Skip to main content
PMC home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2016 Sep 6;113(38):10655–10660. doi: 10.1073/pnas.1611596113

Circulating microRNA signature for the diagnosis of very high-risk prostate cancer

Ali H Alhasan a,b,1, Alexander W Scott b,c,1, Jia J Wu a, Gang Feng d, Joshua J Meeks e,f,2, C Shad Thaxton b,e,f,g,2, Chad A Mirkin a,b,c,h,2
PMCID: PMC5035901  PMID: 27601638

Significance

Serum microRNAs (miRNAs) have emerged as potential noninvasive biomarkers to diagnose prostate cancer (PCa), the most common noncutaneous malignancy among Western men. However, intermediate grades of PCa cannot be distinguished from aggressive forms using current miRNA signatures due to the heterogeneity of PCas. Recently, a high-throughput, spherical nucleic acid-based miRNA expression profiling platform, called the Scano-miR bioassay, was developed to measure the expression levels of miRNAs with both high sensitivity and specificity. By studying serum miRNAs of PCa using the Scano-miR bioassay, we identified a unique molecular signature specific for very high-risk aggressive PCa. This molecular signature will assist in differentiating patients who may benefit from therapy from those who can be closely monitored on active surveillance.

Keywords: prostate cancer, microRNA, Scano-miR, spherical nucleic acid, biomarker

Abstract

We report the identification of a molecular signature using the Scano-miR profiling platform based on the differential expression of circulating microRNAs (miRNA, miR) in serum samples specific to patients with very high-risk (VHR) prostate cancer (PCa). Five miRNA PCa biomarkers (miR-200c, miR-605, miR-135a*, miR-433, and miR-106a) were identified as useful for differentiating indolent and aggressive forms of PCa. All patients with VHR PCa in the study had elevated serum levels of miR-200c. Circulating miR-433, which was differentially expressed in patients with VHR versus low-risk (LR) forms of PCa, was not detectable by quantitative real-time PCR in samples from healthy volunteers. In blind studies, the five miRNA PCa biomarkers were able to differentiate patients with VHR PCas from those with LR forms as well as healthy individuals with at least 89% accuracy. Biological pathway analysis showed the predictive capability of these miRNA biomarkers for the diagnosis and prognosis of VHR aggressive PCa.

Prostate cancer (PCa) is the most common noncutaneous malignancy among men in the United States and the second most common cause of cancer mortality (1, 2). Despite the prevalence of PCa, there are no specific and accurate diagnostic or prognostic biomarkers to differentiate tumor aggressiveness. Although serum prostate-specific antigen (PSA) concentration is used as a routine screening tool for PCa, up to 11% of men with a PSA < 2.0 ng/mL may still have PCa, and based on the serum level alone, it is not possible to distinguish between high- and low-risk (LR) PCas (3). Due to the lack of specificity with PSA-based screening and harm associated with overtreatment and overdiagnosis, the US Preventive Services Task Force recommended physicians do not routinely perform PSA-based PCa screening (4–6). The major criticism associated with PSA-based screening is “overtreatment.” This may be reduced by improved risk stratification; men with LR and very low-risk (VLR) PCa can be monitored on active surveillance, whereas those with intermediate and high-risk (HR) PCa benefit from treatment (7–11). Treatment can be avoided in almost 70% of men in active surveillance at 15 y of follow-up (12). However, many urologists and patients are reluctant to monitor their cancer on active surveillance due to concerns for delaying treatment or potentially missing treatment of aggressive cancer during a window of cure. Evidence for inadequacy of staging and risk stratification is demonstrated by the increase in Gleason Grade from Gleason 6–7 or higher in 40% of patients treated with radical prostatectomy (RP) (13, 14). Thus, significant discrepancies between prostate needle biopsy and RP specimens may be attributed to diagnostic pitfalls as only 2% of the prostate is sampled with a biopsy (15). Improved staging, which can result in reduction in overtreatment, patient anxiety, and biopsy-related complications, can be achieved by identifying unique molecular signatures capable of discriminating aggressive forms of PCa (16).

Detection of molecular signatures that are indicative of molecular processes related to aggressive forms of PCa may allow biological insight into differentiating aggressive from indolent PCa. MicroRNAs (miRNA, miR) are critical gene regulatory elements that are present in stable forms in serum and have emerged as potential noninvasive biomarkers for cancer diagnosis (17–21). Exosomes may function as delivery vehicles of circulating miRNAs and transport them from primary cancer sites to metastatic sites while also shielding miRNAs from serum nucleases (22, 23). Therefore, serum exosomal miRNAs may potentially serve as noninvasive biomarkers to identify molecular signatures specific to patients with a higher risk of developing aggressive forms of PCa relative to those with indolent PCa. It is important to note that others have identified miRNA signatures and linked them to PCa progression. Circulating miR-141, miR-200c, and miR-375 have been proposed as potential blood markers for the diagnosis of PCa (20, 24, 25). However, the heterogeneity of PCas might not allow intermediate grades of PCa to be distinguished from aggressive forms using these previously identified miRNA signatures. In this study, we sought to determine the miRNA expression of very high-risk (VHR) PCa and validate this expression pattern in men with differing PCa aggressiveness.

Results

Circulating miRNA Profiling Using the Scano-miR Bioassay.

Current methods for miRNA profiling include miRNA fluorophore-based microarray techniques, deep sequencing, quantitative real-time PCR (qRT-PCR), and more recently, techniques based upon spherical nucleic acid (SNA) gold nanoparticle conjugates and the Scano-miR platform (26–29). The Scano-miR bioassay, which does not rely on target enzymatic amplification and is therefore amenable to massive multiplexing to screen a sample for thousands of relatively short miRNA targets (19–25 nucleotides), can detect miRNA biomarkers down to 1 femtomolar concentrations with the capability to distinguish perfect miRNA sequences from those with single nucleotide mismatches (i.e., SNPs) (29). We used the Scano-miR platform to study the exosomal miRNA profiles of serum samples from patients with VHR PCa and compared them with the miRNA profiles from healthy individuals and ones with LR PCa. We obtained a discovery set of 16 serum samples from healthy donors and patients with varying grades of PCa (Tables S1 and S2). In a typical Scano-miR assay, exosomes were isolated from serum samples followed by miRNA extraction and ligation to a universal miRNA cloning oligonucleotide linker (29). The ligated mixtures were hybridized onto miRNA microarrays (miR-array), and then SNA probes were added to the miR-arrays to bind the ligated miRNA species. A gold enhancement solution consisting of HAuCl4 and NH2OH (29, 30) was added to enhance the scattered light signals from the SNA probes. These signals were measured with a Tecan LS Reloaded Scanner and used to extract the miRNA profiles and to determine the miRNA expression levels from each serum sample.

Table S1.

Clinical annotation for the discovery set of n = 16 patients screened for PCa

Scano-miR ID Sample ID Age, y Patient diagnosis Gleason score PSA, ng/mL Stage
Aggressive 7 161010S* 58 PCa 9 N/A II
Aggressive 2 161051S* 60 PCa 8 8 IV
Aggressive 5 16712S* 76 PCa 8 477 Ill
Aggressive 3 16906S* 54 PCa 8 8.53 IV
Aggressive 8 11518552† 66 PCa 9 6.36 N/A
Aggressive 6 11518535† 57 PCa 9 6.8 N/A
Aggressive 4 16847S* 77 PCa 8 45 II
Aggressive 1 11518542† 63 PCa 8 N/A N/A
Indolent 1 11518536† 71 PCa 6 N/A N/A
Indolent 4 11518558† 53 PCa 6 N/A N/A
Indolent 2 11518537† 62 PCa 6 N/A N/A
Indolent 3 11518539† 73 PCa 6 N/A N/A
Normal 3 D 2213S† 63 Normal N/A N/A N/A
Normal 4 D 2214S† 66 Normal N/A N/A N/A
Normal 1 D 2218S† 65 Normal N/A N/A N/A
Normal 2 D 2241S/Ac† 60 Normal N/A N/A N/A
Open in a new tab

The Gleason score is the combined Gleason score obtained through biopsy and histological examination. Clinical tumor stage and cancer staging were based on pathological examination. All samples are serum, male, and Caucasian. N/A, not available.

