AI and Cancer Research ── ALL ── 2026-11-05
Acute lymphoblastic leukemia (ALL): treatment, trials and AI
- Abstract
High-dose methotrexate (HD-MTX) is essential for pediatric acute lymphoblastic leukemia (ALL) but frequently causes hepatotoxicity. To address the critical lack of early prediction tools, this study investigated untargeted metabolomic profiles to identify predictive biomarkers and establish robust machine learning (ML) models for early clinical risk assessment and therapeutic optimization. This study included 106 pediatric patients with ALL undergoing HD-MTX therapy. Of these, 56 developed moderate-to-severe liver injury. Untargeted plasma metabolomics was performed before and after treatment. Metabolic alterations were identified using multivariate analyses, with pathway annotation performed via KEGG enrichment. Key discriminatory metabolites were further screened via LASSO regression to identify potential biomarkers. Three ML algorithms (Random Forest, Support Vector Machine, and Bayesian logistic regression) were developed for risk prediction, integrating SHAP analysis for model interpretability and feature contribution evaluation. Patients with liver injury showed distinct metabolic signatures compared with controls. Prior to HD-MTX treatment, 13 metabolites-such as prominently elevated glycocholic acid-were identified as potential predictors of liver injury, with overarching pathway alterations primarily involving arginine biosynthesis and glutathione metabolism, among others. Post-treatment, 11 distinct metabolites-including arachidic acid-further distinguished the injury group, reflecting complex regulatory mechanisms and multi-pathway shifts. Crucially, all three ML models developed using pre- or post-treatment metabolic biomarkers exhibited high predictive and discriminative accuracy for hepatotoxicity (AUC > 0.900). These findings characterize the metabolic alterations underlying HD-MTX-induced liver injury, provide a metabolomics-based framework for early risk assessment, and facilitate the development of precision therapeutic strategies in pediatric ALL.
Journal IF-equivalent: 1.7 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Guo K, Wang L, Feng X, Yang J, Zhang X, Zhao Y, et al. Metabolomic Profiling and Machine Learning‐Based Prediction of High‐Dose Methotrexate‐Induced Liver Injury in Pediatric Acute Lymphoblastic Leukemia. Pediatr Blood Cancer. 2026 Jul 23;73(9):e70424. doi:10.1002/1545-5017.70424. PMID: 42490747.Checked: Abstract only - Abstract
Childhood T-lineage acute lymphoblastic leukemia (T-ALL) is an aggressive hematologic malignancy with poor prognosis. Differently from B-cell precursor ALL, T-ALL lacks effective risk stratification strategies. A recent study has integrated whole genome and whole transcriptome data to define 17 distinct molecular subtypes with prognostic significance. However, clinical translation of this knowledge remains challenging due to the complexity of interpreting high-dimensional multi-omics-based data. Here, we present clinTALL, a deep learning based multi-task pipeline for pediatric T-ALL subtype classification and treatment outcome estimation. The model integrates multimodal input data and uses a neural network architecture to generate a shared latent embedding for jointly learned multi-task prediction. The competing risk-based model was used to predict event-specific outcomes. The model was trained on a publicly available multimodal dataset comprising clinical, genomic and transcriptomic features of 1309 pediatric T-ALL samples. We observed that the transcriptomic-only model achieved superior single-modality results, with 92.2% accuracy for subtype prediction and a 65.9% concordance index (C-index) for event-free survival (EFS) in a cross-validation setup. Integrating all data modalities maintained high subtype classification accuracy (91.7%) and improved the overall concordance index for EFS estimation to 67.5%. The competing risk-based model enables accurate predictions of induction failure (C-index = 96.0%) and second malignant neoplasm (C-index = 62.1%). We validated molecular subtype predictions on an internal dataset of 120 pediatric T-ALL samples and obtained an accuracy of 81.8%. To facilitate the broad application of multi-omics based subtype prediction and treatment outcome inference, we provide clinTALL as a Docker based application, allowing for user friendly access to the tool. Together, our machine learning-based framework allows for automated, accurate subtype classification and treatment outcome inference using multimodal input data, advancing precision risk stratification for pediatric T-ALL. The full source code of clinTALL is available on GitHub ( https://github.com/UKWgenommedizin/clinTALL ).
