital databases. Patients aged 18 years or older diagnosed with AML, APL, or ALL in any one of these six hospital databases between March 1, 2012, and Dec 31, 2021, were recruited. 22 routine parameters were collected at the time of initial disease evaluation; variables with more than 25% of missing values in two datasets were not used for model training, leading to the final inclusion of 19 parameters. The performances of the final model were evaluated on internal testing and external validation sets with area under the receiver operating characteristic curves (AUCs), and clinically relevant cutoffs were chosen to guide clinical decision making. The final tool, Artificial Intelligence Prediction of Acute Leukemia (AI-PAL), was developed from this model. FINDINGS: 1410 patients diagnosed with AML, APL, or ALL were included. Data quality control showed few missing values for each cohort, with the exception of uric acid and lactate dehydrogenase for the cohort from Hôpital Cochin. 679 patients from Hôpital Lyon Sud and Centre Hospitalier Universitaire de Clermont-Ferrand were split into the training (n=477) and internal testing (n=202) sets. 731 patients from the four other cohorts were used for external validation. Overall AUCs across all validation cohorts were 0·97 (95% CI 0·95-0·99) for APL, 0·90 (0·83-0·97) for ALL, and 0·89 (0·82-0·95) for AML. Cutoffs were then established on the overall cohort of 1410 patients to guide clinical decisions. Confident cutoffs showed two (0·14%) wrong predictions for ALL, four (0·28%) wrong predictions for APL, and three (0·21%) wrong predictions for AML. Use of the overall cutoff greatly reduced the number of missing predictions; diagnosis was proposed for 1375 (97·5%) of 1410 patients for each category, with only a slight increase in wrong predictions. The final model evaluation across both the internal testing and external validation sets showed accuracy of 99·5% for ALL diagnosis, 98·8% for AML diagnosis, and 99·7% for APL diagnosis in the confident model and accuracy of 87·9% for ALL diagnosis, 86·3% for AML diagnosis, and 96·1% for APL diagnosis in the overall model. INTERPRETATION: AI-PAL allowed for accurate diagnosis of the three main acute leukaemia subtypes. Based on ten simple laboratory parameters, its broad availability could help guide initial therapies in a context where cytological expertise is lacking, such as in low-income countries. FUNDING: None.

Real life evaluation of AlphaMissense predictions in hematological malignancies.

High-throughput sequencing plays a pivotal role in hematological malignancy diagnostics, but interpreting missense mutations remains challenging. In this study, we used the newly available AlphaMissense database to assess the efficacy of machine learning to predict missense mutation effects and its impact to improve our ability to interpret them. Based on the analysis of 2073 variants from 686 patients analyzed for clinical purpose, we confirmed the very high accuracy of AlphaMissense predictions in a large real-life data set of missense mutations (AUC of ROC curve 0.95), and provided a comprehensive analysis of the discrepancies between AlphaMissense predictions and state of the art clinical interpretation.

Targeted High-throughput Sequencing for Hematological Malignancies: A GBMHM Survey of Practice and Cost Evaluation in France.

The objective of this study was to assess the clinical impact and financial costs of next-generation sequencing (NGS) in 5 categories of pediatric and adult hematological cancers. NGS prescriptions were prospectively collected from 26 laboratories, with varied technical and reporting practice (all or only significant targets). Impact was defined by the identification of (1) an actionable mutation, (2) a mutation with prognostic and/or theranostic value, and/or (3) a mutation allowing nosological refinement, reported by local investigators. A microcosting study was undertaken in 4 laboratories, identifying the types and volumes of resources required for each procedural step. Individual index prescriptions for 3961 patients were available for impact analysis on the management of myeloid disorders (two thirds) and, mainly mature B, lymphoid disorders (one third). NGS results were considered to impact the management for 73.4% of prescriptions: useful for evaluation of prognostic risk in 34.9% and necessary for treatment adaptation (actionable) in 19.6%, but having no immediate individual therapeutic impact in 18.9%. The average overall cost per sample was 191 € for the restricted mature lymphoid amplicon panel. Capture panel costs varied from 369 € to 513 €. Unit costs varied from 0.5 € to 5.7 € per kb sequenced, from 3.6 € to 11.3 € per target gene/hot-spot sequenced and from 4.3 € to 73.8 € per target gene/hot-spot reported. Comparable costs for the Amplicon panels were 5-8 € per kb and 10.5-14.7 € per target gene/hot-spot sequenced and reported, demonstrating comparable costs with greater informativity/flexibility for capture strategies. Sustainable funding of precision medicine requires a transparent discussion of its impact on care pathways and its financial aspects.