*

Serum samples were purchased from ProteoGenex, Inc.

†

Serum samples were purchased from ProMedDx, LLC.

Table S2.

Patients screened for PCa categorized based on their risk status

Sample Clinical Gleason score Clinical staging Risk status
Aggressive 1 8 T1cNxMx HR
Aggressive 2 8(5+3) T3bN1M0 VHR
Aggressive 3 8(4+4) T2NxM1 VHR
Aggressive 4 8(3+5) T2NxM0 HR
Aggressive 5 8(4+4) T3NxM0 VHR
Aggressive 6 9 T2cNxMx VHR
Aggressive 7 9(5+4) T2NxM0 VHR
Aggressive 8 9 T2a NxM0 VHR
Indolent 1 6 T1cN0M0 LR
Indolent 2 6 T1cNxM0 LR
Indolent 3 6 T1cNxM0 LR
Indolent 4 6 T1aNxMx LR
Normal 1 N/A N/A Healthy
Normal 2 N/A N/A Healthy
Normal 3 N/A N/A Healthy
Normal 4 N/A N/A Healthy
Open in a new tab

The Gleason score is the combined Gleason score obtained through biopsy and histological examination. Clinical tumor stage and cancer staging were based on pathological examination of the primary tumor. HR, high risk; LR, low risk; N/A, not available; VHR, very high risk. Patients were categorized based on the 2015 NCCN Guidelines for Prostate Cancer (Version 1.2015).

The Scano-miR Molecular Signature Differentiates Clinical Grades of PCa.

The comparison between the serum miRNA expression profiles of patients with a high Gleason score (≥ GS 8, HR and VHR, aggressive) compared with control samples with no PCa and patients with a low Gleason score (GS 6, VLR or LR, indolent) identified five exclusively expressed miRNAs. Circulating miR-200c was the most frequently expressed marker (100%) in patients with a high Gleason score (n = 8) and was below the detection limit of the Scano-miR assay in all other samples (GS 6 and healthy donors, n = 8), whereas miR-219-2-3p, -337-5p, -331-3p, and -409-3p were expressed at different frequencies (75%, 50%, 50%, and 50%, respectively). Importantly, Scano-miR expression analysis identified 58 miRNAs, consisting of 45 experimentally validated miRNAs and 13 predicted miRNAs (Tables S3 and S4), which were coexpressed in all 16 samples. Permutation t tests were performed to identify six differentially expressed miRNAs with significant changes in their expression levels between aggressive and control samples (miR-605, miR-135a*, miR-495, miR-433, miR-371–3p, and miR-106a, with an adjusted P value of 0.0301, 0.0319, 0.0411, 0.0115, 0.0089, and 0.0017, respectively) (Figs. S1 and S2).

Table S3.

Expression data for coexpressed miRNAs in aggressive PCa

miRNA Scano-A1 Scano-A2 Scano-A3 Scano-A4 Scano-A5 Scano-A6 Scano-A7 Scano-A8
let-7b* 20,549 28,421 20,549 20,549 20,549 31,475 25,790 20,549
let-7d* 55,720 29,633 53,004 54,405 40,877 54,498 20,549 47,668
miR-106a 53,004 56,906 45,530 49,859 54,405 46,565 54,958 56,664
miR-106b 37,522 35,793 38,247 32,936 54,498 35,793 56,092 46,565
miR-122 23,263 32,392 24,777 25,790 24,777 38,247 41,639 39,810
miR-135a* 24,777 23,263 29,102 34,772 29,633 27,231 24,777 25,790
miR-144 45,530 47,668 46,179 42,281 43,938 46,179 43,938 35,793
miR-148a 56,287 54,958 54,708 56,287 47,668 54,132 54,498 55,600
miR-15a* 43,938 54,132 44,905 48,375 53,947 47,668 56,906 49,075
miR-17 52,681 49,859 46,565 43,938 29,102 50,673 29,102 29,102
miR-18a 32,392 32,936 27,231 29,633 43,147 20,549 23,263 48,375
miR-18b 31,475 27,231 32,936 28,421 27,231 25,790 42,281 44,468
miR-193a-3p 48,375 53,947 49,075 51,709 56,160 47,081 56,664 45,530
miR-200b 25,790 20,549 30,771 24,777 41,639 23,263 44,905 52,681
miR-214* 47,668 44,905 47,668 47,668 23,263 49,859 40,877 38,247
miR-24-1* 56,515 55,173 56,380 55,988 54,132 56,906 45,530 54,405
miR-24-2* 46,179 53,004 33,473 40,877 55,600 37,522 56,287 24,777
miR-29b-2* 28,421 29,102 25,790 29,102 38,247 33,473 36,417 33,473
miR-302a 32,936 30,771 32,392 35,793 56,380 39,810 56,515 56,380
miR-324-3p 27,231 37,522 23,263 23,263 35,793 29,102 31,475 23,263
miR-337-3p 53,504 36,417 55,720 56,092 45,530 55,720 44,468 47,081
miR-338-5p 56,906 56,287 56,664 56,160 53,504 56,380 48,375 56,092
miR-342-5p 47,081 54,498 40,877 36,417 56,515 43,147 55,988 39,260
miR-371-3p 29,102 33,473 29,633 27,231 34,772 32,392 49,859 29,633
miR-432* 53,947 40,877 54,132 53,004 39,260 55,380 38,247 46,179
miR-433 38,247 54,405 39,260 41,639 55,173 32,936 54,405 56,287
miR-450a 39,810 42,281 36,417 37,522 37,522 39,260 37,522 42,281
miR-455-5p 56,092 55,988 56,092 55,600 56,287 53,947 56,160 43,938
miR-493 46,565 41,639 53,504 46,565 51,709 44,468 49,075 56,160
miR-495 54,958 47,081 55,988 55,380 47,081 54,958 52,681 54,132
miR-502-5p 44,905 55,380 43,938 45,530 56,092 43,938 56,380 56,515
miR-505 40,877 48,375 42,281 44,905 55,988 45,530 54,708 37,522
miR-505* 49,859 56,664 44,468 46,179 55,380 48,375 51,709 56,906
miR-508-3p 36,417 31,475 37,522 33,473 33,473 28,421 32,392 31,475
miR-515-3p 54,132 46,565 53,947 53,504 39,810 56,092 30,771 32,936
miR-519d 54,405 49,075 54,405 54,958 44,468 56,515 43,147 50,673
miR-519e 55,600 39,810 55,380 54,708 28,421 56,160 29,633 30,771
miR-558 41,639 56,380 39,810 44,468 52,681 40,877 47,081 55,380
miR-584 33,473 54,708 51,709 49,075 53,004 49,075 54,132 53,947
miR-595 56,664 50,673 50,673 56,664 48,375 52,681 55,173 40,877
miR-605 55,173 45,530 49,859 50,673 36,417 56,287 39,260 41,639
miR-675 54,708 43,938 54,958 52,681 31,475 55,988 33,473 44,905
miR-871 56,160 56,160 55,600 56,906 54,958 54,405 55,380 53,504
miR-877 50,673 52,681 55,173 55,173 56,664 51,709 55,720 49,859
miR-96* 34,772 25,790 34,772 31,475 44,905 29,633 47,668 55,173
IVGN-novel-miR_3446 35,793 34,772 48,375 47,081 49,075 41,639 50,673 54,498
IVGN-novel-miR_3458 42,281 24,777 35,793 39,810 42,281 36,417 53,004 55,720
IVGN-novel-miR_3513 43,147 39,260 43,147 43,147 46,565 44,905 35,793 43,147
IVGN-novel-miR_3515 30,771 43,147 31,475 30,771 56,906 24,777 55,600 27,231
IVGN-novel-miR_3516 55,380 38,247 54,498 54,498 32,392 53,004 34,772 32,392
IVGN-novel-miR_3517 44,468 44,468 47,081 39,260 30,771 42,281 32,936 36,417
IVGN-novel-miR_3575 55,988 53,504 56,160 56,380 50,673 54,708 46,565 51,709
IVGN-novel-miR_3582 54,498 55,720 56,906 55,720 46,179 55,600 53,504 53,004
IVGN-novel-miR_3645 39,260 46,179 41,639 32,392 32,936 30,771 39,810 34,772
IVGN-novel-miR_3674 51,709 56,515 52,681 54,132 54,708 53,504 27,231 55,988
IVGN-novel-miR_3683 56,380 55,600 56,287 56,515 55,720 56,664 53,947 54,958
IVGN-novel-miR_3696 29,633 51,709 28,421 38,247 25,790 34,772 28,421 28,421
IVGN-novel-miR_3702 49,075 56,092 56,515 53,947 49,859 55,173 46,179 54,708
Open in a new tab

