基于cfDNA甲基化和机器学习的男性胃癌早筛预测模型的建立

Construction of prediction model for early screening in male patients with gastric cancer based on cell-free DNA methylation and machine learning

目的利用新型的细胞游离DNA(cfDNA)甲基化检测技术,建立中国男性人群胃癌患者cfDNA甲基化模型。方法使用cfDNA甲基化免疫沉淀和高通量测序技术(cfMeDIP-seq)开展胃癌患者的全基因组甲基化的检测,利用生物信息学的方法定位胃组织来源的cfDNA,提取区分胃癌患者的特异甲基化标签,通过随机森林算法建立诊断模型,开展胃癌早期筛查的临床验证研究。结果基于胃癌样本和正常对照选取了前63个最为显著的差异甲基化区段,构建了cfDNA甲基化模型,并将此模型应用于胃癌早期筛查,灵敏度达到85%以上,特异性达到95%以上。验证集的灵敏度和特异性分别为98.7%和99.0%,曲线下方面积大小(AUC)为0.999。结论该研究构建的cfDNA甲基化模型具有良好的胃癌预测性能。

Objective To construct a cell-free DNA(cfDNA)methylation model for early screening in male patients with gastric cancer by using novel cfDNA methylation detection technology.Methods Methylation information of the whole genome of gastric cancer patients were detected by cell-free methylated DNA immunoprecipitation and highthroughput sequencing(cfMeDIP-seq)technology and locate gastrogenic cfDNA.Then bioinformation methods were used to extract specific methylation labels which could distinguish GC patients and establish diagnosis model by random forest algorithm.Related validation clinical researches were also conducted.Results 63 most significant DMR were selected to construct the cfDNA methylation model based on GC samples and normal control samples,the goal sensitivity was above 85 percent while the goal specificity was above 95%.The sensitivity and specificity of the validation set were 98.7%and 99.0%while the area under curve(AUC)was 0.999.Conclusion The cfDNA methylation model constructed in this study has good performance in predicting GC.