High miR-449b expression in prostate cancer is associated with biochemical recurrence after radical prostatectomy

There is a great unmet need for better diagnostic as well as predictive tools in prostate cancer (PC) needed for discerning aggressive PC from indolent PC. Although this disease has the second highest cancer mortality amongst men, approximately 250,000 worldwide [1]. The prevalence hereof, as indicated by autopsy studies, is even higher and reported as being up to 12-54% of all men aged over 50 years [2]. Early diagnosis is imperative since curative treatment is only possible in non-metastatic PC. This has led to the widespread use of screening for PC using Prostate Specific antigen (PSA) testing, but since the prevalence of non-lethal, non-progressive PC is very high, screening leads to a substantial risk of overtreatment [3]. Biomarkers associated with aggressive PC have the potential of improving existing prediction models for identifying patients with aggressive disease and thus aid patients and physicians in deciding between curative treatment and active surveillance [4]. Furthermore, biomarkers associated with aggressive disease could be used for monitoring patients in active surveillance and be used as a trigger for offering curative treatment if the biomarker level changes [5].

In brief, microRNAs are small 18 to 25 nucleotide RNA molecules that function as regulators of gene expression through incorporation into the RNA-induced silencing complex (RISC). The incorporated microRNA binds to mRNA sequences with complementary to the microRNA. Following binding, the Argonaute (Ago) protein also incorporated in RISC cleaves the mRNA strand and thus causes degradation of the target transcript and thereby silences the gene [6]. Through this post transcriptional process miRNA can influence gene expression and thus regulate various biological processes. Since the key biological differences between cancer cells and their normal counterparts are the initiation and utilization of cellular processes like increased proliferation, immortalization and invasive properties, altering gene expression is central in tumorigenesis [7]. Previous studies have shown that miRNA can be central in orchestrating the altered gene expression necessary for the cell to undergo transformation to a cancer cell [6]. As each miRNA can have many different putative target mRNAs, a miRNA can essentially be the central tumorigenic factor [8] or it can function through inhibiting mRNAs that in turn function as either oncogenes or tumor suppressors [9, 10].

miRNAs are generally more stable and easier to measure in biological material where degradation is an issue compared to mRNA [11, 12]. Therefore, miRNAs may prove to be good biomarkers for diagnosis as well as for monitoring PC. It has been shown in several studies that miRNA can be assayed in blood from PC patients and that diagnostic miRNAs can be identified [13, 14].

Several studies have shown that miRNAs are aberrantly expressed in PC and several specific miRNAs have been implicated in PC development [15‐19]. While most studies of miRNA in PC have focused on miRNAs differentially expressed between normal and cancerous prostate tissue, only a few studies have focused on miRNA expression associated with aggressive disease. A number of different surrogate end-points for aggressive PC have been used in the studies like presence of perineural invasion [20], Gleason grade [20, 21], extra prostatic growth [16] and risk of recurrence [22]. These differences in end-points, and the limited power of some studies due to few analyzed samples, are probably the main reasons that reproducing results is difficult, as described by Coppola et al. [23].

Here we performed microRNA profiling using multiplex qPCR on laser micro dissected material from 36 PC patients to identify microRNAs associated with recurrence after radical prostatectomy (RP). We validated the association with recurrence of the top ranked miR-449b in an independent cohort of patients using singleplex RT-qPCR.

In total, 36 tumor samples were laser micro dissected and 672 miRNAs profiled using multiplex RT-qPCR to identify novel miRNAs associated with recurrence after RP. Sixty % of the patients had suffered recurrence and the median follow-up of those without recurrence was 66 months (range 31–80 months). Clinical and histopathological information is listed in Table 1. Initially the miRNA expression data was filtered to exclude miRNAs with no detection above background in any of the samples, leaving 536 miRNAs for further analysis. A total of 235 (44%) miRNAs were detected in all samples in the filtered dataset.