Each column denotes the normalized final intensity value averaged between three probes for each miRNA sequence screened. Enriched serum samples were taken from eight aggressive PCa patients (Gleason score >6) and screened using the Scano-miR platform.

Table S4.

Expression data for coexpressed miRNAs in normal and indolent PCa

miRNA Scano-I5 Scano-I2 Scano-I3 Scano-I4 Scano-N1 Scano-N2 Scano-N3 Scano-N4
let-7b* 25,790 23,263 20,549 20,549 28,421 27,231 25,790 32,936
let-7d* 56,092 55,600 53,504 25,790 48,375 51,709 49,859 55,380
miR-106a 45,530 43,147 49,075 40,877 49,075 48,375 46,565 46,179
miR-106b 38,247 36,417 38,247 52,681 44,468 40,877 42,281 42,281
miR-122 36,417 38,247 34,772 39,810 39,260 28,421 24,777 37,522
miR-135a* 23,263 31,475 29,633 29,102 43,938 31,475 32,392 45,530
miR-144 47,668 44,905 46,179 47,081 40,877 39,260 41,639 38,247
miR-148a 52,681 56,092 53,947 49,859 55,988 47,081 54,132 47,668
miR-15a* 47,081 49,075 48,375 56,160 35,793 53,504 56,515 55,720
miR-17 48,375 43,938 47,668 42,281 45,530 50,673 45,530 39,810
miR-18a 32,392 24,777 27,231 30,771 23,263 25,790 28,421 23,263
miR-18b 28,421 35,793 35,793 23,263 24,777 32,392 32,936 27,231
miR-193a-3p 49,859 52,681 49,859 56,906 38,247 49,859 54,708 50,673
miR-200b 20,549 25,790 25,790 46,179 29,102 23,263 20,549 29,102
miR-214* 50,673 53,504 51,709 38,247 41,639 53,004 47,081 34,772
miR-24-1* 56,515 56,380 56,664 43,147 55,600 54,708 54,958 56,287
miR-24-2* 35,793 37,522 41,639 56,664 36,417 41,639 38,247 39,260
miR-29b-2* 29,102 29,102 32,936 55,173 29,633 29,633 23,263 29,633
miR-302a 32,936 34,772 36,417 55,988 34,772 32,936 36,417 33,473
miR-324-3p 30,771 20,549 23,263 43,938 25,790 29,102 27,231 24,777
miR-337-3p 53,004 53,004 53,004 55,720 54,132 54,498 49,075 51,709
miR-338-5p 55,600 55,720 56,515 44,905 54,958 56,515 56,092 55,600
miR-342-5p 46,179 42,281 47,081 55,600 50,673 46,179 54,498 54,405
miR-371-3p 24,777 28,421 29,102 31,475 20,549 20,549 29,633 20,549
miR-432* 54,132 54,498 54,958 28,421 42,281 56,380 52,681 46,565
miR-433 37,522 39,260 37,522 45,530 37,522 38,247 39,810 31,475
miR-450a 40,877 41,639 39,260 29,633 53,004 45,530 37,522 47,081
miR-455-5p 53,947 56,515 54,708 56,287 56,906 55,988 56,664 56,092
miR-493 43,938 46,179 42,281 56,092 52,681 46,565 44,905 54,498
miR-495 56,664 56,664 56,380 54,405 56,160 55,380 53,947 53,504
miR-502-5p 44,905 47,081 43,147 53,504 54,708 44,905 56,906 44,468
miR-505 49,075 46,565 44,468 54,958 56,092 43,938 43,938 53,004
miR-505* 44,468 44,468 44,905 56,380 55,720 47,668 43,147 55,173
miR-508-3p 31,475 29,633 32,392 24,777 31,475 30,771 31,475 36,417
miR-515-3p 55,988 54,708 56,160 32,392 53,504 56,664 51,709 49,859
miR-519d 56,287 54,958 55,720 33,473 47,668 56,287 53,504 48,375
miR-519e 56,160 54,132 56,092 35,793 47,081 54,958 46,179 43,938
miR-558 46,565 45,530 43,938 36,417 54,405 43,147 40,877 55,988
miR-584 43,147 49,859 45,530 44,468 46,565 56,092 56,160 56,664
miR-595 54,405 50,673 54,405 55,380 56,664 55,173 55,380 53,947
miR-605 56,380 55,380 55,988 56,515 53,947 56,906 47,668 41,639
miR-675 54,708 54,405 56,287 27,231 39,810 56,160 48,375 40,877
miR-871 55,720 56,906 55,600 54,708 56,287 54,132 55,988 54,958
miR-877 53,504 51,709 50,673 51,709 56,380 52,681 55,720 54,132
miR-96* 29,633 32,936 33,473 47,668 30,771 35,793 35,793 35,793
IVGN-novel-miR_3446 42,281 47,668 46,565 50,673 46,179 39,810 56,287 56,380
IVGN-novel-miR_3458 33,473 30,771 30,771 54,498 32,936 34,772 39,260 28,421
IVGN-novel-miR_3513 39,810 40,877 39,810 37,522 33,473 37,522 30,771 32,392
IVGN-novel-miR_3515 39,260 32,392 31,475 53,004 32,392 24,777 34,772 30,771
IVGN-novel-miR_3516 55,380 53,947 55,380 32,936 49,859 55,720 50,673 49,075
IVGN-novel-miR_3517 41,639 48,375 40,877 41,639 44,905 49,075 44,468 44,905
IVGN-novel-miR_3575 54,958 55,988 55,173 54,132 54,498 42,281 53,004 52,681
IVGN-novel-miR_3582 55,173 55,173 54,498 46,565 55,173 44,468 55,600 56,160
IVGN-novel-miR_3645 27,231 33,473 24,777 34,772 27,231 36,417 33,473 25,790
IVGN-novel-miR_3674 51,709 39,810 52,681 49,075 56,515 55,600 55,173 56,906
IVGN-novel-miR_3683 56,906 56,287 56,906 53,947 55,380 53,947 54,405 54,708
IVGN-novel-miR_3696 34,772 27,231 28,421 39,260 43,147 33,473 29,102 43,147
IVGN-novel-miR_3702 54,498 56,160 54,132 48,375 51,709 54,405 56,380 56,515
Open in a new tab

Each column denotes the normalized final intensity value averaged between three probes for each miRNA sequence screened. Four indolent (Gleason score = 6) and four normal enriched serum samples were screened using the Scano-miR platform.

Fig. S1.

Fig. S1.

