Long-read sequencing identifies aberrant fragmentation patterns linked to elevated cell-free DNA levels in cancer

Altered circulating cell-free DNA (cfDNA) fragmentation patterns serve as cancer biomarkers, yet standard short-read sequencing fails to capture the full fragment-length spectrum. Although cancer patients often exhibit elevated cfDNA, the relationship between high cfDNA concentration and altered fragmentation remains poorly defined. To address this question, we leverage Oxford Nanopore (ONT) sequencing, which captures the full fragment-length spectrum and enables cell type inference via DNA methylation markers.

We perform ONT whole-genome sequencing on a pan-cancer cohort and a neuroendocrine cancer cohort, both with elevated cfDNA levels. In both cohorts, the highest cfDNA levels are characterized by either hypofragmentation or hyperfragmentation. Hypofragmented samples (characterized by 1-4 kb fragments) exhibit hallmarks of DNASE1L3-mediated fragmentation due to blood-derived DNA release during delayed blood processing, while samples with ultra-long fragments (> 7.5 kb) indicate release due to cell lysis during plasma processing. In contrast, the short (<145 bp) fragments of hyperfragmented cancer samples are not artifactual, and they are characterized by elevated levels of both cancer- and blood-derived DNA, suggesting an inflammatory or other systemic response as opposed to a cancer-specific fragmentation mechanism.

These findings differentiate biological from artifactual fragmentation, broaden our understanding of high cfDNA levels and hyperfragmentation in cancer, and establish long-read sequencing as a robust tool for biomarker discovery.