Analysis of Circulating Biomarkers for Minimally Invasive Early Detection of Breast Cancer

Breast cancer (BC) is the malignancy with the highest incidence and mortality rates among women. Numerous studies explored cell-free circulating (cfc) microRNAs (miRNAs) as diagnostic biomarkers of BC. However, their results were inconsistent with few intersecting miRNA panels. In a meta-analysis, we evaluated the overall diagnostic performance as well as the sources of heterogeneity between studies on BC detection using cfc miRNA. The findings on sources of heterogeneity would then be applied to our second project, which aimed to identify circulating miRNA ratios associated with BC in women attending mammography screening. On 56 studies that investigated diagnostic circulating miRNAs by utilising Real-Time Quantitative Reverse Transcription PCR (RT-qPCR), pooled sensitivity and specificity of 0.85 [0.81 to 0.88] and 0.83 [0.79 to 0.87] were obtained, respectively. Subgroup analysis revealed a comparable pooled diagnostic performance between studies using serum (sensitivity: 0.87 [0.81 to 0.91]; specificity: 0.83 [0.78 to 0.87]) and plasma (sensitivity: 0.83 [0.77 to 0.87]; specificity: 0.85 [0.78 to 0.91]) as specimen type. Additionally, miRNA(s) based on endogenous normalisers tend to have a higher diagnostic performance than miRNA(s) based on exogenous ones. A nested case–control study was conducted on plasma samples of 65 cases and 66 controls (discovery) and 32 cases and 127 controls (validation). Small-RNA sequencing was carried out on the discovery cohort, and to overcome the normalisation issue in RT-qPCR, we computed miRNA ratios and those associated with BC were selected by two-sample Wilcoxon test and lasso penalised logistic regression. Assessment by RT-qPCR of 20 candidate miRNA ratios was carried out as a platform validation. To identify the most promising biomarkers, penalised logistic regression was further applied to candidate miRNA ratios alone or in combination with non-molecular factors. In the resulting model, LASSO regression selected seven miRNA ratios (miR-199a-3p_let-7a-5p, miR-26b-5p_miR-142-5p, let-7b-5p_miR-19b-3p, miR-101-3p_miR-19b-3p, miR-93-5p_miR-19b-3p, let-7a-5p_miR-22-3p and miR-21-5p_miR-23a-3p), together with the interaction term of centred BMI and menopausal status, lifestyle score and breast density. The ROC AUC of the model was 0.79. After applying the model to the validation cohort and recalibrating the predicted probabilities, an ROC AUC of 0.87 was obtained. In this project, we reaffirmed the ability of circulating microRNAs to diagnose BC, analysed the sources of heterogeneity and discussed the problems of standardisation and reproducibility of results. Additionally, we identified cfc miRNAs potentially useful for BC detection in a screening setting.