Zhang Xianglin, Zhang Wei, Zhang Jinyi, Lyu Xiuhong, Pan Haoran, Jia Tianwei, Wang Ting, Wang Xiaowo, Guo Haiyang
Department of Clinical Laboratory, the Second Hospital, Cheeloo College of Medicine, Shandong University, 247 Beiyuan Street, Tianqiao District, Jinan, 250033, Shandong, China.
Shandong Engineering & Technology Research Center for Tumor Marker Detection, Department of Clinical Laboratory, the Second Hospital, Cheeloo College of Medicine, Shandong University, 247 Beiyuan Street, Tianqiao District, Jinan, 250033, Shandong, China.
Brief Bioinform. 2025 May 1;26(3). doi: 10.1093/bib/bbaf252.
CpG island hypermethylation, a hallmark of cancer, exhibits substantial heterogeneity across tumors, presenting both opportunities and challenges for cancer diagnostics and therapeutics. While this heterogeneity offers potential for patient stratification to predict clinical outcomes and personalize treatments, it complicates the development of robust biomarkers for early detection. Understanding the mechanisms driving this heterogeneity is essential for advancing biomarker design. Here, simulation-based analyses demonstrate that tumor purity and the high prevalence of low epi-mutation samples significantly obscure the identification of negative, rather than positive, regulators of CpG island hypermethylation, limiting a comprehensive understanding of heterogeneity sources. By addressing these confounders, we identify impaired DNA methylation maintenance, as indicated by global hypomethylation levels, as the primary contributor to CpG island hypermethylation variability among known regulators. This finding is supported by integrative analyses of datasets from The Cancer Genome Atlas (TCGA) Pan-Cancer Atlas, Genomics of Drug Sensitivity in Cancer (GDSC1000) cancer cell lines, and epi-allele analyses of two independent whole-genome bisulfite sequencing cohorts, using a newly developed method, MeHist (https://github.com/vhang072/MeHist). Furthermore, we assess widely used hypermethylation biomarkers across ten cancer types and find that 65 out of 246 (26.4%) are significantly influenced by impaired methylation maintenance. Incorporating hypomethylation and hypermethylation markers improves the robustness of cancer detection, as validated across multiple plasma cell-free DNA datasets. In summary, our findings highlight the value of simulation-guided integrative analysis in mitigating confounding effects and identify impaired DNA methylation maintenance as a key regulator of CpG island hypermethylation heterogeneity.
CpG岛高甲基化是癌症的一个标志,在肿瘤之间表现出显著的异质性,这给癌症诊断和治疗带来了机遇和挑战。虽然这种异质性为患者分层以预测临床结果和个性化治疗提供了潜力,但它使用于早期检测的强大生物标志物的开发变得复杂。了解驱动这种异质性的机制对于推进生物标志物设计至关重要。在这里,基于模拟的分析表明,肿瘤纯度和低表观突变样本的高患病率显著掩盖了CpG岛高甲基化的负调控因子而非正调控因子的识别,限制了对异质性来源的全面理解。通过解决这些混杂因素,我们确定了如全局低甲基化水平所示的DNA甲基化维持受损是已知调控因子中CpG岛高甲基化变异性的主要贡献者。这一发现得到了来自癌症基因组图谱(TCGA)泛癌图谱、癌症药物敏感性基因组学(GDSC1000)癌细胞系的数据集的综合分析以及使用新开发的方法MeHist(https://github.com/vhang072/MeHist)对两个独立的全基因组亚硫酸氢盐测序队列的表观等位基因分析的支持。此外,我们评估了十种癌症类型中广泛使用的高甲基化生物标志物,发现246个中有65个(26.4%)受到甲基化维持受损的显著影响。纳入低甲基化和高甲基化标志物可提高癌症检测的稳健性,这在多个血浆游离DNA数据集中得到了验证。总之,我们的发现突出了模拟引导的综合分析在减轻混杂效应方面的价值,并确定DNA甲基化维持受损是CpG岛高甲基化异质性的关键调控因子。