Peripheral blood smears remain a cornerstone in the diagnosis of hematological neoplasms, offering rapid and valuable insights that inform subsequent diagnostic steps. However, since neoplastic transformations typically arise in the bone marrow, they may not manifest as detectable aberrations in peripheral blood, presenting a diagnostic challenge. In this paper, we introduce cAItomorph, an explainable transformer-based AI model, trained to classify hematological malignancies based on peripheral blood cytomorphology. Our data comprises peripheral blood single-cell images from 6115 patients with diagnoses confirmed by cytomorphology, cytogenetics, molecular genetics, and immunophenotyping from bone marrow samples, and 495 healthy controls, eight coarse classes. cAItomorph leverages the DinoBloom hematology foundation model and aggregates image encodings via a transformer-based architecture into a single vector. It achieves an overall accuracy of 0.72 in eight disease classification, with F1 scores of 0.76 for acute leukemia, 0.80 for myeloproliferative neoplasms and 0.94 for healthy cases. The overall accuracy increases to 0.87 in top-2 predictions. cAItomorph achieves high sensitivity for acute leukemia cases in external test sets. By analyzing attention heads, we demonstrate clinically relevant cell-level attentions in both internal and external test sets. Moreover, our model's calibrated prediction probabilities reduce the false discovery rate from 13.5% to 8.7% without missing any acute leukemia cases, thereby decreasing the number of unnecessary bone marrow aspirations based on peripheral blood smears. This study highlights the potential of AI-assisted diagnostics in hematological malignancies, illustrating how models trained on real-world data could enhance diagnostic accuracy and reduce invasive procedures.

Bispecific T-cell engagers (BiTE® molecules) have transformed the treatment of B-cell malignancies, yet clinical activity in AML has been modest. Resistance is driven in part by the genetic heterogeneity of AML, most notably TP53 mutations, present in 10-15% of de novo and up to 25% of therapy-related AML. Thus, we hypothesized that TP53 aberrations in AML contribute to cell-intrinsic and extrinsic resistance against T-cell-based immunotherapy. Cytotoxicity against TP53-deleted (DEL) primary AML cells and TP53-knockdown (KD) AML cell lines was reduced in co-cultures with T cells stimulated with the BiTE molecule AMG 330 (CD3×CD33). In addition, T-cell proliferation and proinflammatory cytokine secretion was impaired in co-cultures with TP53 KD cells. Transwell assays identified the secretome of TP53 KD AML cells as a key contributor to the immunosuppressive effects. Proteomic analysis revealed TGF-β1 in TP53 KD co-cultures as a mediator of T-cell suppression. RNA sequencing of T cells co-cultured with TP53 KD cells uncovered a transcriptional shift toward a senescent cell cycle profile. Our data collectively identify the immunosuppressive secretome of TP53-deficient AML as a key barrier to T-cell-engaging immunotherapies, underscoring an unmet clinical need for strategies able to restore T-cell function in TP53 KD AML.

Community-acquired respiratory viruses (CARV), such as influenza-, parainfluenza- or respiratory syncytial virus, pose a significant threat to immunocompromised patients with cancer. Following the COVID-19 pandemic, SARS-CoV-2 has now joined the ranks of endemic respiratory viruses and continues to be a cause of significant morbidity and mortality in patients with cancer. Strategies to protect this vulnerable patient population both by prevention of infection and by early therapeutic intervention in case of infectious disease are therefore of utmost importance. This guideline provides updated evidence-based recommendations on diagnosis, prophylaxis and treatment of CARV infections including COVID-19 in patients with solid tumors or hematologic malignancies to support clinicians in offering optimal care. The guideline is based on a systematic review of currently available data and was developed until the beginning of 2025 by an expert panel of the Infectious Diseases Working Party (AGIHO) of the German Society for Hematology and Medical Oncology (DGHO).

