Abstract 1909: Extracellular vesicle-derived biomarkers surpass cfDNA performance in a next-generation early detection strategy for colorectal cancer

Abstract Introduction: Colorectal cancer (CRC) increasingly impacts younger individuals, emphasizing the urgent need for earlier, more accurate, and minimally invasive screening methods. Current plasma-based liquid biopsy tests, typically relying on cell-free DNA (cfDNA) and/or free proteins, effectively detect CRC but fall short in identifying early pre-cancerous lesions such as advanced adenomas (AAs). We hypothesized that a multi-omic approach combining cfDNA and extracellular vesicle (EV)-derived biomarkers could enhance early detection of CRC, including AAs, which for current cfDNA-based assays have poor performance of detection. Methods: A predefined statistical analysis plan guided the enrollment of approximately 230 individuals undergoing colonoscopy across three independent clinical sites. Plasma cfDNA methylation levels were analyzed using a combination of enzymatic conversion and enrichment of a panel covering over 4 million CpG sites. Comprehensive profiling of EV-derived long RNAs to measure gene expression and splice variants was performed using a hybrid capture panel covering 20, 000 protein coding RNAs and 20, 000 lncRNAs. Unbiased EV small RNA profiling was performed to measure all small non-coding RNAs between 18-200nts in length. Levels for up to 552 cancer associated proteins in EVs were measured using the Olink platform. Stringent quality control ensured reproducibility and no significant batch effects were seen across sites. Feature selection and model development employed a stratified, 5-fold cross-validation approach with a support vector machine (SVM) model. Sensitivities were evaluated using a Clopper-Pearson estimator at a target specificity of 91%, and prevalence-adjusted. Results: Among the omics channels, EV-derived RNA demonstrated the highest sensitivity for detecting both CRC and pre-cancerous AAs at 87% (±6%) and 34% (±7%) respectively, significantly outperforming cfDNA methylation (76% ±7% and 10% ±4% respectively), EV-derived protein (85% ±7% and 24% ±7% respectively), and EV-derived miRNA (49% ± 9% and 21% ±7% respectively). Cross-validation identified key features relevant to CRC biology, with pathway enrichment analysis of EV-derived gene expression revealing significant activation of immune and inflammatory pathways. Conclusion: While most EV derived analyte surpassed the performance achieved by cfDNA methylation, EV long RNAs emerged as the most sensitive analyte for early detection of CRC and pre-cancerous AAs. This study highlights the potential of EV-derived multi-omic biomarkers to address critical gaps in CRC screening, offering a transformative approach to improving early detection and patient outcomes. Combining EV RNA with cfDNA may give synergistic improvements in performance when applied in a larger cohort. Citation Format: Sudipto Kumar Chakrabortty, J Christian Ray, Dulaney Miller, Kyle Manning, Yang Yang, Shuran Xing, T Jeffrey Cole, Christopher J. Benway, Sinead Nguyen, Allan George, Julia Doo, Yiyuan Yao, Pawan Upadhyay, Joe Johnson, Sonia Kumar, Sivakumar Gowrisankar, Brian Haynes, Johan Skog. Extracellular vesicle-derived biomarkers surpass cfDNA performance in a next-generation early detection strategy for colorectal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1909.