Diagnostic accuracy of circulating microRNAs and lipidomic biomarkers for early breast cancer detection: A systematic review and meta-analysis

Abstract

Background Breast cancer outcomes improve substantially with earlier detection, yet mammography performance can be limited in dense breasts and may lead to false-positive investigations. Circulating microRNAs (miRNAs) and lipidomic/metabolomic signatures have emerged as promising minimally invasive biomarkers that could complement imaging for early-stage detection. The aim of this systematic review and meta-analysis is to evaluate the diagnostic accuracy of circulating microRNAs and lipidomic/metabolomic biomarkers for early breast cancer detection and to explore between-study heterogeneity. Materials and methods A PRISMA 2020–compliant systematic review and meta-analysis were conducted. MEDLINE (PubMed), Web of Science, Scopus, Springer, ScienceDirect, and the Cochrane Library were searched for eligible diagnostic studies assessing circulating microRNAs and lipidomic/metabolomic biomarkers measured in serum or plasma. No language restrictions were applied. Non-English reports were screened and, when potentially eligible, translated for full-text assessment and data extraction. Studies using whole blood were excluded. Study quality was assessed using QUADAS-2. Random-effects models (DerSimonian–Laird) pooled sensitivity, specificity, likelihood ratios, diagnostic odds ratio (DOR), and summary ROC (SROC) area under the curve (AUC). Heterogeneity was evaluated using Q and I 2 statistics, and small-study effects were assessed using Deeks’ test. Results Thirty-two studies (2015–2025) comprising 6,935 participants (3,697 breast cancer cases; 3,238 controls) were included. Pooled sensitivity was 0.87 (95% CI [0.83, 0.90]) and pooled specificity was 0.84 (95% CI [0.79, 0.88]), with pooled DOR 46.10 (95% CI [27.80, 76.70]), PLR 5.17 (95% CI [3.99, 6.68]), NLR 0.16 (95% CI [0.13, 0.21]), and SROC AUC 0.92 (95% CI [0.87, 0.94]). Heterogeneity was substantial (I 2  = 88.29% for sensitivity; I 2  = 90.20% for specificity). In subgroup analyses, serum-based studies showed higher pooled specificity than plasma-based studies. A formal threshold effect assessment did not reveal a statistically significant correlation between sensitivity and false-positive rate (Spearman ρ = − 0.334, p  =  0.062). Deeks’ test suggested potential small-study effects (p = 0.043). Conclusions Circulating microRNA and lipidomic/metabolomic biomarkers demonstrate strong overall diagnostic performance for breast cancer detection; however, substantial heterogeneity and potential small-study effects limit their immediate clinical translation. Given that most included studies used retrospective case-control designs rather than prospective screening cohorts, these biomarkers are best regarded as investigational, complementary tools rather than replacements for mammography at this stage. Future large, prospective, standardized studies with harmonized pre-analytics and prespecified thresholds are needed to support the implementation of screening or triage pathways.