Introduction Febrile neutropenia, a common complication of systemic antineoplastic therapy, varies in risk depending on malignant disease, treatment, and patient factors. The risk increases with the depth and duration of neutropenia and can be reduced with prophylactic use of G-CSF. International guidelines are conflicting in several aspects and do not reflect all patient groups. We therefore updated the 2014 Infectious Diseases Working Party (AGIHO) guideline of the German Society of Hematology and Medical Oncology (DGHO) on evidence-based recommendations for the use of G-CSF in patients with cancer. Materials and Methods After a systematic literature search from January 2014 to December 2024 on PubMed and Medline several consensus meetings were held by an expert panel of the AGIHO to evaluate, discuss and consent on the level of evidence and strength of recommendation for the use of G-CSF in patients with cancer based on ESCMID criteria. Results Results from eligible (randomized) studies were grouped in evidence tables. Recommendations for different entities, risk groups, new G-CSF formulations and emerging strategies in immunotherapy were set up and put into comparison to 2014. Discussion Comprehensive literature search and expert panel consensus confirm most of the key recommendations of 2014. New recommendations on the use of G-CSF in immunotherapy settings provide insight and support for day-by-day clinical decision making in the care of patients with cancer.

Secondary hemophagocytic lymphohistiocytosis (sHLH) is a life-threatening hyperinflammatory condition. While few diagnostic scores are established, none exist to predict both clinical course and time-point specific outcome of sHLH patients so far. We present a machine learning (ML)-based tool to predict Initial Disease Severity (IDS; defined as admission to intensive care units (ICU) OR death < 90 days without ICU admission) and mortality across different time points in sHLH patients. 167 adult sHLH patients from six study centers across three European countries were included retrospectively. Clinical and demographic features, course, survival, and laboratory data were assessed. Random forest models were trained with two sets of eight clinical and laboratory features: one to predict IDS, and five to predict mortality at distinct time points (30, 60, 90, 180 or 365 days). After calibration, the models were tested against hold-out test sets containing n = 32 (IDS) or n = 43 (mortality) sHLH patients. Overall, the models demonstrated strong discriminatory ability, overall performance, and accurate prediction of risk. Serum levels of the soluble interleukin-2 receptor (sIL-2R) and albumin (for IDS) or sIL-2R and platelet counts (for mortality prediction) showed the strongest contributions to the models’ predictions. The HLH-Risk-Calculator is an exploratory tool predicting the clinical course of sHLH. External validation is critical to assess its validity, applicability, and robustness for real-world use. To this end, the calculator is available at www.hlh-risk-calculator.com for research use only, and is currently not intended for clinical decision-making.

Refractory disease and relapse are major challenges in acute myeloid leukemia (AML) therapy attributed to survival of leukemic stem cells (LSC). To target LSCs, antibody-drug conjugates (ADCs) provide an elegant solution, combining the specificity of antibodies with highly potent payloads. We aimed to investigate if FLT3-20D9h3-ADCs delivering either the DNA-alkylator duocarmycin (DUBA) or the microtubule-toxin monomethyl auristatin F (MMAF) can eradicate quiescent LSCs. We show here that DUBA more potently kills cell-cycle arrested AML cells compared to microtubule-targeting auristatins. Due to limited stability of 20D9h3-DUBA ADC in vivo, we analyzed both ADCs in advanced in vitro stem cell assays. 20D9h3-DUBA successfully eliminated leukemic progenitors in vitro in colony-forming unit and long-term culture initiating cell assays, both in patient cells and in patient-derived xenograft (PDX) cells. Further, it completely prevented engraftment of AML PDX leukemia-initiating cells in NSG mice. 20D9h3-MMAF had a similar effect in engraftment assays, but a less prominent effect in colony assays. Both ADCs did not affect healthy stem and progenitor cells at comparable doses providing the rationale for FLT3 as therapeutic LSC target. Collectively, we show that FLT3-directed ADCs with DUBA or MMAF have potent activity against AML LSCs and represent promising candidates for further clinical development.

