Whole-Genome Deep Learning Predicts Chemotherapy Response in Colorectal Cancer.

Hossein Sadeghi; Fatemeh Seif

doi:10.1007/s10528-026-11368-4

Home/Archive/PubMed

PubMed · 41964737

Whole-Genome Deep Learning Predicts Chemotherapy Response in Colorectal Cancer.

Hossein Sadeghi·Fatemeh Seif

View original source ↗Export metadata ↓

Abstract

Chemotherapy response in colorectal cancer (CRC) exhibits significant heterogeneity, with current clinical predictors failing to capture complex genomic determinants of resistance. We developed a hybrid deep learning framework integrating convolutional neural networks (CNNs) and bidirectional long short-term memory (BiLSTM) networks to analyze whole-genome somatic mutations, evolutionary conservation, chromatin accessibility, and 3D genome architecture in 2,546 TCGA patients. An attention mechanism identified predictive genomic regions. The model achieved an AUC of 0.92 (95% CI: 0.89-0.94) in cross-validation and 0.88 (95% CI: 0.85-0.91) in independent validation, outperforming clinical models (ΔAUC = +0.18, p < 0.001). Key predictors included non-coding variants in TP53, KRAS, and PIK3CA regulatory regions. Triple-positive patients (mutations in all 3 regions) had significantly worse progression-free survival (HR = 4.7, p < 0.001). Our framework enables accurate chemotherapy response prediction and reveals novel non-coding resistance mechanisms, advancing precision oncology in CRC.

Explore related subjects

Humans Colorectal Neoplasms Deep Learning Convolutional Neural Networks Mutation Tumor Suppressor Protein p53 Long Short Term Memory Genome, Human Class I Phosphatidylinositol 3-Kinases Antineoplastic Agents Chemotherapy response prediction Colorectal cancer genomics Deep learning Precision oncology Whole-genome sequencing Journal Article

Keep this discovery

BibTeX RIS

Hossein Sadeghi, Fatemeh Seif. 2026-04-11. Whole-Genome Deep Learning Predicts Chemotherapy Response in Colorectal Cancer.. https://doi.org/10.1007/s10528-026-11368-4

Cite the original work for its findings. Save a collection to share your selection of sources.

Discover connections

Journal Article resource Antineoplastic Agents subject Colorectal Neoplasms subject Deep Learning subject Humans subject Journal Article subject Mutation subject

Connections use source metadata and explicit phrase matches, not verified experimental comparisons.

KEEP EXPLORING

Related citations

PubMed2026

Beyond predictive performance: A systematic review and critical methodological appraisal of AI/ML and conventional modelling strategies in breast, colorectal, and pancreatic Cancer.

BACKGROUND: Predictive modelling for cancer risk, treatment-related complications, and survival is central to precision oncology. Conventional logistic regression (LR) and Cox proportional hazards (CoxPH) regression remain widely used but are limited when modelling nonlinear interactions, high-dimensional imaging features, and multimodal clinical-metabolic predictors. Artificial intelligence (AI) and machine learning (ML) methods offer expanded capability through automated feature extraction, ensemble learning, and flexible survival modelling, but the evidence on when AI/ML adds value over conventional models across cancer sites and predictive tasks remains fragmented. OBJECTIVE: To systematically evaluate the methodological performance, validation strategies, and translational limitations of AI/ML models compared with conventional statistical models in published predictive-modelling studies for breast, colorectal, or pancreatic cancer. METHODS: PubMed, Scopus, and Web of Science were searched for studies published between January 2019 and March 2025. Two reviewers independently conducted title-and-abstract screening, full-text eligibility assessment, and PROBAST risk-of-bias assessment. Sixty-five studies (n = 907,567 participants) were narratively synthesised by cancer site, predictive task, model family, comparator, validation strategy, predictor modality, and calibration or explainability reporting. RESULTS: The 65 studies comprised breast cancer (n = 35), colorectal cancer (n = 21), and pancreatic cancer (n = 9). AI/ML superiority over LR and CoxPH was task- and data-dependent. CNN- and U-Net-based models predominated in imaging and body-composition tasks, tree-based ensembles consistently outperformed LR for tabular perioperative complication prediction, and CoxPH remained competitive, and in the largest pancreatic risk study, superior to XGBoost (C-index 0.802 vs 0.723) in well-structured datasets. PROBAST analysis-domain risk was moderate in 54 of 65 studies (83%), driven by limited external validation, sparse calibration reporting (11/65), and few decision-curve analyses (7/65). CONCLUSION: AI/ML adds the most methodological value in imaging-derived feature extraction and nonlinear perioperative prediction, while conventional regression remains preferable in large, structured datasets with linear predictors. Clinical translation requires standardised body-composition definitions, external validation, calibration assessment, decision-curve analysis, and explainability, in line with TRIPOD+AI and CLAIM standards.

