Clinica Chimica Acta; International Journal of Clinical Chemistry · Nov 18, 2026

AI-assisted liquid biopsy for early breast cancer detection: a laboratory translation framework for future multimodal integration

By Alhazmi BA, Jaiswal V, Rekha A et al.

Early breast cancer detection is imaging-dependent in clinical laboratory medicine, and there are important limitations for interval cancers and women with dense breasts. Liquid biopsy is being developed as an adjunct to imaging and tissue diagnosis. There are two classes of circulating cell-free DNA (cfDNA) features and a separate class of extracellular vesicle (EV)-associated biomarkers, such as microRNAs, long non-coding RNAs, and proteins, to consider in the proposed framework. These literature reviews primarily focus on combinations of individual modalities, cfDNA-only combinations, cfDNA-protein panels, and multicancer models. There is a lack of prospective clinical validation of a machine learning model that integrates all three modalities for the early detection of breast cancer. We categorized performance evidence based on the purpose of the test for average-risk screening, dense breast triage, symptomatic assessment, high-risk surveillance, recurrence monitoring for breast cancer, and metastatic disease detection, and differentiated between direct breast cancer evidence and indirect methodological analogies. Translation involves modality-specific specimen processing, modality-specific analytical and clinical validation, locked/version-controlled pipelines, patient-, site-, and time-level data partitioning, leakage control, pre-specified thresholds and abstention rules, calibration assessment, external validation, subgroup analyses, reportable quality-control metrics, and defined revalidation triggers.

doi.org/10.1016/j.cca.2026.122987PubMed 43051393Biomarkers

AgingResearch.News