Acute myeloid leukemia (AML) is a complex disease characterized by diverse molecular pathogenesis. Genetic alterations, including germline and somatic variants in the DEAD box helicase 41 gene ( DDX41) located on chromosome 5 play an increasingly recognized role. Recent reports indicate that 5% of intensively treated adult AML patients harbor DDX41 germline mutations ( DDX41MutGL), and their precise impact remains incompletely understood. These studies suggest that DDX41MutGL may define a distinct biological subgroup, associated with e.g. older age, male gender, low blast, and low white blood cell count (WBC). To further elucidate the role of DDX41 in AML, we performed a retrospective analysis of 906 unselected adult AML patients from the AML Cooperative Group (AMLCG) registry (2015-2022), by targeted sequencing. As DDX41MutGL are typically “null mutations”, leading to reduced DDX41 expression, we additionally investigated whether DDX41 gene expression correlates or resembles the observed germline phenotype. Our analysis encompassed >1000 independent gene expression profiles (GSE37642, GSE14468, and GSE106291) of intensively treated adult AML patients. We identified a sub-cohort with low DDX41 ( DDX41-low) expression and re-analyzed them with next-generation sequencing to detect DDX41MutGL. Additionally, we correlated gene expression data with 198 DNA methylation profiles (from patients who had undergone both analyses) to identify potential epigenetic mechanisms underlying DDX41-low expression. Among the 906 patients (median age 61 years; range 18-98 years), we identified 11 unrelated individuals with suspected DDX41MutGL (VAF > 40%). Notably, the overall frequency of DDX41MutGL in the German unselected AML patient population was merely 1%, considerably lower than previously reported (p<0.0001 in comparison to Duployez et al. 2022). Characteristically, DDX41MutGL patients in this cohort were mostly male (73%), with a median age of 67 years displaying low WBC (average 1,3G/l) and a normal karyotype (80%) at initial diagnosis. Remarkably, our analysis of independent large gene expression cohorts revealed a pattern of phenotypic association in patients with decreased DDX41 expression ( DDX41-low) closely resembling those with DDX41MutGL. These associations were consistent across different datasets and included e.g. older age, low WBC, and low blast count. Furthermore, DDX41-low patients had poor overall survival. The DDX41-low subgroup constituted almost 10% of all AML patients, surpassing the reported frequencies of DDX41MutGL. In an exploratory pilot study, we screened 48 patients exhibiting the lowest DDX41 gene expression by sequencing the coding regions for DDX41 mutations but found none, suggesting alternative mechanisms, such as copy number changes, non-coding alterations or aberrant DNA methylation patterns that may replicate the phenotypic effects associated with DDX41MutGL. To further elucidate this mechanism, we correlated the gene expression of DDX41 with all DNA methylation loci (CpG sites covered by the EPIC array) on chromosome 5 using Spearman correlation analysis (GSE106291). We plotted the correlations against the sorted chromosome length, identifying a peak at a specific location corresponding to the PCDH (Protocadherin) cluster (Figure). Thus, DDX41-low was associated with reduced methylation at the PCDH loci and linked to higher PCDH gene expression. Across several data sets, we confirmed the association between higher PCDH cluster gene expression and low DDX41 expression. In summary, our study reveals that the frequency of DDX41MutGL in an unselected population of German AML patients is considerably lower than previously reported. Additionally, we found that apart from DDX41MutGL, DDX41-low exhibits a comparable clinical profile and is linked to specific methylation and gene expression patterns, with a notable emphasis on the PCDH complex located on chromosome 5. Understanding the functional implications of PCDH genes and their interaction with DDX41 in AML may significantly advance our knowledge of AML pathogenesis. Gaining insights into DDX41 alterations may open avenues for personalized therapeutic approaches and further prognostic stratification for AML patients with distinct molecular characteristics.

