早期乳腺癌的单细胞DNA甲基化揭示对驱动突变的表观基因组反应及进展与浸润的独特特征

The pre-cancerous stage of invasive breast ductal carcinoma (IDC), termed ductal carcinoma in situ (DCIS) is a noninvasive neoplasm of the breast duct and is treated by surgical resection and radiation as standard of care. Pathology grading and mutational profiling do not sufficiently predict DCIS progression or recurrence; this uncertainty leads to overtreatment of an estimated 60% of possibly indolent lesions, via invasive surgical interventions and aggressive neoadjuvant therapies. Recent multicancer screening through DNA methylation markers of cell-free DNA show success in identifying staging and type of multiple cancers but fall short on DCIS and early-stage IDC, and do not predict progression nor recurrence. Further, since the methyl- group is covalently linked to DNA it is stable in FFPE-stored clinical samples for decades, allowing retrospective studies on progression and recurrence in a disease which can take decades to progress. We leverage this promising modality to expand upon our knowledge of DNA methylation in normal, DCIS and synchronous (DCIS+IDC) breast samples using our high-throughput single cell methylation (scMET) method. We profiled 23,692 cells across 13 normal, 10 DCIS, and 18 synchronous samples, equating to >650X coverage of the methylome in total. We paired single cell RNA (scRNA) profiles for all samples, resulting in 154,695 scRNA profiles. As a first-of-its-kind data set, we define methylation-based markers for all expected cell types, including cancer-associated fibroblasts and tumor endothelial cells, enriched in the tumor microenvironment and absent in the normal tissue samples. Through paired copy number calling on scRNA and scMET we match transcriptomes and methylomes to 74 subclonal populations across our DCIS and synchronous samples. In addition to patient and subclone specific mutational events, dozens of these subclones share stereotyped early breast cancer driver events such as 1q amplification and 16q loss. Through these shared events we characterize methylome response to gene dosage changes. Through lineage tracing of inherited methylation changes, known as “epimutations”, we track subclonal populations back to their diploid cell of origin: luminal hormone-responsive epithelial cells. Finally, through our whole methylome capture per cell, we match cell types and subclones to telomere lengths. This reveals signatures of aging with cell type specificity and marked decrease in telomere length through cancer progression. In sum, this study measures the the breast methylome as it progresses from a normal to neoplastic to invasive phenotypes through an epigenetic modality that shows great promise as a biomarker. This work will inform clinically viable approaches to patient stratification to better determine DCIS risk of progression and recurrence.