Abstract 4950: Applying fragmentomics profiles of plasma cell-free DNA for breast cancer detection

Abstract

Abstract The liquid biopsy approach focusing on DNA molecules supports the non-invasive diagnosis of breast cancer. Low-pass whole genome sequencing(lpWGS) for breast cancer detection has been validated for accuracy and cost-effectiveness, especially with copy number variant (CNV) analysis. However, detecting breast cancer with cell-free DNA (cfDNA) CNV was limited by tumor fraction. Utilizing cell-free DNA fragmentation characteristics for cancer detection has shown potential in plasma with high precision and applicability with low tumor fraction. In this study, 217 patients with breast cancer and 10 healthy individuals were enrolled. The cfDNA was extracted and performed using PredicineSCORE, a low-coverage whole genome sequencing assay to identify tumor-specific features. The cfDNA fragmentomics profiles including the coverage at specific regions, length and end-motif of fragment were implemented. We also investigated the copy number variation and the inferred tumor fraction which was adjusted by the Expectation-Maximum (EM) algorithm. Combining, an ensembled machine learning model was constructed for the classification of breast cancer and normal control. We observed a high correlation between tumor fraction and features. The absolute correlation coefficient scores between tumor fraction with the LYL1, MAF, FOXA1, and GRHL2 bonding sites' relative depth were over 0.82, 0.71,0.89,0.87. The insert size ratio of short reads and the CCCA end motif were over 0.77 and 0.72, respectively. We then diluted the tumor samples of tumor fraction at 0.1% to 50% by manually mixing reads from tumor samples into normal samples. By applying the XGBoost algorithm, the performance (F1 score > 0.95) showed a promising ability to distinguish tumors from normal samples. The limit of detection (LOD) was less than 1% which is more sensitive than previously proposed models. Those samples detected mutations with high through panel which can confirm the accuracy of the assay. Our results discovered that profiling the cell-free DNA fragment could facilitate the way of non-invasively detection of breast cancer with high precisions. Citation Format: Yaxin Liu, Shuang Gan, Hang Dong, Haoran Tang, Nan Wang, Xinyu Gui, Anjie Zhu, Ruyan Zhang, Pan Du, Shidong Jia. Applying fragmentomics profiles of plasma cell-free DNA for breast cancer detection [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4950.