Integrated multi-omics analysis of cfDNA and protein biomarkers for multi-cancer early detection from a single blood draw

Non-invasive multi-cancer early detection (MCED) represents a promising strategy for enabling timely clinical intervention and improving patient survival outcomes. However, current MCED approaches mainly relying on single-feature analyses have demonstrated suboptimal performance, particularly in detecting early-stage malignancies. To address this limitation, we developed a novel MCED test that integrates the analysis of cell-free DNA (cfDNA) methylation, somatic mutations, fragmentation patterns, and plasma protein biomarkers from a single blood draw. A targeted methylation-sensitive restriction endonuclease sequencing method was established to enable simultaneous profiling of three distinct cfDNA molecular features. We evaluated the performance of this integrated test in a retrospective, large-scale, multicenter cohort comprising 5,552 participants, including 3,193 patients across 11 cancer types and 2,359 non-cancer controls. In an independent validation set, the integrated multi-omics classifier achieved an AUC of 0.957, with an overall sensitivity of 81.1% and a specificity of 95.6%. For stage I cancers, the classifier demonstrated a sensitivity of 65.3%. Additionally, the multi-omics model accurately predicted the tissue of origin of malignancies with a top-3 accuracy of 86.1% in the independent validation set.