Machine Learning-Based Identification of Survival-Associated CpG Biomarkers in Pancreatic Ductal Adenocarcinoma.

Yu Zhang; Yining Zhao; Bo Aber Zhang

doi:10.1016/j.premed.2026.100046

Home/Archive/PubMed

PubMed · 42677177

Machine Learning-Based Identification of Survival-Associated CpG Biomarkers in Pancreatic Ductal Adenocarcinoma.

Yu Zhang·Yining Zhao·Bo Aber Zhang

View original source ↗Export metadata ↓

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is an exceptionally aggressive cancer with a 5-year survival rate of less than 10%, driven by late-stage diagnosis, limited treatment options, and a lack of reliable biomarkers for early detection and prognosis. In this study, we integrated DNA methylation data from TCGA and ICGC cohorts, categorizing samples based on survival time, and identified 688 differentially methylated CpG sites, along with 224 CpG biomarkers significantly associated with patient survival through statistical and machine learning-based analyses. We developed a random forest model to predict patient survival, achieving 85.2% accuracy for short-survival patients and 70.0% for long-survival patients in the validation set. External dataset validation further confirmed the model's robustness and accuracy. De novo motif analysis of genomic regions surrounding the 224 CpG biomarkers identified TWIST1 and FOXA2 as key transcriptional regulators enriched in survival-associated CpG sites, linking their activity to patient survival outcomes. Collectively, our findings highlight valuable epigenetic biomarkers and provide a predictive model to assess PDAC risk levels post-surgery, offering the potential for improved patient stratification and personalized therapeutic strategies.

Explore related subjects

DNA methylation biomarker machine learning pancreatic cancer survival Journal Article

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Yu Zhang, Yining Zhao, Bo Aber Zhang. 2026-07-09. Machine Learning-Based Identification of Survival-Associated CpG Biomarkers in Pancreatic Ductal Adenocarcinoma.. https://doi.org/10.1016/j.premed.2026.100046

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

KEEP EXPLORING

Related citations

PubMed2026

Epigenetic age acceleration is not strongly associated with cardiorespiratory fitness in heart failure: a pilot study.

BACKGROUND: In heart failure (HF), standard measures such as left ventricular ejection fraction and cardiopulmonary exercise testing incompletely capture interindividual differences in disease status or prognosis. DNA methylation (DNAm) epigenetic clocks, which estimate biological age and epigenetic age acceleration (EAA), may provide complementary insight into cardiorespiratory fitness and systemic aging in HF. RESEARCH DESIGN AND METHODS: We analyzed peripheral blood DNAm from fourteen patients enrolled in REDHART2, a clinical trial of interleukin-1 blockade following hospitalization for acute systolic HF. Genome-wide DNAm was assayed using Illumina EPIC arrays and several clocks were applied to these data. Associations between biological age or EAA and cardiorespiratory fitness measures, inflammatory markers, and clinical parameters were evaluated. RESULTS: All epigenetic clocks demonstrated moderate to strong correlations with chronological age. Biological age was consistently associated with measures of cardiorespiratory fitness, particularly oxygen consumption normalized to fat free mass (VO2_FFM). However, chronological age showed similar associations, and biological age did not significantly improve prediction of VO2 parameters beyond chronological age alone. EAA was not significantly associated with cardiorespiratory fitness for any clock. CONCLUSIONS: In this pilot study, neither biological age nor EAA provided significant predictive value beyond chronological age for cardiorespiratory fitness in patients with HF. CLINICAL TRIAL REGISTRATION NUMBER: NCT03797001.

DNA methylation↗

PubMed2026

Unraveling epigenetic and genetic variations in response to cold stress in two lotus ecotypes.

Genetic variations accumulate over long evolutionary timescales, whereas epigenetic modifications can arise rapidly and be inherited across generations. However, the interplay between genetic and epigenetic variations in shaping ecotype-specific phenotypic plasticity remains elusive. Focusing on two lotus ecotypes that evolved under distinct winter temperature regions and display divergent annual growth cycles under cold stress, we generated DNA methylation landscapes across three sequence contexts (CG, CHG, and CHH, where H = A, T, or C) and identified single-cytosine methylation polymorphisms (SMPs) and single-nucleotide polymorphisms (SNPs). Interestingly, only CG methylation patterns mirror population-level genetic variations. Using epigenetic genome-wide association analysis, we identified differentially methylated CG sites that are either cis- or trans-regulated by SNP. Notably, we constructed a multifactorial regulatory network centered on the NnMKK4-NnCYCD5 module, linking cold response with cell cycle regulation. Temperature stress experiments conducted on lotus ecotypes and transgenic Arabidopsis (OE-NnMKK4 and OE-NnCYCD5) confirmed that NnMKK4 acts as a cold receptor and that NnCYCD5 promotes cell cycle progression and growth under cold conditions. Collectively, our findings provide novel insights into the co-evolutionary dynamics of epigenetic and genetic variations that are associated with different growth cycles of lotus ecotypes in response to cold stress.

DNA methylation↗

PubMed2026

Mammalian DNA methyltransferases in DNA methylation and imprinted gene expression in extraembryonic ectoderm of post-implantation embryos.

DNA methylation in mammals is mainly catalyzed by three DNA methyltransferases (DNMTs). Conventionally, DNMT1 is considered the primary DNMT protein for maintenance DNA methylation, whereas DNMT3A and DNMT3B function in de novo DNA methylation. In two previous studies, we demonstrated that DNMT3A and DNMT3B maintain genome-wide DNA methylation in embryonic stem (ES) cells and in the epiblast of post-implantation embryos. Interestingly, DNMT3A and DNMT3B also sustain genome-wide DNA methylation in the extraembryonic ectoderm (EXE) of post-implantation embryos, including repeats, genic and intergenic regions. Although DNMT1 plays a major role in maintaining DNA methylation at the imprinting control regions (ICRs) in the imprinted regions, DNMT3A and DNMT3B are required for preserving DNA methylation at the ICRs of a subset of imprinted regions in EXE, similar to the observations in ES cells and epiblast. Surprisingly, de novo DNA methylation mediated by DNMT3A and DNMT3B leads to increased DNA methylation at a large subset of imprinted regions. These results are consistent with what we previously elucidated in the epiblast of post-implantation embryos. Importantly, loss of DNA methylation at the ICR of an imprinted region, resulting from the absence of DNMT1 or two DNMT3 proteins, causes allelic expression switch of the corresponding imprinted genes in that imprinted region. This study provides further evidence that DNMT3A and DNMT3B exert both maintenance and de novo DNA methylation functions across the genome in post-implantation embryos. It also validates some previous findings for DNA methylation-dependent allelic expression switch of imprinted genes.

DNA methylation↗