Open in a new tab

Heat map of all coexpressed miRNAs. Hierarchical clustering was performed on 45 coexpressed miRNAs using the Pearson correlation metric. Included miRNA are expressed to some extent in aggressive (n = 8) and control samples (indolent PCa and healthy individuals, n = 8). Samples from patients with aggressive PCa generally cluster together and have globally up-regulated serum miRNA expression.

Fig. S2.

Fig. S2.

Open in a new tab

Identification of differentially expressed miRNAs using the Scano-miR assay. Boxplots represent the background-subtracted, normalized distributions of six differentially expressed miRNAs. The red bar represents the median, whereas the blue bar represents the interquartile range of distribution (permutation t test; miR-106a, P = 0.0017; miR-371–3p, P = 0.0089; miR-433, P = 0.0115; miR = 605, P = 0.0301; miR-135a*, P = 0.0319; miR-495, P = 0.0411).

Despite the identification of differentially expressed miRNAs, single biomarkers may not be accurate diagnostics for aggressive PCa. For that reason, we calculated the molecular signature score for the differentially expressed miRNAs to distinguish aggressive PCa from control samples using a published mathematical formula described by Zeng et al. (31) The molecular signature analysis revealed that the diagnostic reliability was increased significantly (P = 0.0036) upon combining the differentially expressed miRNAs (Fig. 1A). The molecular signature score was able to detect 50% of the aggressive PCa samples, which suggests that there might be intermediate grades of PCa that were not distinguished using the Gleason sum of the prostatic needle biopsy specimens. To address this, we performed correlation studies to the clinical pathology of PCa, and we also conducted blinded validation studies using patient serum samples.

Fig. 1.

Fig. 1.

Open in a new tab

Molecular signature score of six miRNAs. (A) The molecular signature score was calculated for the six differentially expressed miRNAs using the procedure described in Zeng et al. (31). Distinct ranges of the combined intensity score show that there is little overlap between aggressive and control group expression when using an aggregate score (P = 0.0036). (B) Aggregate of six miRNAs showing a significant correlation to VHR PCa. The combined signature intensity is correlated to the degree of PCa aggressiveness (n = 16). Correlation between miRNA expression and patient risk was analyzed using the Wilcoxon rank-sum test. Correlation P value; P < 0.1, Trend; P < 0.05, *; no correlation, –.

Prostate needle biopsies often undergrade the actual tumor aggressiveness, with up to 20% of patients having more aggressive tumors on subsequent biopsy or prostatectomy (32). Therefore, we investigated the correlation of our identified molecular signature to the clinicopathologic features of PCa following the 2015 National Comprehensive Cancer Network (NCCN) Guidelines for Prostate Cancer (Version 1.2015). To examine such a correlation, prostatic needle biopsy specimens of patients with GS ≥ 8 were grouped into either VHR or HR PCas. Six patients out of eight were identified as having VHR cancer that progressed into locally advanced or metastatic PCa (GS 9, metastasis, and/or clinical stage T3), and the other two patients had HR PCa (GS 8 and clinical stage < T3) (Table S2). Unsupervised hierarchical clustering revealed that these six miRNA markers were able to identify a subclass of four patients out of six, classified as VHR (Fig. S3). We calculated the statistical correlation of the molecular signature and individual miRNAs to the VHR cancer using the Kaplan–Meier and Wilcoxon rank-sum tests (33, 34). The results show a significant correlation of the identified molecular signature to the clinical pathology of these patients (P = 0.041) (Fig. 1B). In contrast, not all differentially expressed miRNAs show high correlations when examined individually (three miRNAs with P ≤ 0.05), which supports the notion that these individual biomarkers, considered alone, are not uniquely indicative of disease states.

Fig. S3.

Fig. S3.

Open in a new tab

Heat map of clustering and clinical association for six differentially expressed miRNAs. Unsupervised hierarchical clustering performed on expression profiles for 16 serum samples reveals that, in general, samples of similar histology are clustered together. Interestingly, a subgroup of four samples identified to be indicative of VHR aggressive PCa cluster together using this molecular signature. Gleason scores are on the range of 9 (black) to 6 (light gray) to healthy (white). Tumor staging is on the range of T3 (black) to T1 (light gray) to healthy (white). Risk status scale is VHR, very high risk; HR, high risk; LR, low risk; or healthy. Patients were categorized based on the 2015 NCCN Guidelines for Prostate Cancer (Version 1.2015).

Validation of the Scano-miR miRNA Molecular Signature.

To validate the reliability of the identified circulating miRNAs as diagnostic biomarkers, we obtained additional clinical serum samples from deidentified patients (VHR PCa, LR PCa, and healthy donors, with sample size n = 9, n = 9, and n = 10, respectively) and performed a blinded evaluation using qRT-PCR. Within this cohort of nine VHR PCas, we included four patients that were diagnosed with LR tumors but were actually undergraded and had more aggressive tumors at RP. The clinical annotation data for the validation set of samples are included in Table S5. We successfully detected five miRNAs using qRT-PCR (miR-200c, miR-605, miR-135a*, miR-433, and miR-106a) that exhibited the same expression profiles as in our Scano-miR profiling studies using the discovery sample set (Fig. 2 and Fig. S4). Four of these miRNAs (miR-605, miR-135a*, miR-433, and miR-106a) were differentially expressed between VHR and undergraded LR PCas relative to LR PCa samples with fold changes >1.5 (Fig. 2 A–D). Circulating miR-433 was differentially expressed in VHR versus LR PCa serum samples (Fig. 2D) but was not detected in normal serum samples (Fig. S4). In addition, miR-200c was only detected in serum samples from patients with VHR PCa (Fig. 2E and Fig. S4). The data suggest that this miRNA biomarker panel can be used as noninvasive biomarkers to distinguish between patients with VHR and LR forms of PCa.

Table S5.

Clinical annotation for the validation set of n = 28 blinded donors screened for PCa

Sample ID Age, y Category Risk status First biopsy total Gleason RP total Gleason Pathological stage
415162* 49 Aggressive VHR 5+4 5+4 T3a
415163* 67 Aggressive VHR 4+5 4+5 T3b
415164* 73 Aggressive VHR 4+5 4+5 T3b
415165* 76 Aggressive VHR 4+4 5+3 T3b
62985* 60 Aggressive VHR 4+5 4+5 T3a
33220* 66 Undergraded LR 3+3 4+4 T3b
415173* 76 Undergraded LR 3+3 4+4 T3b
41145* 69 Undergraded LR 3+3 3+5 T3a
415172* 74 Undergraded LR 3+3 5+4 T3a
36957* 67 Indolent LR 3+3 3+3 T2c
36936* 71 Indolent LR 3+3 3+3 T2c
38123* 65 Indolent LR 3+3 3+3 T2a
415166* 79 Indolent LR 3+3 3+3 T3a
415167* 76 Indolent LR 3+3 3+3 T2a
415168* 66 Indolent LR 3+3 3+3 T2a
415169* 67 Indolent LR 3+3 3+3 T2c
415170* 59 Indolent LR 3+3 3+3 T2b
415171* 77 Indolent LR 3+3 3+3 T2c
BRH980191† 50 Normal Healthy N/A N/A N/A
BRH991717† 59 Normal Healthy N/A N/A N/A
BRH991716† 50 Normal Healthy N/A N/A N/A
BRH991715† 54 Normal Healthy N/A N/A N/A
BRH991714† 59 Normal Healthy N/A N/A N/A
BRH991712† 50 Normal Healthy N/A N/A N/A
BRH991711† 50 Normal Healthy N/A N/A N/A
BRH991710† 65 Normal Healthy N/A N/A N/A
BRH991709† 54 Normal Healthy N/A N/A N/A
BRH991708† 57 Normal Healthy N/A N/A N/A
Open in a new tab

The Gleason sum is the combined Gleason score obtained through histological examination of both prostatic needle biopsy and RP. Clinical tumor stage and cancer staging were based on pathological examination. All samples are serum, male, and Caucasian. HR, high risk; LR, low risk; N/A, not available; VHR, very high risk. Patients were categorized based on the 2015 NCCN Guidelines for Prostate Cancer (Version 1.2015).

