MicroRNAs are key regulators of gene expression and play critical roles in both normal physiology and pathology. Recent research has demonstrated that these molecules are present in body fluids, such as serum, plasma, and urine, and can be readily measured using a variety of techniques. More importantly, emerging evidence suggests that circulating or urine miRNAs are useful indicators of disease. Here, we consider the potential utility of such miRNAs as noninvasive biomarkers of prostate cancer, a disease that would benefit substantially from novel diagnostic and prognostic tools. The studies aimed at identifying diagnostic, prognostic, and/or predictive miRNAs for prostate cancer are summarised and reviewed. Finally, practical considerations that will influence the translation of this recent research into clinical implementation are discussed.
Prostate cancer is the second most common solid tumour in men worldwide and, despite significant advances in early diagnosis and management, it remains a leading cause of cancer-related death in men [
Prostate cancer is characterised by distinctly unpredictable outcomes from latent, slow growing tumours to aggressive, rapidly lethal tumours. Although much effort has been put into finding biomarkers that would improve diagnosis, the pertinent clinical issue is the detection of aggressive forms of the disease at an early, curable stage. A significant proportion of cases follow an indolent course and may not require curative treatment. In fact, up to 40% of elderly men will harbour cancer within their prostates at autopsy [
Considering these issues, biomarkers that could improve diagnostic accuracy and better discriminate indolent from aggressive prostate cancers at an early stage would revolutionise the clinical management of this important disease. Moreover, identifying predictive biomarkers for the multitude of new treatment strategies being developed for metastatic prostate cancer [
MicroRNAs (miRNAs) are ~22 nucleotide-long, single-stranded, noncoding RNAs that were first reported in the nematode
It is currently estimated that the human genome encodes over 1800 distinct miRNAs (miRBase 19; [
Given their physiological importance, it is not surprising that miRNAs also play important roles in the genesis and progression of cancer. This concept was first demonstrated by Calin and colleagues, who found that a genomic region at 13q14 containing two miRNAs (miR-15a and miR-16-1) is frequently deleted in leukemia [
Mechanisms by which miRNA function is altered in cancer include deletion/amplification of miRNA genes, modulation of miRNA gene expression through epigenetic mechanisms or dysregulation of transcription factors, and mutation of miRNA loci or their target sequences [
The realisation that miRNAs are deregulated in human cancers has generated considerable interest with regard to their potential as biomarkers. miRNAs have a number of desirable characteristics for such an application. Perhaps most importantly, miRNA expression profiles are often tissue, developmental, and disease specific. For example, early work demonstrated that miRNA expression signatures accurately distinguished between different tumour types and could accurately identify cancers of histologically uncertain origin [
Recent research has shown that miRNAs possess one additional feature of an ideal biomarker, namely, an ability to be sampled noninvasively. In 2008, a number of groups reported the presence of circulating miRNAs in cell-free fractions of blood (i.e., serum and plasma) and presented evidence suggesting that a subset of these molecules could be useful indicators of disease [
The potential of miRNAs derived from body fluids as noninvasive biomarkers for different tumor entities has been discussed extensively in recent review articles (see, e.g., [
Studies investigating the potential of circulating miRNAs as biomarkers of prostate cancer.
Body fluid | Sample size | Methodology | Key findings | Reference |
---|---|---|---|---|
Plasma | 25 patients (metastatic PCa), 25 healthy controls | qRT-PCR |
miR-141 levels could differentiate metastatic PCa patients from healthy subjects | Mitchell et al., 2008 [ |
| ||||
Serum | 6 patients (stages 2–4 PCa), 8 healthy controls | Microarray (custom) (547 miRNAs) | 15 miRNAs were elevated in PCa patients. However, serum miRNAs could not distinguish between different cancer types | Lodes et al., 2009 [ |
| ||||
Serum | 56 patients (20 localized PCa, 20 androgen-dependent PCa, 10 CRPC2), 6 BPH3 controls | qRT-PCR |
miR-21 was elevated in CRPC patients compared to BPH and associated with resistance to docetaxel in CRPC patients | Zhang et al., 2010 [ |
| ||||
Serum | 29 patients (9 low risk, 11 intermediate risk, and 9 high risk)1, 9 healthy controls | qRT-PCR |
10 miRNAs were altered in PCa patients compared to healthy controls. 7 miRNAs were correlated with different risk groups. | Moltzahn et al., 2011 [ |
| ||||
Serum | Profiling: 7 high grade, 14 low grade patients. Validation: 116 patients (various grades) | qRT-PCR |
