AbstractIntroduction:Cancer remains a leading cause of mortality worldwide. The multi-cancer early detection (MCED) test complements current screening methods improving early detection and treatment outcomes. While most MCED tests focus on community populations, our MCTarg models were specifically designed to address both low-risk and high-risk populations (e.g., those with conditions such as ulcerative colitis, adenomatous polyps, chronic bronchitis, tuberculosis, atrophic gastritis, and H. pylori infection), tailoring the approach to the unique characteristics and needs of each group. Here, we present the performance of our Multiple Cancer Target (MCTarg), which utilizes a single plasma metabolite test combined with machine learning technology to screen for the most prevalent cancer types—specifically lung cancer (LC), gastric cancer (GC), and colorectal cancer (CRC).Methods:We enrolled 951 cancer patients (540 LC, 203 GC, 208 CRC) and 889 non-cancer individuals (healthy controls and those with benign diseases) across three centers. Plasma samples were analyzed using GC-MS and LC-MS multi-platforms. Participants were divided into a discovery cohort for identifying cancer signatures and optimizing models, and an internal validation cohort for performance evaluation. External validation was conducted on an independent cohort (108 cancer patients, 125 non-cancer individuals) from two additional centers. Furthermore, the discriminatory ability of these metabolites between the non-cancer and multi-cancer groups was confirmed using targeted metabolomic analysis.Results:Two screening models, MCTarg-1 for low-risk populations and MCTarg-2 for high-risk populations, were established for various clinical scenarios. MCTarg-1 for low-risk populations exhibited 98.9% sensitivity at 98.0% specificity in the internal validation cohort and 93.5% sensitivity at 95.0% specificity in the external validation cohort. MCTarg-2 for high-risk populations yielded 59.9% sensitivity at 94.4% specificity internally, and 64.8% sensitivity at 85.6% specificity externally. For early-stage (I-II) patients in the external cohort, sensitivities were 79.1% for MCTarg-1 and 69.2% for MCTarg-2. With 66 metabolite biomarkers identified, MCTarg-1 exhibited 80.6% sensitivity at 98.0% specificity in the internal validation cohort, and 73.3% sensitivity at 86.7% specificity in the external validation cohort. MCTarg-2 also showed 69.4% sensitivity at 91.7% specificity, and 57.4% sensitivity at 84.0% specificity, respectively.Conclusions:Our MCTarg has demonstrated outstanding and competitive performance across various risk groups. With further large-scale validation and the inclusion of additional cancer types, MCTarg has the potential to become a universally applicable, simple, and cost-effective method, enabling early detection and localization of common cancers in large populations.Citation Format:Yan Li, Xinping Xu, Chunyan Zeng, Bei Qing, Yun He, Yanlong Liu, Guodong Song, Jianhua Hu, Tianqi Shao, Li Liu, Qingyan Wei, Shuqi Yu, He Wen, Junyuan Hu, Wei Zhang, Youxiang Chen, Zhenkun Xia. MCTarg: A plasma-based metabolic biomarker model for multi-cancer early detection [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr LB255.