Introduction Patients (pts) with core-binding factor (CBF) AML (defined by inv(16)(p13.1q22) or t(16;16)(p13.1;q22)/CBFB::MYH11 or t(8;21)(q22;q22.1)/RUNX1::RUNX1T1 gene fusions) are considered to have favorable risk disease when treated with intensive chemotherapy regimens. However, since these pts were excluded from pivotal trials evaluating venetoclax in combination with hypomethylating agents (HMA/Ven) or low-dose cytarabine (LDAC/Ven), little is known about the efficacy of these regimens in CBF AML pts ineligible for intensive treatment. Case series suggest that pts with CBFB::MYH11 respond well to HMA/Ven, whereas those with RUNX1::RUNX1T1 may be less responsive (Zhang et al., 2023, Shen et al., 2025). Given the lack of approved and effective therapies for older or frail pts with CBF AML besides HMA/Ven, this study aimed to assess its efficacy in this specific patient population. Methods Based on data from seven European study group registries (PETHEMA, DATAML, CELL, ALFA, NCRI, PALG, and SAL), we retrospectively analyzed adult pts with newly diagnosed CBF AML treated with HMA/Ven. CBF fusions were identified using conventional karyotyping, FISH, PCR and/or NGS. For response evaluation, ELN 2022 criteria were used, and best response within six HMA/Ven cycles is reported. The median overall survival (mOS) was estimated using the Kaplan-Meier method. Statistical analyses were performed using GraphPad Prism v.10.5.0. Results We identified 53 pts with newly diagnosed CBF AML (27 (51%) with CBFB::MYH11 and 26 (49%) with RUNX1::RUNX1T1) treated with HMA/Ven between September 2019 and July 2025. The median age of the entire cohort was 76 years (range, 30-86), and 21 pts (40%) were female. ECOG performance status was ≥2 in 17 pts (35%). Eight pts (20%) had AML post-cytotoxic therapy. According to the European LeukemiaNet (ELN) 2024 risk classification, 70%, 28%, and 2% of pts had favorable, intermediate, and adverse risk disease, respectively. The most common secondary genetic alterations were mutations in NRAS (31%), TET2 (30%), KIT (29%), KRAS (23%) and DNMT3A (14%). Except for a higher KRAS mutation prevalence in pts with CBFB::MYH11 (44% vs. 5%; p=0.051), there were no significant differences in baseline characteristics between CBFB::MYH11 or RUNX1::RUNX1T1 subgroups. None of the pts had previously been exposed to HMA or Ven, and most pts (95%) received Ven in combination with azacitidine. The median planned Ven schedule (cycle one) was 21 days (range, 7-28), and the median number of administered HMA/Ven cycles was 4 (range, 1-24). Two pts underwent subsequent allogeneic hemopoietic cell transplantation. The composite complete remission rate (CRc, CR + CRi) of the entire cohort was 69%, and was similar for pts with CBFB::MYH11 orRUNX1::RUNX1T1 fusions(67% vs. 72%,p=0.564). Pts with ELN 2024 intermediate risk mutations (i.e., FLT3-ITD, NRAS and/or KRAS) had a CRc rate similar to those without any of these variants (68% vs. 73%,p=0.752). Also, the CRc rate of pts with KIT mutations was comparable to wild-type pts (68% vs 79%,p=0.29). The 60-day mortality rate was 8%. After a median follow-up of 14 months, the mOS of the entire cohort was 13 months, with corresponding 12-month and 24-month OS rates of 54% and 39%, respectively. The mOS of pts with RUNX1::RUNX1T1 was similar to those with CBFB::MYH11 (12 months vs. 14 months, p=0.447). As compared to the corresponding wild-type pts, neither ELN 2024 intermediate risk variants nor KIT mutations adversely impacted mOS (FLT3-ITD/NRAS/KRAS, 11 months vs. NR, p=0.191; KIT, 14 months vs. 12 months, p=0.681). Conclusions Based on our analysis in a larger cohort of international pts, treatment with HMA/Ven seems to be a valuable treatment option for CBF AML pts ineligible for intensive chemotherapy. In contrast to previous studies in mostly Chinese cohorts, we did not observe a significant difference in response or survival between pts with AML with CBFB::MYH11 and those with RUNX1::RUNX1T1. With a mOS of around 12 months, outcomes of HMA/Ven treated CBF AML pts appear less favorable than those previously reported for pts with ELN 2024 favorable risk disease, and similar to those observed in ELN 2024 intermediate risk non-CBF AML pts, irrespective of FLT3-ITD, NRAS and/or KRAS mutations. Therefore, we propose that all pts with CBF AML treated with HMA/Ven should be provisionally classified as having ELN 2024 intermediate risk disease.

