EMERGING LIQUID BIOPSY BIOMARKERS FOR EARLY CANCER DETECTION: SYSTEMATIC REVIEW OF CFDNA, CTDNA, METHYLATION, AND FRAGMENTOMICS

Abstract

Background: Early cancer detection is central to reducing cancer-associated morbidity and mortality, yet established population screening is available for only a subset of malignancies. Liquid biopsy permits minimally invasive interrogation of tumor-derived molecular signals in blood, particularly circulating cell-free DNA (cfDNA) and its tumor-derived component, circulating tumor DNA (ctDNA). Objective: To systematically review the biological basis, diagnostic performance, clinical evidence, strengths, limitations, and translational potential of cfDNA concentration/integrity, mutation-based ctDNA, DNA methylation, and fragmentomics for early and multi-cancer detection. Methods: A structured literature search was undertaken for human studies published through June 2026. The search identified 676 records: 412 from PubMed/MEDLINE, 238 from other academic databases, and 26 through citation/reference searching. After removal of 124 duplicate records and 8 other records, 544 records underwent title and abstract screening; 454 were excluded. Ninety reports were sought for retrieval, 6 were not retrieved, and 84 full-text reports were assessed for eligibility. Sixty-six full-text reports were excluded for predefined reasons, leaving 18 primary studies in the qualitative synthesis. Reporting followed PRISMA 2020 and PRISMA-DTA principles, and diagnostic-accuracy studies were appraised using QUADAS-3. Owing to marked clinical and analytical heterogeneity, quantitative meta-analysis was not performed. Results: Simple cfDNA concentration and integrity measures showed cancer-associated changes but limited tumor specificity. Mutation-based ctDNA assays offered high molecular specificity but were constrained by extremely low tumor fractions in stage I disease and by clonal hematopoiesis. Methylation-based assays provided a broader signal and strong cancer-signal-origin information; in an independent CCGA validation cohort, specificity was 99.5%, while sensitivity increased from 16.8% in stage I to 90.1% in stage IV disease. Fragmentomic approaches captured genome-wide differences in fragment size, coverage, nucleosome positioning, and end characteristics and demonstrated high discriminative performance with relatively shallow sequencing. Multimodal strategies generally improved information density. Among 17 diagnostic-accuracy studies evaluated with QUADAS-3, participant selection was the dominant concern: 13/17 (76.5%) had high risk in this domain. The 2026 NHS-Galleri randomized trial did not meet its prespecified primary endpoint for reducing combined stage III-IV cancers, although stage IV cancers decreased and stage I-II diagnoses increased. Conclusion: Methylation and fragmentomics currently provide the most information-dense cfDNA-based approaches for early cancer detection, while mutation-based ctDNA retains high biological specificity. Multimodal integration is likely to provide the greatest diagnostic potential. However, high diagnostic accuracy does not establish population-screening benefit; randomized evidence demonstrating reductions in advanced cancer and mortality, with acceptable harms and costs, remains essential.