Journal IF-equivalent: 11.3 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Stoiber L, Antić Ž, Rebellato S, Fazio G, Rademacher A, Lenk L, et al. clinTALL: machine learning-driven multimodal subtype classification and treatment outcome prediction in pediatric T-ALL. Genome Med. 2026 Jul 22;18(1):108. doi:10.1186/s13073-026-01733-8. PMID: 42487139; PMCID: PMC13393873.Checked: Full text checked - Abstract
Molecular subtypes of B-cell acute lymphoblastic leukemia (B-ALL) are essential in modern clinical treatment. However, the fast emerging subtypes and the requirement of complex multi-omic diagnostics are continuously challenging the clinical subtypes classification in sensitivity, cost, and turnaround time. We develop B-cell Acute Lymphoblastic Leukemia Subtype Identification based on gene eXpression (BALL6), a robust deep learning framework for cross-platform B-ALL subtyping. BALL6 utilizes a recurrent neural network trained on rank-transformed expression values of feature genes, capturing subtype signals while inherently minimizing technical noise. We implement in BALL6 the rank-based augmentation framework which further enhances its performance on data-limited or imbalanced datasets. BALL6 includes two integrated models: an AL model distinguishing B-ALL, T-ALL, and AML, and a B-ALL model identifying the most updated 20 established molecular subtypes. BALL6 demonstrates robust accuracy across multiple independent datasets, achieving 99.38% (AL model) and 93.84% (B-ALL model) accuracy on previously unseen data. Notably, BALL6 is robust to missing values, which enables its cross-platform application. This is demonstrated through reliable subtype prediction by BALL6 with sparse gene expression profiles from scRNA-seq data. BALL6 is open-sourced with an accessible web tool ( https://cccg.ronglian.com/#/analysis ), facilitating its broad applications in leukemia research and clinical diagnostics.
Journal IF-equivalent: 9.8 (OpenAlex 2-year mean citedness, value as of 2026-11-04, retrieved 2026-11-05; not the official Clarivate IF)Reference: Cui B, Sun H, Zhao S, Fan R, Rao J, Wu W, et al. Deep learning based individualized cross-platform molecular subtype classification of B-lineage acute lymphoblastic leukemia. NPJ Precis Oncol. 2026 Jun 10;10(1):355. doi:10.1038/s41698-026-01556-1. PMID: 42265190; PMCID: PMC13594159.Checked: Full text checked - Abstract
Acute lymphoblastic leukemia (ALL) is a highly heterogeneous hematologic malignancy with poor prognosis in refractory and relapsed cases. Aurora kinase B (AURKB), a member of the Aurora kinase family, is markedly overexpressed in ALL patients with cytogenetic abnormalities and is associated with unfavorable clinical outcomes. Targeting AURKB represents a promising therapeutic approach to address this unmet clinical need. This study aimed to develop and validate a novel, selective degradation strategy against AURKB in ALL, utilizing artificial intelligence-assisted drug design to create a peptide-based degrader. The goal was to demonstrate the efficacy of this degrader in reducing AURKB expression and suppressing leukemic cell growth. We performed database analyses and confirmed AURKB overexpression in patient-derived ALL samples. Using AI-assisted design, we developed ATPD (AURKB Targeting Peptide Degrader), a selective peptide-based degrader of AURKB. The activity of ATPD was evaluated through in vitro cell proliferation assays, in vivo leukemia models, and ex vivo cytotoxicity tests using primary ALL cells. ATPD effectively induced the degradation of AURKB and significantly inhibited the proliferation of ALL cells both in vitro and in vivo. Furthermore, ATPD exhibited potent cytotoxic activity against primary leukemic cells derived from ALL patients, B-All PDX, and mini-PDX models. Our findings demonstrate that ATPD is the first AI-designed, selective AURKB degrader with potent anti-leukemic activity. This study highlights ATPD's potential as a novel, precise therapeutic strategy for the treatment of ALL, addressing a critical gap in managing refractory and relapsed disease.
Journal IF-equivalent: 9.9 (OpenAlex 2-year mean citedness, value as of 2026-11-03, retrieved 2026-11-04; not the official Clarivate IF)Reference: Xie Y, Feng H, Huang Y, Yang R, Lei H, He W, et al. Development of a selectively AURKB targeting peptide degradation drug with artificial intelligence-assisted design for the treatment of acute lymphoblastic leukemia. J Adv Res. 2026 Jul;85:1175-88. doi:10.1016/j.jare.2025.10.061. PMID: 41161489; PMCID: PMC13316543.Checked: Full text checked