Understanding the genetic evolutionary pathways of tumors as they progress from normal to malignant states is crucial for cancer biology, early tumor detection and determination of optimal therapeutic strategies. While traditional methods for assessing cancer progression rely on somatic mutation analysis, the relatively low mutation burden in many tumors as well as low tumor purity limits their resolution at the individual patient level. In this context, we propose that methylation data can significantly enrich tumor datasets, offering a novel approach to enhance the precision of phylogenetic reconstruction and temporal resolution in cancer analysis. Our study focuses on leveraging the extensive landscape of DNA methylation, with over 28 million methylated sites available for analysis. A noticeable fraction of these sites exhibits differential methylation between tumor and adjacent normal tissues, providing a high-density map that surpasses the resolution of somatic mutation data.

We validated the proposed concept utilizing Oxford Nanopore Technologies long-read whole genome sequencing of longitudinal tumor samples. Through nanopore long-read sequencing, we achieved haplotype estimation, which facilitated a more accurate evaluation of tumor purity and ploidy. We obtained comprehensive profiles of both tumor methylation changes and mutations across multiple timepoints. We have developed a novel version of our previously developed tool, Phylogic, which is now capable of conducting high-resolution temporal evolution and phylogenetic analyses from methylation data.

Importantly, somatic methylation sites possess unique dynamics, appearing and disappearing throughout tumor evolution, necessitating specialized modeling algorithms. Our newly developed tools allow for the first-time clock-like temporal estimation of methylation at the level of individual samples and cohorts, while simultaneously reconstructing phylogenies and subclonal populations by integrating methylation and mutation data from longitudinal samples.

The new approach not only enhances our understanding of cancer progression, but also provides a powerful framework for analyzing mixed cell-type samples, thereby facilitating more accurate assessments of cancer phylogeny and timing. Through this work, we aim to open new avenues for early detection and targeted intervention in cancer development.