Abstract 6599: Neomer and fragmentation profiles of cell-free DNA with low-pass whole genome sequencing to predict prostate cancer biopsy outcomes

Abstract

Abstract Prostate cancer is one of the most common types of cancer in men. Current prostate-specific antigen (PSA)-based diagnostic strategy for prostate cancer may frequently cause overdiagnosis. In this study, we prospectively recruited 380 individuals, including 167 with pathologically confirmed prostate cancer, 127 with negative prostate biopsy results, and 86 assessed healthy cohort with no need for biopsy. We collected a plasma sample from each participant, performed whole-genome sequencing, and profiled cell-free DNA (cfDNA) neomer and fragmentation characteristics. We then developed a machine learning model using these cfDNA profiles to predict prostate cancer biopsy outcomes. Our results showed that this assay could achieve high accuracy (Accuracy: 85.3%, AUROC: 0.919, 95% CI: 0.892 - 0.946) in repeated 5-fold cross validations. Patients with negative biopsy results had higher prediction scores than healthy individuals (P < 2.2X10-16), making them less distinguishable from prostate cancer patients (AUROC: 0.878, 95% CI: 0.837 - 0.918 vs. AUROC: 0.980, 95% CI: 0.963 - 0.997). We further combined this cfDNA-based assay with the PSA testing and constructed an ensemble model. The classification performance was further improved in the ensemble model, showing AUROC metrics of 0.951 (95% CI: 0.932 - 0.969) between cancer and non-cancer, 0.919 (95% CI: 0.889 - 0.949) between cancer and negative biopsy, and 0.998 (95% CI: 0.994 - 1.000) between cancer and healthy individuals. At a sensitivity of 90.4%, the ensemble assay accurately identified 85.4% of non-cancerous conditions, including 76.4% of negative biopsy patients and 98.8% of healthy individuals. We concluded that our cfDNA fragmentation and neomer-based assay could be incorporated into the PSA-based predicting system for prostate cancer to increase the accuracy and reduce the overdiagnosis. Citation Format: Fei Liu, Kaoqing Peng, Rui Chen, Mingzhao Li, Tao Zeng, Yan Zhao, Ming Wang, Hongxing Liu, Jianhao Wu, Min Wu, Haimeng Tang, Hua Bao, Xue Wu, Yang Shao, Di Gu, Shancheng Ren. Neomer and fragmentation profiles of cell-free DNA with low-pass whole genome sequencing to predict prostate cancer biopsy outcomes [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6599.