PO.BCS01.11 · 生物信息与计算
利用纳米孔测序的cfDNA甲基化特征检测早期癌症
Detecting early cancer using cfDNA methylation signatures with nanopore sequencing
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
早期癌症筛查通过在疾病进展前实现及时治疗来提高生存率。然而，当前的全人群筛查工具仅限于少数几种癌症类型，突显了对可常规用于检测多种癌症的新型检测方法的需求。血浆中肿瘤来源的无细胞DNA（cfDNA）携带肿瘤的遗传和表观遗传改变，可用于开发新的筛查方法。在此，我们旨在识别癌前和早期肿瘤相关的cfDNA甲基化特征。我们分析了对乳腺癌或前列腺癌患者临床诊断前长达10年采集的血浆、以及乳腺癌患者II期诊断血浆所进行的纳米孔测序数据。初步数据揭示，诊断性乳腺癌样本中存在全基因组低甲基化和患者间变异性增加，二者在较年轻患者中更为明显。值得注意的是，全基因组低甲基化见于三阴性乳腺癌（TNBC）和ER+/HER2-乳腺癌，但不见于其他亚型。这些全基因组改变未在两种癌症类型的癌前病变中观察到，提示此类表观遗传改变可能在恶性转化过程中产生或变得可检测，而非在肿瘤早期起始阶段。为识别与癌前表观遗传改变相关的位点，在基因组不同功能元件上选取了差异甲基化区域（DMR）。使用这些DMR，在发现队列（癌前乳腺：n=33；癌前前列腺：n=35）中为每种癌症类型构建了集成随机森林模型。训练后的模型在验证集中对乳腺癌（AUC=0.76，癌前乳腺：n=12；II期乳腺：n=12）和前列腺癌（AUC=0.83，n=5）均显示出中等的区分能力，且两个模型在年龄较大的组中表现更好。此外，启动子甲基化特征也能够将患者分层为发生乳腺癌的高风险和低风险。将进行通路富集分析以研究所选特征的致癌相关性。本项目的成功将界定如何通过液体活检检测早期癌症，并推进早期检测的时间线。本项目有潜力开发出用于多癌种检测和监测的单次血液检测。
查看英文原文 English abstract
Early cancer screening improves survival by enabling timely treatment before the disease progresses. However, current population-wide screening tools are limited to a few cancer types, highlighting the need for novel detection methods that can be routinely used to detect multiple cancers. Plasma tumour-derived cell-free DNA (cfDNA) carries the tumour's genetic and epigenetic alterations and can be used to develop new screening methods. Here, we aim to identify precancerous and early tumour-associated cfDNA methylation signatures. We analyzed nanopore sequencing data performed on plasma collected up to 10 years before clinical diagnosis from patients with breast or prostate cancer, and stage II diagnostic plasma from breast cancer patients. Preliminary data revealed global hypomethylation and increased inter-patient variability in the diagnostics breast cancer samples, both of which were more pronounced in younger patients. Notably, global hypomethylation was found in Triple Negative Breast Cancer (TNBC) and ER+/HER2- Breast Cancer, but not in other subtypes. These global alterations were not observed in precancerous lesions in both cancer types, suggesting such epigenetic alterations may arise or become detectable during malignant transformation rather than early tumour initiation. To identify loci that are associated with precancerous epigenetic alterations, differentially methylated regions (DMRs) were selected across different functional elements in the genome. Using these DMRs, an ensemble random forest model was built in the discovery cohort (pre-breast: n = 33; pre-prostate: n = 35) for each cancer type. The trained model showed moderate discriminability for both breast (AUC = 0.76, pre-breast: n = 12; stage II breast: n = 12) and prostate (AUC = 0.83, n = 5) cancer in the validation set, and both models performed better in the older age group. In addition, promoter methylation signatures are also capable of stratifying patients into high-risk and low-risk of developing breast cancer. Pathway enrichment analysis will be performed to investigate the oncogenic relevance of selected features. Success in this project will define how early cancer can be detected through liquid biopsy and advance the timeline of early detection. This project has the potential to develop a single blood test for multi-cancer detection and monitoring.
利益披露 Disclosure
Y. Hou, None..
N. Cheng, None.