*

Serum samples were obtained from the NU Prostate SPORE Serum Repository.

†

Serum samples were purchased from BioreclamationIVT.

Fig. 2.

Fig. 2.

Open in a new tab

qRT-PCR validation of the blinded samples. (A–D) Blinded qRT-PCR analysis of patient serum samples successfully validated four coexpressed miRNAs (miR-605, miR-135a*, miR-433, and miR-106a) (fold change >1.5). (E) Blinded qRT-PCR analysis of a validated, exclusively expressed miRNA: miR-200c.

Fig. S4.

Fig. S4.

Open in a new tab

Ct profiles of validated miRNAs. Five successfully validated miRNAs (miR-200c, miR-605, miR-135a*, miR-433, and miR-106a) using blinded qRT-PCR analysis from patient serum samples showed distinct patterns that correlate with healthy specimens, VHR aggressive PCa, or indolent PCa.

One of the potential applications of the miRNA score is to accurately risk-stratify patients with biopsy-detected PCa. Thus, we sought to compare the area under the curve (AUC) of our miRNA score compared with the gold standard—the biopsy identified Gleason score—for predicting aggressive compared with indolent cancer. Our receiver operating characteristic (ROC) analyses showed that the miRNAs identified by the Scano-miR bioassay exhibit very high diagnostic capabilities in differentiating between VHR aggressive PCa versus controls with an ROC of 1.0, 0.98, 0.98, 0,92, and 0.89 for miR-200c, miR-433, miR-135a*, miR-605, and miR-106a, respectively (Fig. 3 A–E). The prostatic needle biopsy Gleason grading showed the lowest diagnostic capability with an ROC of 0.81 (Fig. 3F).

Fig. 3.

Fig. 3.

Open in a new tab

Specificity and sensitivity analysis. ROC curves were generated to compare the ROC of the Scano-miR miRNAs (A–E) to the Gleason sum from the first prostatic needle biopsy (FB) (F). The miRNAs identified by the Scano-miR bioassay yielded an ROC of at least 0.89 in differentiating between VHR PCa versus control group.

Mapping the validated miRNAs to PCa pathways is important toward understanding their significance in PCa progression. In silico analyses generated a total of 42 candidate pathways (Table S6) from which five common pathways are targeted by the validated miRNA biomarker panel. The identified pathways are primarily involved in cancer progression (including PCa) and phosphatidylinositol-4,5-bisphosphate 3-kinase–Akt serine/threonine kinase (PI3K-Akt) signaling (Table 1). We found the PI3K-Akt signaling pathway to be among the top common candidate pathways, which is a major driver of PCa growth in advanced cancer stages. Additionally, genes that are known to be involved in PCa progression were significant targets of the validated miRNAs (corrected P value threshold of <0.05; Fig. 4). The results suggest that the validated miRNAs target different genes within the same candidate pathways involved in the transition from localized PCa to metastatic PCa.

Table S6.

Enriched KEGG pathway analysis

KEGG pathway P value No. of genes No. of miRNA
TGF-beta signaling pathway, hsa04350 2.55E-11 19 2
Endocytosis, hsa04144 3.28E-09 35 3
Hepatitis B, hsa05161 2.51E-08 24 4
ErbB signaling pathway, hsa04012 1.34E-06 18 4
Colorectal cancer, hsa05210 1.67E-06 14 2
Chronic myeloid leukemia, hsa05220 1.73E-06 16 4
Pathways in cancer, hsa05200 2.94E-06 44 5
Acute myeloid leukemia, hsa05221 5.51E-06 13 3
PI3K-Akt signaling pathway, hsa04151 5.51E-06 44 5
PCa, hsa05215 1.06E-05 17 5
Pancreatic cancer, hsa05212 2.34E-05 15 3
Hepatitis C, hsa05160 4.23E-05 21 4
Neurotrophin signaling pathway, hsa04722 4.59E-05 20 4
Insulin signaling pathway, hsa04910 1.38E-04 21 4
Protein processing in endoplasmic reticulum, hsa04141 6.11E-04 25 3
Dorso-ventral axis formation, hsa04320 1.23E-03 6 3
Regulation of autophagy, hsa04140 1.45E-03 7 2
Axon guidance, hsa04360 1.47E-03 20 2
Gap junction, hsa04540 1.49E-03 12 4
Endometrial cancer, hsa05213 1.93E-03 10 4
Focal adhesion, hsa04510 1.93E-03 26 5
Taurine and hypotaurine metabolism, hsa00430 2.21E-03 3 3
Pantothenate and CoA biosynthesis, hsa00770 2.23E-03 5 2
p53 signaling pathway, hsa04115 2.23E-03 12 3
Maturity onset diabetes of the young, hsa04950 2.77E-03 5 3
Spliceosome, hsa03040 3.45E-03 18 4
MAPK signaling pathway, hsa04010 3.65E-03 31 4
Ubiquitin-mediated proteolysis, hsa04120 5.51E-03 19 3
Wnt signaling pathway, hsa04310 5.53E-03 22 3
Nonsmall cell lung cancer, hsa05223 8.77E-03 9 4
Shigellosis, hsa05131 1.28E-02 10 2
ARVC, hsa05412 1.31E-02 10 2
Glioma, hsa05214 1.36E-02 11 5
beta-Alanine metabolism, hsa00410 1.84E-02 6 2
Circadian rhythm, hsa04710 1.84E-02 6 2
Chagas disease, American trypanosomiasis, hsa05142 1.84E-02 14 3
Thyroid cancer, hsa05216 2.90E-02 6 2
mRNA surveillance pathway, hsa03015 2.96E-02 12 3
Small cell lung cancer, hsa05222 3.19E-02 11 3
Melanoma, hsa05218 3.45E-02 10 4
Valine, leucine, and isoleucine biosynthesis, hsa00290 3.82E-02 1 1
Epithelial cell signaling in Helicobacter pylori infection, hsa05120 4.27E-02 9 2
Open in a new tab

The table shows the biological pathways associated with validated miRNA PCa biomarkers. Common pathways associated with all five miRNAs are shown in bold font.

Table 1.

Common KEGG pathways

KEGG pathway P value No. of genes No. of miRNA
Pathways in cancer, hsa05200 2.94E-06 44 5
PI3K-Akt signaling pathway, hsa04151 5.51E-06 44 5
PCa, hsa05215 1.06E-05 17 5
Focal adhesion, hsa04510 1.93E-03 26 5
Glioma, hsa05214 1.36E-02 11 5
Open in a new tab

The table shows the biological pathways shared with the five validated miRNA PCa biomarkers.

Fig. 4.

Fig. 4.

Open in a new tab

KEGG pathway analysis of the validated miRNAs. Target genes and biological pathways for up-regulated miRNAs (red oval) and down-regulated miRNAs (green oval) were identified using microT-CDS and TarBase to classify the GO category and KEGG pathway enrichment with a corrected P value threshold of <0.05. The yellow squares represent target genes potentially altered by the expression of the validated miRNAs, and blue squares represent genes that are not directly targeted by the validated miRNAs.

Discussion

Risk stratified treatment of PCa is critically dependent on staging through PSA, physical examination, and tissue biopsy. To address the inherent gaps in cancer staging, we successfully applied and validated the Scano-miR profiling platform and ultimately identified a unique panel of miRNA biomarkers associated with different grades of PCa.