[Preventable cancers: Is it enough to change our behaviour?] / Cancers évitables - Suffit-il de changer nos comportements ?

In France, part of 40 % of preventable cancers can be attributed to lifestyle habits. Epidemiological data show that occupational exposures are a major cause of these cancers. However, despite this evidence, the prevention actions promoted by public authorities are focused on changing individual behaviors. In this article, we seek to understand the reasons of the erasure of the role of socio-environmental factors in cancer prevention discourse.

Title: Cancers évitables - Suffit-il de changer nos comportements ? Abstract: En France, une partie des 40 % de cancers évitables peut être attribuée aux habitudes de vie. Les données épidémiologiques révèlent que les expositions professionnelles et domestiques à des substances cancérogènes sont aussi responsables d'une fraction significative de ces cancers. Pourtant, en dépit de ces évidences, les actions de prévention promues par les pouvoirs publics se focalisent sur le changement des comportements individuels. Dans cet article, nous cherchons à comprendre les raisons de l'effacement de la place des facteurs socio-environnementaux des discours portant sur la prévention des cancers.

C/EBPα Confers Dependence to Fatty Acid Anabolic Pathways and Vulnerability to Lipid Oxidative Stress-Induced Ferroptosis in FLT3-Mutant Leukemia.

Although transcription factor CCAAT-enhancer binding protein α (C/EBPα) is critical for normal and leukemic differentiation, its role in cell and metabolic homeostasis is largely unknown in cancer. Here, multiomics analyses uncovered a coordinated activation of C/EBPα and Fms-like tyrosine kinase 3 (FLT3) that increased lipid anabolism in vivo and in patients with FLT3-mutant acute myeloid leukemia (AML). Mechanistically, C/EBPα regulated the fatty acid synthase (FASN)-stearoyl-CoA desaturase (SCD) axis to promote fatty acid (FA) biosynthesis and desaturation. We further demonstrated that FLT3 or C/EBPα inactivation decreased monounsaturated FA incorporation to membrane phospholipids through SCD downregulation. Consequently, SCD inhibition enhanced susceptibility to lipid redox stress that was exploited by combining FLT3 and glutathione peroxidase 4 inhibition to trigger lipid oxidative stress, enhancing ferroptotic death of FLT3-mutant AML cells. Altogether, our study reveals a C/EBPα function in lipid homeostasis and adaptation to redox stress, and a previously unreported vulnerability of FLT3-mutant AML to ferroptosis with promising therapeutic application. SIGNIFICANCE: FLT3 mutations are found in 30% of AML cases and are actionable by tyrosine kinase inhibitors. Here, we discovered that C/EBPα regulates FA biosynthesis and protection from lipid redox stress downstream mutant-FLT3 signaling, which confers a vulnerability to ferroptosis upon FLT3 inhibition with therapeutic potential in AML. This article is highlighted in the In This Issue feature, p. 1501.

VEXAS: is it time to reshape the nosology of clonal hematopoiesis?

INTRODUCTION: The recent description of VEXAS (for vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) challenges the nosological framework of myelodysplastic syndromes. AREAS COVERED: Clonal expansion driven by somatic mutations in cancer genes has been largely described in healthy aging individuals. Regarding hematopoiesis, the prevalence of clonal hematopoiesis has blurred the line between normal and pathological, especially for the definition of myelodysplastic syndromes. VEXAS syndrome further challenges the nosology as this clonal disease of hematopoiesis is also associated with dysplastic features and cytopenias. EXPERT OPINION: In this perspective, we discuss whether VEXAS should be considered a genuine myelodysplastic syndrome and propose a conceptual framework to refine the nosology, based on the distinction of clonal hematopoiesis of indeterminate potential (CHIP), clonal hematopoiesis of hematological significance, and clonal hematopoiesis of other significance.

Next-CLL, a New Next-Generation Sequencing-Based Method for Assessment of IGHV Gene Mutational Status in Chronic Lymphoid Leukemia.