miR-141, miR-200b, and miR-375 were elevated in serum from high-grade patients and correlated with clinicopathological parameters | Brase et al., 2011 [ |
| ||||
Plasma | 21 patients (metastatic PCa) | qRT-PCR |
miR-141 levels were associated with clinical progression and positively correlated with prostate specific antigen | Gonzales et al., 2011 [ |
| ||||
Plasma | 51 patients (18 localized PCa, 8 local advanced, and 25 metastatic), 20 healthy controls | qRT-PCR (miR-21, miR-141, and miR-221) | miR-21 and miR-221 levels were elevated in PCa patients compared to healthy controls. miR-21, miR-141, and miR-221 levels were higher in metastatic compared to localised tumours |
Agaoglu et al., 2011 [ |
| ||||
Serum | 45 patients (37 localized PCa, 8 metastatic), 18 BPH controls, and 20 healthy controls | qRT-PCR (5 miRNAs) | miR-26a, miR-195, and let-7i levels were elevated in PCa compared to BPH samples | Mahn et al., 2011 [ |
| ||||
Serum | Profiling: 14 TRAMP mice, 14 healthy controls. Validation: 25 patients (metastatic CRPC), 25 healthy controls | Microarray (Affymetrix; 609 murine miRNAs), qRT-PCR (10 human miRNAs) | miR-141, miR-298, miR-346, and miR-375 levels were elevated in metastatic CRPC compared to healthy controls. Expression of miR-375 in primary tumours was associated with biochemical relapse |
Selth et al., 2012[ |
| ||||
Plasma, serum, and urine | Profiling: 78 patients (various grades, 15 with diagnosed metastases), 28 healthy controls. Validation: 119 patients (47 recurrent after RP4, 72 nonrecurrent) | qRT-PCR |
12 circulating miRNAs were at altered levels in PCa patients compared to healthy controls. 16 circulating miRNAs were at altered levels in metastatic versus localised PCa (including miR-141 and miR-375). Urinary levels of miR-107 and miR-574-3p exhibited significant diagnostic value | Bryant et al., 2012 [ |
| ||||
Serum | 84 patients (28 low risk localised disease, 30 high risk localised disease, and 26 metastatic CRPC | qRT-PCR |
miR-375, miR-141, miR-378, and miR-409-3p were at altered levels in metastatic CRPC compared localised cancer | Nguyen et al., 2013 [ |
| ||||
Plasma | Profiling: 25 patients (localised and metastatic PCa), 17 BHP controls. Validation: 80 patients (localised and metastatic PCa), 44 BHP controls, and 54 healthy controls | Microarray (Illumina; 1146 miRNAs), qRT-PCR (8 miRNAs) |
5 miRNAs with significant diagnostic value were identified (let-7c, let-7e, miR-30c, miR-622, and miR-1285) | Chen et al., 2012 [ |
| ||||
Plasma | 23 patients (15 androgen dependent PCa, 8 CRPC), 20 healthy controls | qRT-PCR |
miR-221 was elevated in PCa patients compared to healthy controls, and higher in androgen-dependent PCa compared to CRPC | Zheng et al., 2012 [ |
| ||||
Plasma | 82 patients (various risk scores1,5) | qRT-PCR (5 miRNAs) | miR-20a, miR-21, miR-145, and miR-221 were associated with tumour risk scores1,5 | Shen et al., 2012 [ |
| ||||
Serum | 72 patients (24 localised prostate cancer, 24 bladder cancer, and 24 renal cell carcinoma), 48 noncancer controls | qRT-PCR (8 putative reference small RNAs and miR-21) |
|
Sanders et al., 2012 [ |
2Castration-resistant prostate cancer
3Benign prostatic hyperplasia
4Radical prostatectomy
5D’Amico score
The Tewari laboratory was the first to demonstrate an association between a circulating miRNA and prostate cancer [
A number of studies have assessed the potential of circulating miRNAs to diagnose early-stage, localised prostate cancer. Moltzahn and colleagues compared circulating miRNA profiles in men with early-stage prostate cancer and healthy men [
Other studies have focussed on identifying miRNAs associated with metastatic disease that could be applied as prognostic markers at the time of diagnosis or to detect recurrence following primary treatment. Our group utilized the TRansgenic Adenocarcinoma of Mouse Prostate (TRAMP) model to discover circulating miRNAs that demarcated mice with advanced disease from healthy mice [
The utility of circulating miRNAs as predictors of treatment response is a particularly exciting concept. Although the application of circulating miRNAs for this purpose is in its infancy, two recent studies have provided evidence for their potential in prostate cancer. Zhang and colleagues measured miR-21 levels in patients with localised and metastatic prostate cancer and found that this miRNA was significantly higher in CRPC patients who exhibited resistance to the chemotherapeutic docetaxel [
For physiological and anatomical reasons, urine may represent a valuable source of miRNA biomarkers for urological cancers. Bryant and colleagues were the first to test this concept [
Collectively, the studies described above suggest that circulating miRNAs may assist in the diagnosis, prognosis, and prediction of prostate cancer. Unfortunately, there is little agreement between most of these studies: some of the reasons that may underlie these conflicting results are discussed below. One positive finding is the robust association of miR-141 and miR-375 with metastatic disease [
