A novel expressed prostatic secretion (EPS)-urine metabolomic signature for the diagnosis of clinically significant prostate cancer

Abstract

Objective: Significant efforts are currently being made to identify novel biomarkers for the diagnosis and risk stratification of prostate cancer (PCa). Metabolomics can be a very useful approach in biomarker discovery because metabolites are an important read-out of the disease when characterized in biological samples. We aimed to determine a metabolomic signature which can accurately distinguish men with clinically significant PCa from those affected by benign prostatic hyperplasia (BPH).

Methods: We first performed untargeted metabolomics using ultrahigh-performance liquid chromatography tandem mass spectrometry on expressed prostatic secretion urine (EPS-urine) from 25 patients affected by BPH and 25 men with clinically significant PCa (defined as Gleason score ≥ 3 + 4). Diagnosis was histologically confirmed after surgical treatment. The EPS-urine metabolomic approach was then applied to a larger, prospective cohort of 92 consecutive patients undergoing multiparametric magnetic resonance imaging for clinical suspicion of PCa prior to biopsy.

Results: We established a novel metabolomic signature capable of accurately distinguishing PCa from benign tissue. A metabolomic signature was associated with clinically significant PCa in all subgroups of the Prostate Imaging Reporting and Data System (PI-RADS) classification (100% and 89.13% of accuracy when the PI-RADS was in range of 1-2 and 4-5, respectively, and 87.50% in the more critical cases when the PI-RADS was 3).

Conclusions: A combination of metabolites and clinical variables can effectively help in identifying PCa patients that might be overlooked by current imaging technologies. Metabolites from EPS-urine should help in defining the diagnostic pathway of PCa, thus improving PCa detection and decreasing the number of unnecessary prostate biopsies.

Keywords: EPS-urine; Prostate; cancer; diagnosis; metabolomics; prediction.

Figure 1

Untargeted metabolomics analysis of BPH and PCa patients (C18 chromatographic separation): unsupervised principal component analysis score for C18 and BEH amide column separation in positive (A, C) and negative (B, D) modes assessing the clustering of the BPH (red) and PCa (green) patients performed on identified metabolites upon probabilistic quotient normalization. BPH, benign prostatic hyperplasia; PCa, prostate cancer; BPH, benign prostatic hyperplasia.

Figure 2

EPS metabolomic signature of BPH and PCa patients. (A) Seventeen significantly (P < 0.05) different metabolites between BPH and PCa patients. Chromatographic column (C18 or BEH) and mass spectrometry polarity (positive or negative) are indicated. (B) The most significant metabolic pathways for BPH and PCa are represented by bigger/red dots with higher −log P value. EPS, expressed prostatic secretion; BPH, benign prostatic hyperplasia; PCa, prostate cancer; pos, positive; neg, negative.

Figure 3

Structure of the meta classifier: the PI-RADS value is used to choose a different set of variables as input for the corresponding classifier. The metabolites for each PI-RADS level are indicated along with the polarity (positive or negative) for mass spectrometry acquisition. PI-RADS, Prostate Imaging Reporting and Data System; pos, positive; neg, negative.

Figure 4

Confusion matrix: true-positive, true-negative, false-positive, and false-negative rates in the prediction model. (A) PI-RADS 1-2. (B) PI-RADS 3. (C) PI-RADS 4-5.