The miRNA biomarker panel was discovered and validated by investigation of the serum miRNA profiles from two experimental sample sets. The first set was profiled using the Scano-miR bioassay to identify differentially expressed miRNAs specific to VHR PCa samples that were previously clinically graded based upon Gleason biopsy scoring. A blinded qRT-PCR study was then performed on the second sample set, which served to validate the identified miRNA biomarkers in patient samples with known pathological grading. For example, although individual miRNA biomarkers such as miR-433 and miR-135a* did not fully agree with the clinical grading of PCa, known pathological grading of the blinded qRT-PCR study validated the significant diagnostic capabilities of the identified miRNA biomarkers including circulating miR-433 and miR-135a*. The molecular signature generated from the validated miRNAs enabled us to accurately distinguish between patients with indolent or aggressive forms of PCa at rates higher than typical prostatic needle biopsy Gleason scoring. This miRNA biomarker panel represents a simple tool for the diagnosis of PCa without the need for surgical intervention.

The majority of the identified miRNAs were linked previously to the pathogenesis of PCa either as oncogenes or tumor suppressors. Circulating miR-200c in plasma can be used as a marker to distinguish localized PCa from metastatic castration-resistant PCa (25). miR-106a was significantly dysregulated in PCa, whereas a single nucleotide polymorphism in miR-605 was found to correlate with the biochemical recurrence of PCa (35, 36). However, to our knowledge, circulating miR-433 and miR-135a* have not been linked to PCa previously, and the selected miRNA panel (miR-200c, miR-605, miR-135a*, miR-433, and miR-106a) has not been proposed to have a predictive value for the management of PCa.

Identifying genetic clues to the molecular basis of PCa growth is a major challenge, as the number of mutated genes is often higher than the actual mutations that drive cancer. Our analysis with the selected miRNA panel in the PCa pathway suggested a list of target genes [Phosphatase and Tensin Homolog (PTEN), PI3K, Tumor Protein p53 (TP53), Retinoblastoma 1 (RB1), MDM2, TGFA, NFKB1, CASP9, CDKN1A, E2F1, SOS1, MAPK1, CREB5, TCF7L1, CCND1, BCL2, PDGFD, PDGFRA, GRB2, LEF1, and TCF4]. Although many of these target genes might act as passengers, some of them are known drivers of PCa tumorigenesis. For example, somatic mutations of TP53 and RB1 in PCa are well-established genetic alterations (37). Loss of the tumor suppressor PTEN causes activation of the PI3K-Akt signaling pathway, which is a critical oncogenic pathway in PCa (38). The PI3K-Akt pathway is an important driver of epithelial–mesenchymal transition (EMT) to reduce intercellular adhesion of cancer cells while increasing motility (39). Recent reports suggest a cross-talk between PI3K-AKT and the androgen receptor (AR) pathway in PCa with an inactivated PTEN gene (40), where activated PI3K/AKT causes PCa to become metastatic and hormone-independent. As a result, the validated miRNAs might play an important role in the regulation of aggressive PCa.

In conclusion, we have identified circulating miRNAs that serve as a molecular signature to detect VHR PCa. These biomarkers (miR-200c, miR-605, miR-135a*, miR-433, and miR-106a) showed significant correlation to VHR PCa in clinical samples. This work serves as an initial proof-of-principle study that circulating miRNA biomarkers can identify specific grades of PCa, and further investigations with larger numbers of patient samples are currently underway to validate the utility of using these biomarkers and scoring signature to unambiguously diagnose the disease. Finally, additional work will be required to determine the role of the identified serum miRNAs in PCa development and tumorigenesis.

Materials and Methods

Clinical Samples.

The discovery set of serum samples were purchased from two vendors, as specified in Table S1 (ProteoGenex, Inc. and ProMedDx, LLC). The validation set of serum samples with different grades of PCas and negative for metastasis were obtained from the Northwestern University (NU) Prostate Specialized Programs of Research Excellence (SPORE) serum repository following the institutional review protocol, whereas healthy serum samples were purchased from BioreclamationIVT. Serum samples were collected from donors with matched ethnicity and sex (Caucasian and male). Samples were stored at –80 °C upon arrival and thawed on ice before use.

Isolation of Serum Exosomal RNA.

Exosomes were isolated from the discovery set of serum samples using ExoQuick Exosome Precipitation Solution (System Biosciences, part #EXOQ5A-1) following the manufacturer’s protocol. In short, serum samples were centrifuged to remove cell debris (3,000 rpm, 15 min) (Eppendorf Centrifuge 5424R, rotor FA-45-24-11). We added 1 mL serum supernatant to 252 μL ExoQuick exosome precipitation solution and mixed and incubated it at 4 °C for 30 min. Following incubation, the mixture was recentrifuged and the exosome pellet was collected. RNA isolation from the exosome pellet was performed using mirVana PARIS miRNA isolation kit (Ambion, part #AM1556) following the manufacturer’s protocol by suspending the exosome pellet in 300 μL of cell disruption buffer solution followed by adding 300 μL of 2× denaturing solution and was allowed to incubate on ice for 5 min. Next, 600 μL of acid-phenol:chloroform was added to the mixture, vortexed, and centrifuged to collect 300 μL of the aqueous phase (10,000 rpm, 5 min). The aqueous phase was mixed with 100% ethanol at a 1:1.25 volume ratio and then column filtered, followed by RNA elution with 100 μL of elution buffer. Total RNA from the filtrate was precipitated by adding 0.3 M NaCl, 20 μg glycogen, and 1 volume of isopropanol and allowed to incubate at –80 °C for 12 h. The mixture was centrifuged to collect the pellet (15,000 rpm, 30 min, 4 °C), followed by one wash with 1 mL of 70% (vol/vol) ethanol. The pellet was washed once with 1 mL 70% ethanol, air-dried, and suspended in 10 μL RNase-free water. Total RNA was stored at –80 °C until profiling studies using the Scano-miR bioassay.

Synthesis of the Universal SNA Nanoconjugates.

SNAs were synthesized by chemisorbing 4 μM of a propylthiol-modified ssDNA recognition sequence [5′-propylthiol-(A)10-TCCTTGGTGCCCGAGTG-3′] complementary to miRNA Cloning Linker II (IDT) onto 10 nM of citrate-stabilized gold nanoparticles (13 nm in diameter) following a published protocol (29). The mixture was allowed to incubate for 1 h at room temperature, followed by a salt aging process consisting of 0.01% SDS, 10 mM phosphate buffer (pH 7.4), and 0.1 M sodium chloride (NaCl), for an additional 1 h at room temperature. Two additional aliquots of 0.1 M NaCl were added, and the mixture was allowed to incubate for 1 h between each addition and subsequently incubated overnight (room temperature, shaking at 130 rpm). SNAs were purified through three successive rounds of centrifugation (16,000 × g for 20 min), supernatant removal, and resuspension in PBS (137 mM NaCl, 10 mM phosphate, 2.7 mM KCl, pH 7.4). All experiments were carried out with RNase-free materials.

miRNA Profiling Using the Scano-miR Platform.

Isolated serum miRNAs were added to a ligation mixture (200 U Truncated T4 RNA Ligase 2, 900 ng miRNA cloning linker II, 12% PEG 8000, and 1× T4 RNL2 buffer) from New England Biolabs following the manufacturer’s protocol and allowed to incubate for 3 h at 37 °C. The ligation mixture was suspended in 400 μL RNase-free 2× SSC hybridization buffer (0.3 M NaCl, 0.03 M sodium citrate, pH 7.0) and hybridized onto NCode Human miRNA microarray V3 (Invitrogen) for 12 h at 52 °C (29). Following the incubation, the miR-arrays were washed to remove unbound miRNAs using prewarmed 2× SSC (52 °C), 2× SSC, PBS (137 mM NaCl, 10 mM phosphate, 2.7 mM KCl, pH 7.4), and nanopure water. We hybridized 1 nM of the synthesized SNAs suspended in 400 μL 2× SSC onto the miR-array at 56 °C for 1 h. The washing steps were repeated to remove unreacted SNAs. All experiments were performed using RNase-free materials. Finally, the light scattering of the gold nanoparticles was increased using three rounds of gold enhancing solution [a freshly mixed 1:1 (vol/vol) solution of 1 mM HAuCl4 and 10 mM NH2OH] (5 min each, at room temperature). The miR-array was imaged with a LS Reloaded scanner (Tecan).