Many factors are likely to impact on our ability to identify
Pipeline for developing circulating and urine miRNAs as biomarkers of disease, with important considerations shown. Study design is critical and will influence all aspects of the methodology. Clinical evaluation and translation, including clinical trials, commercialisation, and approval, have been discussed elsewhere (e.g., see [
Robust, standardised methodology for sampling the biological material is critically important. For example, contamination of blood fluids with intact or lysed (i.e., haemolysis) blood cells during phlebotomy and sample processing can have a profound effect on the resultant miRNA profile [
Similarly, although miRNA profiling from urine is in its infancy, it is reasonable to suggest that the sampling strategy will have a significant impact on the measurable miRNA milieu. Urine samples should ideally be taken as first pass samples immediately after a DRE to enrich urine sample for prostate cells. The commercially available urine-based PCA3 test for prostate cancer is normally performed after a modified DRE (3 strokes per lobe). A study aimed at analysing performance characteristics of the PCA3 test found that, in the absence of a DRE, an unacceptably low number of patients (75.9% versus 96.7% following DRE) had sufficient prostate cells in their urine for robust measurement [
Many different protocols for isolating miRNAs from serum/plasma and urine have been developed. In general, these protocols comprise guanidinium-phenol (Trizol, Qiazol, etc.) extraction of the sample followed by purification of miRNAs using either alcohol-mediated precipitation or column-based methods [
Measuring specific miRNAs or profiling the complete miRNA population is generally achieved using either qRT-PCR, microarrays, or next-generation sequencing (NGS). Of these, by far the most commonly employed is qRT-PCR, probably because of its increased sensitivity and accuracy. Microarrays and NGS are less sensitive but can profile many more miRNAs. Moreover, NGS has the ability to identify previously unknown miRNAs that would not be amplified by qRT-PCR or anneal to microarray chips. Given that the miRNAome is likely to expand further and the emerging notion that miRNA 5′- and 3′-end structural variants, termed isomirs, are commonly expressed and have been linked to cancer [
Each of the three aforementioned methodologies is associated with a number of unresolved issues. Arguably, the most important of these is how to best normalise miRNA measurements to account for biological and technical variability. Quantitation of small RNAs extracted from 50–400
Two other issues in the profiling/data processing phase must be taken into account. First, it is important to use false discovery rate (FDR) correction when profiling large numbers of miRNAs with any of the methods described above. Second, it is strongly advised that differentially expressed miRNAs identified by microarrays or NGS are validated by qRT-PCR, a more sensitive and accurate technique. Both of these factors are likely to reduce false positive and other erroneous discoveries and result in more robust disease-associated signatures.
Finally, it is worth highlighting that urine miRNA concentrations can differ significantly based on the hydration status of patients. Whilst the gold standard to account for such differences is to measure 24-hour urine volumes, a more feasible method may be to normalise expression data to urinary osmolarity or specific gravity.
Even if the experimental workflow (comprising collection of the biomaterial, extraction, miRNA profiling, and data processing) is robust, the experiment can be impaired by a poor study design. This factor is probably a major reason why very few markers can be validated in further studies. A number of factors need to be considered with the complete biomarker development pipeline firmly in mind [
While there is genuine potential for circulating and urine miRNAs in diagnostic, prognostic, and predictive applications, clinical implementation of a noninvasive miRNA test for prostate cancer is still a distant goal. The studies that have been conducted thus far are heterogeneous in terms of objectives and methodology, which have often yielded conflicting data and outcomes. Improving the consistency and standardisation of these factors is of critical importance. Moreover, cohorts with long clinical followup to validate some of the promising findings, such as the association between miR-141 and miR-375 and metastasis, are yet to be analysed. Despite these challenges, and in light of the fact that circulating miRNAs were discovered just 4 years ago, we believe that the outlook in this field is bright.
The authors would like to acknowledge the following funding sources: the Prostate Cancer Foundation of Australia (to L. A. Selth, ID: YI0810) and Cancer Australia (ID: 1012337). L. A. Selth is a Young Investigator of the Prostate Cancer Foundation. N. Sapre is a recipient of postgraduate scholarships from the Cancer Council Victoria and the Cybec Foundation.