Circulating microRNAs combined with PSA for accurate and non-invasive prostate cancer detection

Authors:
Publisher website
Google Scholar
Abstract
The dosage of prostate-specific antigen (PSA), an easily evaluable and non-invasive biomarker, has made early detection of prostate cancer (PCa) possible. However, it leads to high percentages of unnecessary biopsies and may miss aggressive tumors in men with PSA levels below 4 ng/ml. Therefore, we propose to combine circulating microRNAs (miRs) with PSA, to improve the diagnostic route for PCa. Plasma miR profiling identified candidate diagnostic miRs in a discovery cohort of 60 tumors and 60 controls (men with benign prostatic hyperplasia or healthy donors). Linear models with an empirical Bayesian approach and multivariate penalized logistic regression were applied to select tumor-associated miRs and/or clinical variables. A classifier was developed and tested on a validation cohort of 68 tumors and 174 controls consecutively collected, where miRs were evaluated by quantitative real-time polymerase chain reaction. A classifier based on miR-103a-3p, let-7a-5p and PSA could detect both overall and clinically significant tumors better than PSA alone, even in 50-69 years aged men with PSA ≤ 4 ng/ml. Even in the validation cohort, the classifier performed better than PSA alone in terms of specificity and positive predictive value, allowing to correctly identify eight out of nine tumors undetected by PSA, including three high-risk and three tumors in 50-69 years old men. Of carriers of non-malignant lesions with PSA in the 4-16 ng/ml interval, who may avoid unnecessary biopsies, 34% were correctly identified. Coupling two circulating miRs with PSA could be a useful strategy to diagnose clinically significant PCa and avoid an important fraction of unnecessary biopsies.
Funding Information
  • Associazione Italiana per la Ricerca sul Cancro (MFAG 11742)
  • Compagnia di San Paolo
  • Italian Institute for Genomic Medicine
  • Fondazione Umberto Veronesi
  • Post-Doctoral

This publication has 30 references indexed in Scilit: