Abstract 3685: Improved detection of minimal residual disease in colorectal cancer patients using adaptive noise cancellation algorithm

Abstract Objective: Existing methods for detecting circulating-tumor DNA (ctDNA) through fixed panels exhibit constrained sensitivity, while customized panels are time-consuming and rely on prerequisite tumor mutation profiles. Here, we aimed to introduce an adaptive noise cancellation approach for efficient Minimal Residuals Disease (MRD) detection in colorectal cancer patients employing a fixed panel under both tumor-informed and tumor-naïve situations. Methods: 108 plasma samples were collected from 52 colorectal cancer patients (11 relapse, 41 non-relapse) at various post-surgical timepoints. Comprehensive mutational profiling of plasma samples was conducted using a fixed MRD panel (SHIELDINGTM ULTRA) that encompasses hotspots in 2365 cancer-related genes. Background errors were removed by an adaptive noise cancellation algorithm relying on DNA fragment profiling, normal pool, and white blood cell background and intra-run plasma sampling. Variants were called both with (tumor-informed) and without (tumor-naïve) the present of tumor tissue. Results: Our longitudinal analysis of 108 plasma samples yielded 90. 9% sensitivity, 97. 6% specificity and 97. 6% negative predictive value in the tumor-informed scenario. The ctDNA positive patients exhibited significantly higher risk than those with negative ctDNA status (Hazard Ratio = 87. 2, p = 3. 5 × 10^ (-5) ). For landmark analysis, 33 plasma samples (≤7 days post-operation) were assessed, yielding a remarkable performance of 66. 7% sensitivity and 100% specificity. Our approach showed robustness under tumor-naïve scenario, illustrating 90. 9% sensitivity, 92. 7% specificity and 97. 4% NPV for the longitudinal analysis. This corresponded to an increased risk of 49 times in the ctDNA positive patients (HR = 49. 0, p = 2. 2 × 10^ (-4) ). Notably, the results of tumor-naïve samples closely aligned with those of tumor-informed samples, showing a concordance of 92. 6% and substantial agreement (Cohen’s Kappa = 0. 691). Conclusion: Our method stands as an ultra-sensitive tool for identifying colorectal patients at high risk of recurrence. By employing an adaptive noise cancellation algorithm, our fixed panel approach demonstrates superiority over existing fixed panels in accuracy and proves to be a cost-effective alternative to personalized panels. Significantly, it extends potential clinical utilization towards tumor-naïve patients, which is not feasible for personalized panels. Citation Format: Hua Bao, Wanxiangfu Tang, Ningyou Li, Min Wu, Jinfeng Zhang, Baihan Zhu, Xue Wu, Yang Shao. Improved detection of minimal residual disease in colorectal cancer patients using adaptive noise cancellation algorithm 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 (6Suppl): Abstract nr 3685.