Preleukemic clonal evolution revealed using single-cell DNA sequencing and computational modelling

Preleukemic clonal evolution revealed using single-cell DNA sequencing and computational modelling
复制标题

使用单细胞 DNA 测序和计算模型揭示白血病前期克隆进化

DOI:
10.1101/2023.12.15.571872
复制
发表时间:
2023
期刊:
--
影响因子:
--
通讯作者:
Blundell J
Blundell J
中科院分区:
--
文献类型:
--
作者:
Blundell J

文献摘要

相似文献

癌症驱动突变的获得可以比诊断早几十年。然而,我们对这些癌症早期阶段发生的进化的理解仍然有限,因为大多数研究只关注“赢得”进化竞赛的恶性克隆的测序。在干细胞中不转化为恶性克隆的驱动突变的代表提供了这种癌前演变的窗口。在这里,我们从16例急性髓系白血病(AML)患者的骨髓中分离出白血病前造血干细胞(pHSC),并对数千个pHSC进行单细胞DNA测序,以重建每位患者白血病前演变的系统发育树。在大多数情况下,AML需要在同一细胞谱系中获得3或4个驱动突变。然而,重建的系统发育树是高度可变的16名患者。树的结构和大小的中间白血病前克隆编码的重要信息的作用,选择,突变和遗传漂变癌前演变过程中的操作。使用模拟和机器学习方法,我们表明,在pHSC树中观察到的高度可变的模式自然地从体细胞进化模型中出现,在该模型中,在AML的整个多击轨迹中存在强烈的正选择作用。我们推断,每个驱动突变的收购提供了一个额外的选择优势,每年约15%的pHSC克隆相对于他们的直系祖先。使用这些推断水平的积极选择,我们表明,个人注定要发展AML在生命的后期很可能窝藏多突变克隆,已经出现了不幸的生命早期。我们的研究结果表明,在成年早期识别携带2个驱动突变的单个HSC可能为识别AML进展高风险个体提供合理的策略。
The acquisition of cancer driver mutations can predate diagnosis by decades. However, our understanding of the evolution that occurs in these early stages of cancer remains limited because most studies focus only on sequencing the malignant clone that" wins" the evolutionary race. The representation of driver mutations in the stem cells which do not transform into the malignant clone provides a window into this precancerous evolution. Here, we isolate preleukemic haematopoietic stem cells (pHSCs) from the bone marrow of 16 patients with Acute Myeloid Leukaemia (AML) and perform single-cell DNA sequencing on thousands of pHSCs to reconstruct phylogenetic trees for the preleukemic evolution in each patient. In the majority of cases AML required the acquisition of 3 or 4 driver mutations in the same cell lineage. However the reconstructed phylogenetic trees are highly variable across the 16 patients. The structure of the trees and the sizes of intermediate preleukemic clones encode important information about the roles of selection, mutation and genetic drift operating during precancerous evolution. Using simulations and machine learning approaches, we show that the highly variable patterns observed in pHSC trees emerge naturally from a model of somatic evolution in which there is strong positive selection acting throughout the entire multi-hit trajectory towards AML. We infer that the acquisition of each driver mutation provides an additional selective advantage of~ 15% per year to pHSC clones relative to their immediate ancestors. Using these inferred levels of positive selection we show that individuals destined to develop AML late in life are likely to harbour multiple-mutant clones that have arisen anomalously early in life. Our findings suggest that the identification of a single HSC carrying 2 driver mutations in early adulthood may provide a rational strategy to identify individuals at high risk of AML progression.