Data Analysis and miRNA Clustering.

Raw Scano-miR expression data were extracted from 4,608 probes using GenePix Pro-6 software (Molecular Devices). Expression values below background threshold as well as abnormal probe shape index were filtered from downstream data analysis. An average of three probe replicates per miRNA target were used for expression analysis. In total, 705 human miRNAs were screened for each sample. The identities and frequencies of the expression profiles were calculated for five exclusively expressed miRNAs that were detected solely in aggressive serum samples, where frequency denotes the number of times the miRNA was detected in the serum sample divided by the number of aggressive samples. We filtered 583 miRNAs that were not expressed in all 16 samples from further expression analysis. Quantile normalization was performed on 16 samples with 167 coexpressed features. Heat maps were clustered using Pearson correlation as a distance metric and visualized using MATLAB.

Molecular Signature and Clinical Analysis.

A permutation t test was performed to obtain six differentially expressed miRNAs between aggressive and control samples. Permutation t tests were used to estimate the null distribution of the t test statistic (41). P values were corrected for false discovery rate (FDR) using the Storey–Tibshirani procedure (41). A molecular signature score was calculated based on the differences between the expression levels of down-regulated and up-regulated miRNAs using a published formula (31). The Kaplan–Meier and Wilcoxon rank-sum tests were used to assess the correlation of the signature score and individual miRNA to HR patient profiles.

qRT-PCR Validation of the Blinded Samples.

The serum exosomes were isolated from the validation set of serum samples using the previously described protocol and suspended in a denaturing lysis buffer (Ambion, part #AM1560). We spiked 100 pM of synthetic cel-miR-40-3p (Applied Biosystems, part #MC10631) into denatured exosomes. RNA isolation from the denatured exosome was performed using mirVana miRNA isolation kit (Ambion, part #AM1560) following the manufacturer’s protocol. Using TaqMan RT kit (part #4366597), TaqMan hsa-miR-200c, hsa-miR-106a, hsa-miR-605, hsa-miR-371-3p, hsa-miR-135a*, hsa-miR-433, hsa-miR-495, and cel-miR-40 RT primers, 1 ng (5 μL) of total RNA from each sample was reverse-transcribed in 15 μL reaction volumes following the manufacturer’s protocol (Applied Biosystems, TaqMan MicroRNA Assays PN 4364031E). qRT-PCR reactions were conducted in 96-well plates with 1.33 μL of RT product with TaqMan PCR master mix (part #4364343) and TaqMan probes for each miRNA in a total volume of 20 μL. An ABI Prism Model 7900 HT instrument was used to perform the qRT-PCR reactions with data analyzed using the comparative Ct method with cel-miR-40-3p used as an exogenous control. Known concentrations of cel-miR-40-3p were used to generate qRT-PCR standard curve. For statistical evaluation of the qRT-PCR validation test, the Mann–Whitney t test was used, where a P value less than 0.05 and a cutoff of >1.5-fold change was considered statistically significant (GraphPad Prism 6).

Sensitivity and Specificity Calculation.

The tradeoff between sensitivity (true positive rate) and false positive rate (1-specificity) using the Gleason scoring sum of the first prostatic needle biopsy (FB), individual miRNA biomarkers, and molecular signature score for predicting VHR PCa was assessed using the area under the ROC curve.

Target Genes and Pathway Analysis of the Validated miRNAs.

In silico analysis was performed to identify miRNA target genes and molecular pathways potentially altered by the expression of single or multiple miRNAs. Putative target genes of miRNA were determined using the homology search algorithm microT-CDS and a database of published, experimentally validated miRNA-gene interactions, TarBase (42, 43). For microT-CDS, a microT prediction threshold of >0.8 was set. DIANA-miRPath was used to perform functional annotation clustering and pathway enrichment analysis of multiple miRNA target genes (44). Two-sided Fisher’s exact test and the X2 test were used to classify the Gene Ontology (GO) category and KEGG pathway enrichment, and the FDR was calculated to correct P values. A corrected P value threshold of <0.05 was used to select significant GO categories and KEGG pathways.

Acknowledgments

The research was supported by the Prostate Cancer Foundation and the Center for Cancer Nanotechnology Excellence (CCNE) initiative of the National Institutes of Health (NIH) under Awards U54 CA151880 and U54 CA199091. A.H.A. acknowledges the King Abdullah Scholarships Program funded by the Ministry of Higher Education of Saudi Arabia for doctoral scholarship support. C.S.T. and J.J.M. acknowledge the Northwestern Memorial Foundation Dixon Translational Research Grant Initiative Award SP0010333. J.J.M is supported by National Cancer Institute (NCI) SPORE Grant P50 CA180995. The contributions of G.F. are partially supported by NCI Cancer Center Support Grant NCI CA060553 and National Center for Advancing Translational Sciences Grant 8UL1TR000150. J.J.W. was supported in part by National Science Foundation (NSF) Grants DGE-1007911 and NIH T32 GM008449. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect those of the sponsors.

Footnotes

The authors declare no conflict of interest.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1611596113/-/DCSupplemental.

