Extracellular RNAs from Whole Urine to Distinguish Prostate Cancer from Benign Prostatic Hyperplasia

Abstract

RNAs, especially non-coding RNAs (ncRNAs), are crucial players in regulating cellular mechanisms due to their ability to interact with and regulate other molecules. Altered expression patterns of ncRNAs have been observed in prostate cancer (PCa), contributing to the disease's initiation, progression, and treatment response. This study aimed to evaluate the ability of a specific set of RNAs, including long ncRNAs (lncRNAs), microRNAs (miRNAs), and mRNAs, to discriminate between PCa and the non-neoplastic condition benign prostatic hyperplasia (BPH). After selecting by literature mining the most relevant RNAs differentially expressed in biofluids from PCa patients, we evaluated their discriminatory power in samples of unfiltered urine from 50 PCa and 50 BPH patients using both real-time PCR and droplet digital PCR (ddPCR). Additionally, we also optimized a protocol for urine sample manipulation and RNA extraction. This two-way validation study allowed us to establish that miRNAs (i.e., miR-27b-3p, miR-574-3p, miR-30a-5p, and miR-125b-5p) are more efficient biomarkers for PCa compared to long RNAs (mRNAs and lncRNAs) (e.g., PCA3, PCAT18, and KLK3), as their dysregulation was consistently reported in the whole urine of patients with PCa compared to those with BPH in a statistically significant manner regardless of the quantification methodology performed. Moreover, a significant increase in diagnostic performance was observed when molecular signatures composed of different miRNAs were considered. Hence, the abovementioned circulating ncRNAs represent excellent potential non-invasive biomarkers in urine capable of effectively distinguishing individuals with PCa from those with BPH, potentially reducing cancer overdiagnosis.

Keywords: BPH; PCa; diagnosis; liquid biopsy; lncRNA; miRNA; molecular signature; ncRNA.

Figure 6

Receiver operating characteristic (ROC) curves of dysregulated miRNAs obtained by computing real-time PCR data. The area under the curve (AUC) value of each target is shown in Table 4. (A) Univariable and multivariable ROC curves; (B) ROC curves of miRNA molecular signatures.

Figure 7

ROC curves of dysregulated lncRNAs and mRNAs obtained by computing real-time PCR data. The AUC of each target is shown in Table 5.

Figure 1

Expression levels of prostate cancer associated 3 (PCA3) and prostate cancer associated transcript 18 (PCAT18) in whole urine, pellet, and supernatant samples. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the endogenous control. A test set of 8 urine samples from prostate cancer (PCa) patients and 8 from benign prostatic hyperplasia (BPH) patients was used for the analyses. To determine the statistical significance, a t-test was applied (* = p-value < 0.05).

Figure 2

Expression levels of PCAT18, GAPDH, miR-16-5p, and miR-29c-3p in concentrated and non-concentrated urine samples. The graphs reveal a negligible gain in Ct values within the concentrated samples. The analysis was conducted on a set of 5 BPH urine samples. For each transcript, expression values are shown as 40-Ct in all tested samples, namely, whole and concentrated urine. For each sample, the starting volume is shown on the right y-axis. For the concentrated urine, the waste fraction was also analyzed to assess the eventual elution and loss of RNA molecules.

Figure 3

The graphs illustrate the mean 2^−ΔCt of selected long RNAs (A) and microRNAs (miRNAs) (B) in PCa and BPH whole urine samples. The threshold we set (cycle threshold [Ct] < 30) corresponds to a value of 2^−ΔCt < 0.1. A test set of 15 urine samples from PCa patients and 15 from BPH patients was used for the analyses.

Figure 4

Box plots showing the expression levels of long non-coding RNAs (lncRNAs) and mRNAs selected and analyzed in a cohort of 50 PCa samples and 50 BPH samples used as controls. Y-axes report the value of ΔCt multiplied by −1. To determine the statistical significance, a t-test was applied (* = p-value < 0.05).

Figure 5

Box plots showing the expression levels of the miRNAs selected and analyzed in a cohort of 50 PCa samples and 50 BPH samples used as controls. Y-axes report the value of ΔCt multiplied by −1. To determine the statistical significance, a t-test was applied (* = p-value < 0.05, ** = p-value < 0.01, *** = p-value < 0.005).

Figure 8

Box plots showing the expression levels of the lncRNAs and mRNAs selected and analyzed by ddPCR in a sub-cohort of 40 PCa samples and 40 BPH samples. Y-axes report the value of copies/µL. To determine the statistical significance, a t-test was applied (* = p-value <0.05).

Figure 9

Box plots showing the expression levels of the miRNAs selected and analyzed by ddPCR in a sub-cohort of 40 PCa samples and 40 BPH samples used as controls. Y-axes report the value of copies/µL. To determine the statistical significance, a t-test was applied (*** = p-value < 0.001, **** = p-value < 0.0001).

Figure 10

ROC curves generated using ddPCR data and prostate-specific antigen (PSA) values. The AUC of each target is shown in Table 8. (A) Univariable and multivariable ROC curves; (B) ROC curves of miRNA molecular signatures.

Figure 11

Heatmap showing correlations between the expression of miRNAs (expressed as counts/µL obtained by ddPCR) and clinicopathological data (PSA and International Society of Urological Pathology grade [ISUP] values). The scale bar shows r-values calculated through Spearman correlation tests, ranging from green (negative correlation) to red (positive correlation). Statistical significance is represented by the number of asterisks (* = p-value < 0.05; ** = p-value < 0.005; *** = p-value <0.0005).