Comprehensive Analysis of Cell-Free DNA Fragmentation Across Cancer Stages

Circulating cell-free DNA (cfDNA) in the bloodstream displays cancer-derived fragmentation patterns, offering a non-invasive diagnostic avenue for cancer patients. However, the association between cfDNA fragmentation patterns and cancer progression remains largely unexplored. In this study, we analyzed this relationship using 214 whole-genome cfDNA samples across seven solid cancer types and revealed that the relation between cfDNA fragmentation patterns and cancer stages vary across cancer types. Among them, cfDNA fragmentation patterns in colorectal cancer (CRC) showed a strong correlation with cancer stages. This finding is further validated with an independent targeted cfDNA dataset from 29 CRC samples. Inspired by these findings, we designed “frag2stage”, a machine learning model that exploits cfDNA fragmentation data to differentiate cancer stages of CRC. Evaluated on two independent cfDNA datasets, our model can distinguish cancer stages of CRC with the area under the curve (AUC) values ranging from 0.68 to 0.99. The results suggest that cfDNA fragmentation patterns might carry yet undiscovered genetic and epigenetic signals, highlighting their promising potential for broader diagnostic applications in oncology.