Current guidelines for patients with chronic lymphocytic leukemia (CLL) recommend mutation status determination of the clonotypic IGHV gene before treatment initiation to guide the choice of first-line therapy. Currently, commercially available next-generation sequencing (NGS) solutions have technical constraints, as they necessitate at least a 2 × 300 bp sequencing, which restricts their use for routine practice. The cost of the commercial kits also represents an important drawback. We present a new method called Next-CLL, a ready-to-use strategy to evaluate IGHV gene mutation status using any NGS device (including 2 × 150 bp sequencers) in routine diagnostic laboratories. The performance of Next-CLL was validated on genomic DNA and cDNA obtained from 80 patients with CLL at diagnosis. Next-CLL identified a productive clone in 100% of cases, whereas PCR with Sanger sequencing led to a 12.5% failure rate. Next-CLL had 100% concordance with the reference technique for IGHV gene identification and allowed assessment of the IGHV mutation status from the leader sequence, following international guidelines. Comparing a large retrospective series of samples, analyzed by using Sanger sequencing (n = 773) or Next-CLL (n = 352), showed no bias in IGHV usage or mutational status, further validating our strategy in the real-life setting. Next-CLL represents a straightforward workflow for IGHV analysis in routine practice to assess clonal architecture and prognosis of patients with CLL.

VEXAS: where do we stand 2 years later?

PURPOSE OF REVIEW: Two years after the recognition of VEXAS (for Vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) syndrome, we propose an extensive review of the current understanding of VEXAS pathophysiology and therapeutic options. RECENT FINDINGS: Among the nearly 150 articles published about VEXAS, some have provided determinant insights into VEXAS pathophysiology and treatment. Clinical data from retrospective series support the JAK inhibitor ruxolitinib as the most efficient strategy to control inflammation, and interesting results were also described with azacytidine. Allogeneic stem cell transplantation remains the only curative option, but should be proposed to carefully selected patients. SUMMARY: Although waiting for more robust evidence from prospective clinical trials, therapeutic options emerge from retrospective studies. We propose a set of criteria that should be systematically reported to harmonize the evaluation of therapeutic outcomes. This will allow the collection of high-quality data and facilitate their subsequent meta-analysis with the overall aim of improving the management of VEXAS patients.

Real-life challenges using personalized prognostic scoring systems in acute myeloid leukemia.

Personalized medicine is a challenge for patients with acute myeloid leukemia (AML). The identification of several genetic mutations in several AML trials led to the creation of a personalized prognostic scoring algorithm known as the Knowledge Bank (KB). In this study, we assessed the prognostic value of this algorithm on a cohort of 167 real life AML patients. We compared KB predicted outcomes to real-life outcomes. For patients younger than 60-year-old, OS was similar in favorable and intermediate ELN risk category. However, KB algorithm failed to predict OS for younger patients in the adverse ELN risk category and for patients older than 60 years old in the favorable ELN risk category. These discrepancies may be explained by the emergence of several new therapeutic options as well as the improvement of allogeneic stem cell transplantation (aHSCT) outcomes and supportive cares. Personalized medicine is a major challenge and predictions models are powerful tools to predict patient's outcome. However, the addition of new therapeutic options in the field of AML requires a prospective validation of these scoring systems to include recent therapeutic innovations.

BTG1 inactivation drives lymphomagenesis and promotes lymphoma dissemination through activation of BCAR1.

Understanding the functional role of mutated genes in cancer is required to translate the findings of cancer genomics into therapeutic improvement. BTG1 is recurrently mutated in the MCD/C5 subtype of diffuse large B-cell lymphoma (DLBCL), which is associated with extranodal dissemination. Here, we provide evidence that Btg1 knock out accelerates the development of a lethal lymphoproliferative disease driven by Bcl2 overexpression. Furthermore, we show that the scaffolding protein BCAR1 is a BTG1 partner. Moreover, after BTG1 deletion or expression of BTG1 mutations observed in patients with DLBCL, the overactivation of the BCAR1-RAC1 pathway confers increased migration ability in vitro and in vivo. These modifications are targetable with the SRC inhibitor dasatinib, which opens novel therapeutic opportunities in BTG1 mutated DLBCL.

Impact on Outcome of Minimal Residual Disease after Hematopoietic Stem Cell Transplantation with Fludarabine, Amsacrine, and Cytosine Arabinoside-Busulfan Conditioning: A Retrospective Monocentric Study.

