Integration of breaking-point transition probabilities and specific open chromatin region features for early cancer...

Early detection of cancer using cell-free DNA (cfDNA) remains challenging due to the limited release of circulating tumor DNA (ctDNA) in early-stage disease. This study aimed to enhance detection sensitivity by integrating cfDNA fragmentation features with epigenomic context related to chromatin accessibility.

We performed low-pass whole-genome sequencing (LP-WGS) on cfDNA samples from 506 participants, including lung cancer patients, benign cases, and healthy controls. Fragmentation patterns were characterized by calculating 16-bp breakpoint transition probabilities and extracting features within tissue-specific open chromatin regions (OCRs). These features were combined into an model termed SOTP_BP16_OCR, and its performance in multi-cancer classification on additional datasets was further evaluated.

cfDNA fragments exhibited strong sequence bias within 16 bp around breakpoints, extending beyond the conventional 5'-end motif window. The selected lung cancer-specific OCRs were enriched in biologically meaningful regions and showed clear differences between healthy and malignant samples. The SOTP_BP16_OCR model achieved an AUC of 0.96 (95% CI: 0.94-0.99) for distinguishing lung cancer patients from healthy individuals, including those with benign nodules, and maintained an AUC of 0.95 (95% CI: 0.92-0.98) for early-stage cases. Using features derived from two external datasets (n = 129 and n = 225), the SOTP_BP16_OCR model achieved AUCs of 0.94 and 0.99 for case-control classification. Tissue-of-origin prediction, which could only be performed in one dataset, reached an accuracy of 0.802.

By integrating sequence transition probabilities and open chromatin features, our model captures biologically relevant fragmentation signals under limited ctDNA release, providing a sensitive, affordable, and scalable approach for cfDNA-based cancer early screening.