Humans↗

PubMed2026

Mechanisms linking the gut microbiota to colorectal cancer development and progression.

Colorectal cancer remains a leading cause of global cancer mortality, with a concerning rise in early-onset cases driven by complex interactions between environmental exposures, lifestyle factors, and host genetics. Mounting evidence indicates that gut microbiota dysbiosis critically modulates this oncogenic process, acting as an active participant rather than a passive bystander. This review systematically synthesizes the dichotomous roles of the intestinal microbiome in colorectal tumorigenesis through the conceptual framework of the driver-passenger model. We discuss how early initiating driver bacteria, such as Polyketide synthase-positive Escherichia coli and enterotoxigenic Bacteroides fragilis, compromise mucosal barriers, induce chronic mucosal inflammation, and inflict direct genomic instability. As the local tumor microenvironment undergoes profound metabolic remodeling, opportunistic passenger pathogens, notably Fusobacterium nucleatum, become enriched, further promoting cellular proliferation and facilitating tumor immune evasion. Conversely, protective commensals, exemplified by Clostridium butyricum and Streptococcus thermophilus, exert robust tumor-suppressive effects through multifaceted mechanisms. These beneficial microbes actively antagonize malignant progression by redirecting tumor metabolic fluxes toward oxidative stress, orchestrating deep epigenetic reprogramming, and degrading core oncoproteins to reverse chemoresistance. Transitioning from fundamental mechanisms to clinical application, we evaluate a comprehensive spectrum of microbiota-targeted interventions, encompassing non-invasive diagnostic biomarkers, fecal microbiota transplantation, engineered bacteria, phage therapy, and postbiotics. Finally, we critically address the formidable translational challenges associated with microbial heterogeneity, long-term safety, and regulatory standardization, aiming to provide a balanced perspective on integrating microbiome-based strategies into next-generation precision oncology for colorectal cancer.

Humans↗

PubMed2026

A machine learning-derived and functionally validated circadian rhythm signature predicts clinical outcomes and in silico drug sensitivity in colorectal cancer.

BACKGROUND: Colorectal cancer (CRC) displays considerable heterogeneity in clinical outcomes, highlighting the need for reliable prognostic biomarkers. While the aberrant expression of circadian rhythm-related genes has been implicated in cancer pathogenesis, its comprehensive role in CRC progression and predicted therapeutic vulnerabilities remains inadequately characterized. METHODS: Bulk and single-cell RNA-sequencing data were integrated from multiple CRC cohorts. A circadian rhythm signature (CRS) was developed through machine learning algorithms and validated for prognostic value. Comprehensive analyses of tumor microenvironment, genomic alterations, and drug sensitivity were performed. Furthermore, the biological function of the core gene, BHLHE40, was validated in CRC cell lines through CCK-8, EdU, and wound healing assays. RESULTS: Single-cell analysis demonstrated an elevated expression signature of circadian rhythm-related genes in dendritic cells. The optimized CRS, comprising 14 circadian rhythm-related genes, successfully categorized patients into high- and low-risk groups. Patients with a high CRS showed markedly poorer overall survival and computationally inferred immunosuppressive features, including reduced CD8+ T cell infiltration and increased M2 macrophage polarization. Genomic analysis revealed enhanced mutation burden in TP53 and alterations in RTK-RAS/WNT pathways. Notably, in vitro assays confirmed that BHLHE40 is significantly overexpressed in CRC cells. Knockdown of BHLHE40 markedly inhibited tumor cell proliferation and migration. Drug sensitivity profiling identified bexarotene and SMER-3 as potential therapeutic options for high-CRS patients. A nomogram integrating CRS with clinical parameters demonstrated superior predictive accuracy for 1-, 3-, and 5-year survival. CONCLUSIONS: The CRS represents a promising prognostic biomarker that reflects tumor immune status and genomic features, providing valuable insights for personalized treatment strategies in CRC.

Circadian rhythm↗