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) after conditioning with a sequential association of fludarabine, amsacrine, and cytosine arabinoside (FLAMSA) followed by a reduced-intensity conditioning regimen has emerged for patients with high-risk acute myeloid leukemia (AML), especially in refractory or relapsing patients. Here we aimed to address retrospectively the impact of pretransplantation minimal residual disease (MRD) by flow cytometry on the outcomes of high-risk AML patients who underwent allo-HSCT after sequential FLAMSA-busulfan (FLAMSA-Bu)-based conditioning regimens. We included 165 high-risk AML patients who underwent transplantation after FLAMSA-BU in this retrospective single-center "real life" study. All patients received in vivo T cell depletion with antithymocyte globulin (5 mg/kg). MRD detection was based on a leukemia-associated immunophenotype using the European LeukemiaNet recommendations, with a threshold of .1%. Univariate and multivariate analyses were performed using R version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria). With a median follow-up of 4.0 years post-transplantation, the median overall survival (OS) was 54.9 months. Overall, 41 patients (24.8%) relapsed post-transplantation, with a resulting cumulative incidence of relapse (CIR) of 26.7% at 2 years and 34.0% at 5 years. Detectable MRD preceding allo-HSCT and refractory status were associated with worse median OS and CIR rates compared with patients without detectable MRD; however, OS was not significantly different between pre-HSCT MRD-positive and refractory patients (median, .7 year versus 2.0 years; P = .3). Conversely, pre-HSCT MRD negativity was associated with a reduced 2-year CIR. Neither European LeukemiaNet risk stratification nor age had a significant influence on OS. In the multivariate analysis, only pre-HSCT MRD positivity and lower conditioning regimen intensity were significantly associated with a poorer OS. The cumulative incidence of extensive chronic graft-versus-host disease at 2 years was 26.15%. The estimated nonrelapse mortality (NRM) of the entire cohort at 2 years was 23.1%, with age and unrelated donor identified as risk factors for higher NRM. Our data ahow that FLAMSA-Bu conditioning did not reverse the pejorative effect of detectable pre-HSCT MRD, suggesting that such patients should be offered alternative strategies before HSCT to reach deeper remission.

Panel Informativity Optimizer: An R Package to Improve Cancer Next-Generation Sequencing Panel Informativity.

Mutation detection by next-generation sequencing is routinely used for cancer diagnosis. Selecting an optimal set of genes for a given cancer is not trivial as it has to optimize informativity (ie, the number of patients with at least one mutation in the panel), while minimizing panel length to reduce sequencing costs and increase sensitivity. We propose herein Panel Informativity Optimizer (PIO), an open-source software developed as an R package with a user-friendly graphical interface to help optimize cancer next-generation sequencing panel informativity. Using patient-level mutational data from either private data sets or preloaded data set of 91 independent cohorts from 31 different cancer types, PIO selects an optimal set of genomic intervals to maximize informativity and panel size in a given cancer type. Different options are offered, such as the definition of genomic intervals at the gene or exon level and the use of optimization strategy at the patient or patient per kilobase level. PIO can also propose an optimal set of genomic intervals to increase informativity of custom panels. A panel tester function is also available for panel benchmarking. Using public databases, as well as data from real-life settings, we demonstrate that PIO allows panel size reduction of up to 1000 kb, and accurately predicts the performance of custom or commercial panels.

[Oncology and psychiatry: Towards mutual connections/relationships]. / Oncologie et psychiatrie - Pour une relation réciproque féconde.

Oncology has been proposed as a model of scientificity, in order to promote scientific approaches to psychiatry. In this article, another type of relation between oncology and psychiatry is explored, which promotes the idea of a mutually enriching dialogue and underlines the contributions of psychiatry to oncology. The ways in which both fields address epistemological and ethical issues in their respective approaches to disease is also examined. We argue that these two disciplines can learn from one another in the common context of chronic conditions, thanks to the potential of big data collection and their biostatistics treatment for the identification of markers - sources of individualization -, as well as thanks to the renewed attention given to the temporal and processual dimension of these diseases, in particular within the framework of "staging" models.

Title: Oncologie et psychiatrie - Pour une relation réciproque féconde. Abstract: L'oncologie est souvent considérée par les défenseurs d'une psychiatrie scientifique comme un modèle médical à imiter. Psychiatres, oncologues et philosophes, nous proposons dans cet article une autre manière d'envisager les relations entre oncologie et psychiatrie, en promouvant l'intérêt d'un dialogue entre ces disciplines, convaincus de leur potentiel enrichissement réciproque et, en particulier, des apports possibles de la psychiatrie à l'oncologie. Nous proposons d'étudier chacune des manières de faire face aux difficultés épistémologiques et éthiques rencontrées dans l'approche des maladies. Nous pensons que psychiatrie et oncologie peuvent apprendre l'une de l'autre, dans le contexte commun de maladies chroniques qu'il s'agit de gérer plutôt que de guérir, grâce au potentiel qu'offrent le recueil de données massives et leur traitement biostatistique pour l'identification de marqueurs permettant d'individualiser les traitements, ainsi que grâce à l'attention renouvelée accordée à la dimension temporelle et processuelle de ces maladies, notamment dans le cadre de modèles de « stadification ¼ (ou staging).