Liquid biopsy, multi-cancer early detection, and artificial intelligence: new frontiers in cancer screening from a...

Early cancer detection remains a central challenge in oncology because many lethal tumors are diagnosed after curative opportunities have narrowed, whereas current organ-specific screening methods cover only a limited number of cancer types and may be constrained by invasiveness, cost, accessibility or stage-dependent sensitivity. Liquid biopsy, multi-cancer early detection (MCED) and artificial intelligence (AI) are rapidly reshaping this field, but their clinical implications require careful interpretation. This review critically evaluates major liquid-biopsy analytes, including circulating tumor DNA, cell-free DNA methylation and fragmentomics, circulating tumor cells, extracellular vesicles, non-coding RNAs, tumor-educated platelets and multi-omics signatures, with emphasis on intended use, clinical maturity, tissue-of-origin value and translational limitations. A distinctive feature of this review is the integration of tumor-derived signals with host-response and immunological readouts, including peripheral blood mononuclear cell-based monitoring, immune-cell-derived extracellular vesicles, exosomal immune-checkpoint molecules and inflammatory confounders, thereby framing liquid biopsy as both a cancer-detection tool and a window into tumor-immune interactions. We further discuss MCED as a clinical care pathway rather than an isolated blood test, highlighting the importance of positive and negative predictive values, cancer prevalence, diagnostic-resolution pathways, false-positive workup, overdiagnosis, mortality benefit, cost-effectiveness and equitable access. The role of AI is examined in relation to model development, multimodal fusion, tissue-of-origin prediction, calibration, interpretability, bias, generalizability and clinical implementation. Across these technologies, a key translational message is that technical detectability is not equivalent to clinical readiness. While selected assays have entered defined clinical or guideline-supported settings, many emerging biomarkers and AI-enabled models remain investigational or translational. Future progress will depend on standardized workflows, prospective validation in representative populations, evidence of clinical utility, regulatory and ethical oversight, and integration with established screening and diagnostic systems.