A computational framework for sensitive tumor detection and accurate subtyping using shallow cell-free DNA methylome sequencing

Plasma circulating tumor DNA (ctDNA) enables non-invasive monitoring of metastatic cancer. However, the detection of low tumor content (TC) via tumor tissue-agnostic approaches remains challenging. We introduce METER, a computational strategy exploiting tumor-type specific DNA methylation patterns for sensitive ctDNA detection, accurate quantification, and subtyping from plasma low-pass (0.5-1x) whole-methylome sequencing. In longitudinal samples from metastatic breast cancer patients, METER demonstrated a stronger association with clinical outcomes than both state-of-the-art ctDNA methods and matched circulating tumor cell (CTC) counts, even at TC below 3%. METER (https://github.com/caos-lab-unifi/METER) integrates TC estimation and subtyping in a single framework, enabling sensitive and accurate analyses for precision oncology.