通过全基因组测序和单读纠错进行超灵敏分子残留疾病检测。
Ultra-sensitive molecular residual disease detection through whole genome sequencing with single-read error correction.
发表日期:2024 Aug 20
作者:
Xinxing Li, Tao Liu, Antonella Bacchiocchi, Mengxing Li, Wen Cheng, Tobias Wittkop, Fernando L Mendez, Yingyu Wang, Paul Tang, Qianqian Yao, Marcus W Bosenberg, Mario Sznol, Qin Yan, Malek Faham, Li Weng, Ruth Halaban, Hai Jin, Zhiqian Hu
来源:
EMBO Molecular Medicine
摘要:
虽然游离 DNA (cfDNA) 的全基因组测序 (WGS) 对于分子残留疾病 (MRD) 的检测具有巨大前景,但其性能受到 WGS 错误率的限制。这里我们介绍AccuScan,一种高效的cfDNA WGS技术,可实现单读长级别的全基因组纠错,实现4.2×10-7的错误率,比以读长为中心的去噪低约两个数量级方法。 AccuScan 在 MRD 中的应用证明了分析灵敏度低至 10-6 循环变异等位基因频率,样本水平特异性为 99%。 AccuScan 在预测结直肠癌复发方面表现出 90% 的标志性敏感性(术后 6 周内)和 100% 的特异性。使用手术后一周内收集的样本,它还显示出对食道癌的 67% 敏感性和 100% 特异性。当 AccuScan 用于监测黑色素瘤患者的免疫治疗时,循环肿瘤 DNA (ctDNA) 水平和动态特征与临床结果一致。总体而言,AccuScan 为 MRD 检测提供了高度准确的 WGS 解决方案,支持百万分之一范围内的 ctDNA 检测,无需大量样本输入或个性化试剂。© 2024。作者。
While whole genome sequencing (WGS) of cell-free DNA (cfDNA) holds enormous promise for detection of molecular residual disease (MRD), its performance is limited by WGS error rate. Here we introduce AccuScan, an efficient cfDNA WGS technology that enables genome-wide error correction at single read-level, achieving an error rate of 4.2 × 10-7, which is about two orders of magnitude lower than a read-centric de-noising method. The application of AccuScan to MRD demonstrated analytical sensitivity down to 10-6 circulating variant allele frequency at 99% sample-level specificity. AccuScan showed 90% landmark sensitivity (within 6 weeks after surgery) and 100% specificity for predicting relapse in colorectal cancer. It also showed 67% sensitivity and 100% specificity in esophageal cancer using samples collected within one week after surgery. When AccuScan was applied to monitor immunotherapy in melanoma patients, the circulating tumor DNA (ctDNA) levels and dynamic profiles were consistent with clinical outcomes. Overall, AccuScan provides a highly accurate WGS solution for MRD detection, empowering ctDNA detection at parts per million range without requiring high sample input or personalized reagents.© 2024. The Author(s).