Potential of miRNAs in urinary extracellular vesicles for management of active surveillance in prostate cancer patients

Abstract

Background: Active surveillance is an alternative to radical treatment for patients with low-risk prostate cancer, which could also benefit some patients with intermediate risk. We have investigated the use of miRNA in urinary extracellular vesicles to stratify these patients.

Methods: NGS was performed to profile the miRNAs from small urinary extracellular vesicles in a cohort of 70 patients with prostate cancer ISUP Grade 1, 2 or 3. The most promising candidates were then analysed by RT-qPCR in a new cohort of 60 patients.

Results: NGS analysis identified nine miRNAs differentially expressed in at least one of the comparisons. The largest differences were found with miR-1290 (Grade 3 vs. 1), miR-320a-3p (Grade 3 vs. 2) and miR-155-5p (Grade 2 vs. 1). Combinations of 2-3 miRNAs were able to differentiate between two ISUP grades with an AUC 0.79-0.88. RT-qPCR analysis showed a similar trend for miR-186-5p and miR-30e-5p to separate Grade 3 from 2, and miR-320a-3p to separate Grade 2 from 1.

Conclusions: Using NGS, we have identified several miRNAs that discriminate between prostate cancer patients with ISUP Grades 1, 2 and 3. Moreover, miR-186-5p, miR-320a-3p and miR-30e-5p showed a similar behaviour in an independent cohort using an alternative analytical method. Our results show that miRNAs from urinary vesicles can be potentially useful as liquid biopsies for active surveillance.

Fig. 1. Protein analysis of the 10,000×g and 100,000×g pellets.

Urine samples were first sequentially centrifuged at 2000×g and 10,000×g. Then the 10,000×g pellet was washed with 20 ml PBS and pelleted again at the same speed, while the supernatant was centrifuged at 100,000×g, washed twice with PBS and pelleted again at 100,000×g. Both pellets were resuspended in 100 µl PBS and the same volume of sample was analysed by silver staining (a) and western blot (b).

Fig. 2. NGS -identified miRNAs in urinary EVs of prostate cancer patients showing differential expression between ISUP grades.

The plots show the normalised counts of miRNAs whose expression was significantly changed between Grade 3 and Grade 1 (a–e), Grade 3 and Grade 2 (f–h) and Grade 2 and Grade 1 (i). Central bars represent the mean and whiskers represent the standard deviation of all the samples.

Fig. 3. Evaluation of the predictive power of the miRNAs identified by NGS.

a–c ROC curves and AUC values for the best models generated using the NGS data for each one to one-grade comparison. Five miRNAs: miR-1290, miR-1246, miR-320b, miR-204-3p and miR-143-3p. Three miRNAs: miR-320a-3p, miR-30e and miR-186. d Repeated fivefold cross-validation was used to estimate the behaviour of the miRNAs in an independent cohort. The graph shows, for each comparison, the calculated AUC of each individual miRNA and the best combinations of several miRNAs.

Fig. 4. miRNAs in urinary EVs from a new cohort of 60 prostate cancer patients (20 per ISUP grade) were analysed using RT-qPCR.

a–c Comparison of the expression of miR-186-5p and miR-30e-5p between Grades 3 and 2, and miR-320a-3p between Grades 2 and 1. P values: a—0.049, b—0.1, c—0.012. d–f ROC curves and AUC of the predictive models generated with miRNAs and PSA levels.