Isavuconazole is effective against invasive aspergillosis (IA) and mucormycosis (IM) and may improve clinical outcomes compared to alternative antifungal treatments. However, real-world evidence regarding its clinical use and the health economic burden of inpatient treatment for IA and IM of patients with haematological malignancies remains limited. A retrospective, matched, multicentre cohort study was conducted in six German tertiary care centres. The study included adults with haematological or oncological diseases who were diagnosed with proven, probable, or possible IA or IM. We compared clinical and health economic outcomes under first-line treatment initiated with isavuconazole (case group) vs. liposomal amphotericin B (L-AmB) and/or voriconazole (control group) between 2016 and 2021. A micro-costing approach was used to assess direct treatment costs. We included 198 patients (99 per group), most with a probable or possible classification. Median length of hospital stay was 44 days (interquartile range [IQR] 27–74) in the isavuconazole group and 39 days (IQR 26–56) in the control group (p = 0.285). All-cause mortality rates were 29

Febrile Neutropenia is an emergency in the treatment of cancer patients. It requires prompt and evidence-based clinical and antimicrobial management. The implementation of standard operating procedures (SOP) across hospitals and outpatient cancer departments can improve the outcome of FN patients by reducing FN-related morbidity and mortality and by the continuation of cancer treatment. This guideline describes an evidence-based approach to risk stratification, epidemiology, diagnosis, and treatment. It is provided by the Infectious Diseases Working Party (AGIHO) of the German Society of Hematology and Medical Oncology (DGHO) and is an update of the 2017 version. Emerging aspects in epidemiology, diagnostic procedures, risk stratification, first-line antimicrobial treatment, empiric antifungal treatment and the duration of antimicrobial treatment are discussed and rated on evidence-based strength of recommendation and quality of evidence as described by the European Society of Clinical Microbiology and Infectious Diseases (ESCMID). By this, the aim of this guideline is to provide evidence-based recommendations on the management of febrile neutropenia in cancer patients for the practicing clinician.