References

  • 1.Jemal A, et al. Cancer statistics, 2007. CA Cancer J Clin. 2007;57(1):43–66. doi: 10.3322/canjclin.57.1.43. [DOI] [PubMed] [Google Scholar]
  • 2.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin. 2015;65(1):5–29. doi: 10.3322/caac.21254. [DOI] [PubMed] [Google Scholar]
  • 3.Heijnsdijk EA, et al. Overdetection, overtreatment and costs in prostate-specific antigen screening for prostate cancer. Br J Cancer. 2009;101(11):1833–1838. doi: 10.1038/sj.bjc.6605422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Grubb RL, 3rd, et al. PLCO Project Team Serum prostate-specific antigen hemodilution among obese men undergoing screening in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Cancer Epidemiol Biomarkers Prev. 2009;18(3):748–751. doi: 10.1158/1055-9965.EPI-08-0938. [DOI] [PubMed] [Google Scholar]
  • 5.Andriole GL, et al. PLCO Project Team Mortality results from a randomized prostate-cancer screening trial. N Engl J Med. 2009;360(13):1310–1319. doi: 10.1056/NEJMoa0810696. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Crawford ED, Andriole GL, Marberger M, Rittmaster RS. Reduction in the risk of prostate cancer: Future directions after the Prostate Cancer Prevention Trial. Urology. 2010;75(3):502–509. doi: 10.1016/j.urology.2009.05.099. [DOI] [PubMed] [Google Scholar]
  • 7.Schröder FH, et al. ERSPC Investigators Screening and prostate-cancer mortality in a randomized European study. N Engl J Med. 2009;360(13):1320–1328. doi: 10.1056/NEJMoa0810084. [DOI] [PubMed] [Google Scholar]
  • 8.Epstein JI. An update of the Gleason grading system. J Urol. 2010;183(2):433–440. doi: 10.1016/j.juro.2009.10.046. [DOI] [PubMed] [Google Scholar]
  • 9.Hugosson J, et al. Mortality results from the Göteborg randomised population-based prostate-cancer screening trial. Lancet Oncol. 2010;11(8):725–732. doi: 10.1016/S1470-2045(10)70146-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Schröder FH, et al. ERSPC Investigators Prostate-cancer mortality at 11 years of follow-up. N Engl J Med. 2012;366(11):981–990. doi: 10.1056/NEJMoa1113135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Conti SL, et al. Pathological outcomes of candidates for active surveillance of prostate cancer. J Urol. 2009;181(4):1628–1633, discussion 1633–1634. doi: 10.1016/j.juro.2008.11.107. [DOI] [PubMed] [Google Scholar]
  • 12.Klotz L, et al. Long-term follow-up of a large active surveillance cohort of patients with prostate cancer. J Clin Oncol. 2015;33(3):272–277. doi: 10.1200/JCO.2014.55.1192. [DOI] [PubMed] [Google Scholar]
  • 13.Bostwick DG, Myers RP, Oesterling JE. Staging of prostate cancer. Semin Surg Oncol. 1994;10(1):60–72. doi: 10.1002/ssu.2980100110. [DOI] [PubMed] [Google Scholar]
  • 14.Isariyawongse BK, et al. Significant discrepancies between diagnostic and pathologic Gleason sums in prostate cancer: The predictive role of age and prostate-specific antigen. Urology. 2008;72(4):882–886. doi: 10.1016/j.urology.2008.02.021. [DOI] [PubMed] [Google Scholar]
  • 15.Chun FK, et al. Optimizing performance and interpretation of prostate biopsy: A critical analysis of the literature. Eur Urol. 2010;58(6):851–864. doi: 10.1016/j.eururo.2010.08.041. [DOI] [PubMed] [Google Scholar]
  • 16.Barbieri CE, et al. The mutational landscape of prostate cancer. Eur Urol. 2013;64(4):567–576. doi: 10.1016/j.eururo.2013.05.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Lee RC, Feinbaum RL, Ambros V. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell. 1993;75(5):843–854. doi: 10.1016/0092-8674(93)90529-y. [DOI] [PubMed] [Google Scholar]
  • 18.Lim LP, et al. Microarray analysis shows that some microRNAs downregulate large numbers of target mRNAs. Nature. 2005;433(7027):769–773. doi: 10.1038/nature03315. [DOI] [PubMed] [Google Scholar]
  • 19.Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005;120(1):15–20. doi: 10.1016/j.cell.2004.12.035. [DOI] [PubMed] [Google Scholar]
  • 20.Mitchell PS, et al. Circulating microRNAs as stable blood-based markers for cancer detection. Proc Natl Acad Sci USA. 2008;105(30):10513–10518. doi: 10.1073/pnas.0804549105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Selth LA, et al. Discovery of circulating microRNAs associated with human prostate cancer using a mouse model of disease. Int J Cancer. 2012;131(3):652–661. doi: 10.1002/ijc.26405. [DOI] [PubMed] [Google Scholar]
  • 22.Valadi H, et al. Exosome-mediated transfer of mRNAs and microRNAs is a novel mechanism of genetic exchange between cells. Nat Cell Biol. 2007;9(6):654–659. doi: 10.1038/ncb1596. [DOI] [PubMed] [Google Scholar]
  • 23.Alhasan AH, Patel PC, Choi CHJ, Mirkin CA. Exosome encased spherical nucleic acid gold nanoparticle conjugates as potent microRNA regulation agents. Small. 2014;10(1):186–192. doi: 10.1002/smll.201302143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Brase JC, et al. Circulating miRNAs are correlated with tumor progression in prostate cancer. Int J Cancer. 2011;128(3):608–616. doi: 10.1002/ijc.25376. [DOI] [PubMed] [Google Scholar]
  • 25.Watahiki A, et al. Plasma miRNAs as biomarkers to identify patients with castration-resistant metastatic prostate cancer. Int J Mol Sci. 2013;14(4):7757–7770. doi: 10.3390/ijms14047757. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Grasedieck S, et al. Circulating microRNAs in hematological diseases: Principles, challenges, and perspectives. Blood. 2013;121(25):4977–4984. doi: 10.1182/blood-2013-01-480079. [DOI] [PubMed] [Google Scholar]
  • 27.Mirkin CA, Letsinger RL, Mucic RC, Storhoff JJ. A DNA-based method for rationally assembling nanoparticles into macroscopic materials. Nature. 1996;382(6592):607–609. doi: 10.1038/382607a0. [DOI] [PubMed] [Google Scholar]
  • 28.Taton TA, Mirkin CA, Letsinger RL. Scanometric DNA array detection with nanoparticle probes. Science. 2000;289(5485):1757–1760. doi: 10.1126/science.289.5485.1757. [DOI] [PubMed] [Google Scholar]
  • 29.Alhasan AH, et al. Scanometric microRNA array profiling of prostate cancer markers using spherical nucleic acid-gold nanoparticle conjugates. Anal Chem. 2012;84(9):4153–4160. doi: 10.1021/ac3004055. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Kim D, Daniel WL, Mirkin CA. Microarray-based multiplexed scanometric immunoassay for protein cancer markers using gold nanoparticle probes. Anal Chem. 2009;81(21):9183–9187. doi: 10.1021/ac9018389. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Zeng X, et al. Circulating miR-17, miR-20a, miR-29c, and miR-223 combined as non-invasive biomarkers in nasopharyngeal carcinoma. PLoS One. 2012;7(10):e46367. doi: 10.1371/journal.pone.0046367. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Dijkstra S, Hamid AR, Leyten GH, Schalken JA. Personalized management in low-risk prostate cancer: The role of biomarkers. Prostate Cancer. 2012;2012(327104):327104. doi: 10.1155/2012/327104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Amer Statist Assn. 1958;53(282):457–481. [Google Scholar]
  • 34.Mann HB, Whitney DR. On a test of whether one of two random variables is stochastically larger than the other. Ann Math Stat. 1947;18(1):50–60. [Google Scholar]
  • 35.Volinia S, et al. A microRNA expression signature of human solid tumors defines cancer gene targets. Proc Natl Acad Sci USA. 2006;103(7):2257–2261. doi: 10.1073/pnas.0510565103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Huang SP, et al. Genetic variants in microRNAs and microRNA target sites predict biochemical recurrence after radical prostatectomy in localized prostate cancer. Int J Cancer. 2014;135(11):2661–2667. doi: 10.1002/ijc.28904. [DOI] [PubMed] [Google Scholar]
  • 37.Sellers WR, Sawyers CA. Somatic genetics of prostate cancer: Oncogenes and tumor suppressors. In: Kantoff PW, editor. Prostate Cancer: Principles and Practice. 1st Ed. Lippincott Williams & Wilkins; Philadelphia, PA: 2002. pp. 53–76. [Google Scholar]
  • 38.Majumder PK, Sellers WR. Akt-regulated pathways in prostate cancer. Oncogene. 2005;24(50):7465–7474. doi: 10.1038/sj.onc.1209096. [DOI] [PubMed] [Google Scholar]
  • 39.Larue L, Bellacosa A. Epithelial-mesenchymal transition in development and cancer: Role of phosphatidylinositol 3′ kinase/AKT pathways. Oncogene. 2005;24(50):7443–7454. doi: 10.1038/sj.onc.1209091. [DOI] [PubMed] [Google Scholar]
  • 40.Marques RB, et al. High efficacy of combination therapy using PI3K/AKT inhibitors with androgen deprivation in prostate cancer preclinical models. Eur Urol. 2015;67(6):1177–1185. doi: 10.1016/j.eururo.2014.08.053. [DOI] [PubMed] [Google Scholar]
  • 41.Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proc Natl Acad Sci USA. 2003;100(16):9440–9445. doi: 10.1073/pnas.1530509100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Reczko M, Maragkakis M, Alexiou P, Grosse I, Hatzigeorgiou AG. Functional microRNA targets in protein coding sequences. Bioinformatics. 2012;28(6):771–776. doi: 10.1093/bioinformatics/bts043. [DOI] [PubMed] [Google Scholar]
  • 43.Vlachos IS, et al. DIANA-TarBase v7.0: Indexing more than half a million experimentally supported miRNA:mRNA interactions. Nucleic Acids Res. 2015;43(Database Issue):D153–D159. doi: 10.1093/nar/gku1215. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Vlachos IS, et al. DIANA Mirpath V.2.0: Investigating the combinatorial effect of microRNAs in pathways. Nucleic Acids Res. 2012;40(Web Server Issue):W498–W504. doi: 10.1093/nar/gks494. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES