Applications of AI-Assisted Liquid Biopsy for Early Cancer Detection and Treatment Monitoring: A Systematic Review of Current Evidence

Le résumé fourni par la source

Background Early cancer diagnosis is also a significant clinical issue because both traditional methods of imaging and tissue biopsy are not capable of achieving early-stage, minimal residual disease (MRD) and newdeveloped treatment resistance. Liquid biopsy is an approach that is based on circulating tumor DNA (ctDNA), circulating tumor cells (CTCs), exosomal RNA/miRNA, and epigenetic signatures and provides a minimally invasive instrument that can be used to profile tumors in real-time. The recent developments in high-throughput sequencing and methylation-based assays have greatly enhanced the level of analytical sensitivity and therefore, the detection of early malignancy is better and also therapeutic response can be effectively monitored. Objective To conduct a systematic review and synthesis of the evidence regarding the diagnostic accuracy, prognostic utility and treatment-monitoring performance of the liquid biopsy biomarkers in the detection of early cancer in the solid tumours, to evaluate the methodological trends and sources of heterogeneity on clinical applicability. Methods There was a systematic search of PubMed, EMBASE, Scopus, Web of Science, Cochrane Library, IEEE Xplore, and Google Scholar to identify different papers published during 2010-2025. Research evaluating ctDNA, CTCs, exosomes, or methylation analyzes in the diagnosis, prognostication, or MRD monitoring was included. Two reviewers were screening data and extracting data independently according to PRISMA 2020 guidelines. Quality of the methodology was evaluated based on the QUADAS-2 and ROBINS-I. Random-effects pooled meta-analyses were based on the sensitivity, specificity, odds ratios (ORs), and area under the receiver-operating curve (AUC). Correlations were used to examine correlations between variables of study design and diagnostic performance. Results Out of 6,812 original entries, 42 studies were taken into account according to which over 28,400 cancer patients and 33,000 controls were included, and ctDNA and methylation based assays proved to be the most successful in the diagnostic process with pooled sensitivity of 0.78 and specificity of 0.83 and AUC of 0.87. CtDNA panels utilizing methylation reached an AUC of as high as 0.92 and multi-omics 0.94. Exosomal miRNA biomarkers performed well (AUC 0.86 -0.92) and CTC tests were of moderate accuracy (AUC 0.79). In treatment monitoring, early ctDNA clearance showed a therapeutic response (HR 0.42) and increasing levels of ctDNA predicted radiologic relapse 3-6 months before it happened. Diagnostic (r = 0.72) and serial (r = 0.68) sampling correlated with depth of sequencing, whereas preanalytic variability decreased the accuracy (r = - 0.54). External validation was only done in 36% of studies. Conclusions Liquid biopsy offers a solid and minimally invasive platform of early cancer diagnosis, therapeutic response, and MRD, and longitudinal ctDNA dynamics provides powerful predictive information of therapeutic response and recurrence. Nevertheless, large-scale clinical utilization is still hampered by inconsistencies in assay procedures, poor standardization and sub-optimal external validation. The future studies ought to focus on multi-center prospective research, integrative pipelines, and multi-omics to facilitate clinical translation.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.