Cellular senescence is a dynamic cancer cell condition with both tumor-suppressive and relapse-promoting features. While terminating cell proliferation, it may sustain disease persistence through its senescence-associated secretory phenotype (SASP), immune alteration, and stem-like reprogramming. Here, we identify therapy-induced senescence (TIS) as a unifying, plastic state across newly diagnosed (nd), genetically diverse non-M3 acute myeloid leukemias (AML) that reprograms blasts towards an acute promyelocytic leukemia (APL)-like phenotype with distinct plasticity-related therapeutic vulnerabilities. We established a short-term ex vivo chemotherapy assay to quantify the patient-individual TIS capacity of primary AML blasts by fluorescent senescence-associated β-galactosidase activity, further characterized by p16INK4a expression, Ki67 loss, and H3K9me3 heterochromatin remodeling. Like daunorubicin or cytarabin, hypomethylating agents, hydroxyurea, and anti-CD33 antibody-drug conjugates also triggered TIS to varying extents, showing that diverse agents converge on senescence and prime AML cells for potential senolytic elimination. Probing baseline transcriptomes of nd AML samples by an AML TIS-high signature of the top-100 differentially expressed genes failed to stratify outcomes across TCGA (n=172), OHSU Beat AML (n=405), and MLL Munich Leukemia Laboratory (n=433) cohorts. Unexpectedly, a 13-gene subclassifier, determined by unsupervised cluster analysis of the top-100 genes and unrelated to the M3-typical t(15;17) PML::RARA translocation, identified 65 of 67 genetically defined APL cases across these nearly 1,000 all-subtype AML transcriptomes, suggesting that TIS rendered non-M3 AML APL-like. Multi-omic analyses including bulk and single-cell RNA sequencing plus PRC2/SUZ12 and H3K27me3 ChIP-seq unveiled TIS-related loss of polycomb repression, epigenetic remodeling, and transcriptional reprogramming, yet independent of PML::RARA, as the underlying molecular mechanism. Functionally, TIS-associated APL-like plasticity conferred sensitivity to differentiation therapies such as all-trans retinoic acid (ATRA) and histone deacetylase inhibitors, and to Bcl2-targeting agents with senolytic activity. Patient-derived xenograft models validated these vulnerabilities as sequential TIS followed by senescence targeting achieved durable control exclusively in models able to mount a senescence response. Given the biologically unifying and ATRA-sensitizing role of TIS-associated M3-like plasticity, we assessed its clinical relevance in three independent cohorts totaling 92 newly diagnosed patients. TIS capacity at diagnosis robustly stratified outcomes, predicted superior disease-free and overall survival, and was associated with favorable ELN classification, normal cytogenetics, and NPM1 or DNMT3A mutations, while being lowest in adverse-risk SRSF2-mutant AML. Our findings uncover TIS-related M3-like plasticity as a novel state exploitable by a two-punch strategy of senesence induction first followed by targeted interventions through pro-differentiation or senolytic approaches, establishing a new therapeutic paradigm with immediate clinical potential and broader relevance to other malignancies where therapy-induced plasticity dictates outcome.

Mutations of the FMS-like tyrosine kinase (FLT3) occur in acute myeloid leukemia (AML) and are associated with very poor prognosis. Available FLT3 inhibitors are potent but either show a lack of selectivity regarding other tyrosine kinases or only transient efficacy due to emerging resistance under therapy. Water-soluble derivatives of the tyrosine kinase inhibitor Marbotinib are highly selective dual-type I/II inhibitors of FLT3. The bisarylmethanone-based compound 29 and its carbamate derivative 42 show excellent results in various biological tests. They inhibit FLT3-ITD (internal tandem duplication) as well as therapy-associated FLT3-TKD point mutations. Additionally, good water solubility and consequently biological availability was achieved by attaching amine functions to appropriate scaffold positions, suggested by modeling of inhibitor binding at inactive and active FLT3 states. Subsequent formation of different salts led to very promising results in in vivo studies and improvements compared to midostaurin (1b), Quizartinib (7), Marbotinib 10 and its carbamate 11c.

The addition of gemtuzumab ozogamicin (GO) to intensive chemotherapy (IC) has become a mainstay in treating patients with core binding factor acute myeloid leukemia (CBF-AML). However, evidence for the efficacy of GO in this particular subgroup is primarily based on meta-analytic data from different trials conducted more than a decade ago. In this registry-based study, we evaluated the impact of adding GO to IC in 265 CBF-AML patients from the SAL, AMLCG, and CELL cooperative study groups. Patients receiving GO had a 2-year overall survival of 90% compared with 80% in those without GO (hazard ratio [HR] 0.45, 95% confidence interval [CI] 0.21–0.95, P = 0.036) and a 2-year event-free survival of 51% versus 36% (HR 0.69, 95% CI 0.48–0.99, P = 0.046). While complete remission rates in GO vs. non-GO patients were comparable (89% vs. 90%, P = 0.81), more GO patients achieved measurable residual disease-negative remission (77% vs. 49%, P < 0.001), resulting in numerically reduced cumulative incidence of relapse (HR 0.67, 95% CI 0.43–1.02, P = 0.06). Despite delayed platelet recovery, high-grade toxicities were not increased in GO-treated patients. These findings support the integration of GO into treatment protocols for IC-eligible patients with CBF-AML.