1BackgroundProstate cancer (PCa) diagnosis remains challenged by the limited specificity of prostate-specific antigen (PSA) testing, which cannot reliably distinguish malignancy from benign prostatic hyperplasia (BPH). MicroRNAs (miRNAs) are emerging candidates for liquid biopsy-based diagnostics, but most studies assess expression in isolation within a single compartment (biological source - Tissue, blood, serum, urine etc.), overlooking both compartment-specific behavior and the coordinated relationships among miRNAs.
MethodsWe profiled four candidate miRNAs -- miR-19b-3p, miR-21-5p, miR-101-3p and miR-375-3p, across four biological compartments (prostate tumor tissue, urine, serum, and blood) in 179 patients undergoing prostate biopsy for clinical suspicion of PCa (104 PCa, 75 BPH) using qRT-PCR. Urinary exosomal RNA was isolated with a commercial exosome isolation kit so from here onwards this compartment will be referred to as urine. Differential expression was quantified using Cohens d; inter-miRNA coordination was assessed via Spearman correlation and differential correlation ({Delta}r) analysis; and a compartment-level network rewiring score was derived as the sum of |{Delta}r| across miRNA pairs. Cross-compartment structural alignment was evaluated by comparing correlation patterns at the population level. Diagnostic models combining PSA, age, and urinary exosomal-miRNA features were evaluated using Logistic Regression, Elastic Net Logistic Regression and Naive Bayes classifiers under leave-one-out cross-validation (LOOCV).
Singh, S. et al. · CC-BY 4.0