Our study revealed 31 miRNAs differentially expressed between patients suffering recurrence and patients with no recurrence. Furthermore the top ranked miR-449b was successfully validated in an independent cohort of 163 patients using RT-qPCR and found to be an independent predictor of biochemical recurrence after prostatectomy.

miRNA in relation to cancer constitutes an interesting field of research due to their role in gene expression regulation and because they are generally more stable than mRNA. There is however a number of conflicting results from miRNA profiling studies including irreproducible results. One major issue is the sample itself, and the tissue actually being profiled. Samples where the carcinoma cell percentage is low, will inevitably reveal a miRNA profile which is a mix the miRNAs from the carcinoma cells and the profile from the surrounding normal tissue, thus diluting the miRNA expression in the carcinoma cells. Regarding studies of aggressive PC, tumor heterogeneity has often been overlooked. Normally only a single tumor biopsy is used and the miRNA profile produced is a measurement of the miRNA expression in that given part of the tumor. If the tumor contains more aggressive clones in other parts of the prostate a discrepancy will arise between the clinical performance of the tumor and the miRNA profile obtained from the more indolent tumor cells. Another issue influencing reproducibility is the technical differences between studies. Although good reproducibility exist within RT-qPCR profiling platforms, the correlation with microarray platforms is often not high, as reported by Chen et al. [26]. The number of miRNAs actually being profiled also differs between studies ranging from 119 [19] to 676 [22], in addition since different platforms are used the specificity and sensitivity of the probes detecting the miRNAs can differ leading to greater variability of the expression measurements. However, in spite of these challenges, several previously identified miRNAs associated with adverse characteristics in PC were also identified in our study. The miRNAs, miR-449b and miR-10b were seen previously by Fendler et al. [22], differentiating between early or late recurrence. Furthermore, several miRNAs (miR-126, miR-125-5p, miR-10a/b) overlapped with the miRNAs identified by Prueitt et al. [20] being associated with perineural invasion. Finally, miR-449b and miR-484 overlapped with the miRNAs associated with extraprostatic disease extension identified by Ambs et al. [16]. Among the down regulated microRNAs found in this study, miR-154 was also seen down regulated in patients with biochemical recurrence after prostatectomy by Tong et al. [19]. In addition to the microRNAs previously linked to aggressive traits in PC several new microRNAs were identified to be associated with recurrent disease.

The use of biochemical recurrence as surrogate end-point of aggressive prostate cancer has limitations. Other end-points like time to metastasis or progression to death are stronger end-points for the clinical aggressiveness of prostate cancer, unfortunately this data was not available for the cohort. The Gleason grade included in the multivariate model of the validation cohort is based on the original pathology grading. Gleason grading has changed over time and a re-evaluation of the prostatectomy specimens would strengthen the model.

The miR-449b gene is situated on chromosome 5 in the second intron of the CDC20B gene. MiR-449b shares seed sequence with miR-449a which is also transcribed in the second intron of CDC20B. This off course implies similar mRNA targets and thus similar functions. In our study we found that miR-449a had a similar fold change difference of 3.8 compared with 3.2 for miR-449b between the recurrent and non-recurrent cases, but miR-449a expression was not significantly associated with recurrence in Mann–Whitney test. This difference can be due to specificity and/or sensitivity of the PCR probes for miR-449a and miR-449b respectively.

As described above the miR-449 cluster has previously been associated with adverse outcomes in PC. In contrast to this miR-449 functional studies have provided conflicting results. The miR-449 cluster has been implicated in different cellular functions primarily in cell cycle control and in cellular differentiation [27]. MiR-449b has been shown to inhibit androgen receptor expression resulting in inhibited androgen mediated proliferation [28]. Studies using the Saos-2 osteosarcoma cell line show that mir-449 cluster transcription is inducible by the E2F1 transcription factor. Induced miR-449 has been shown to cause causes cell cycle arrest and promote apoptosis by inhibiting CDK6 and CDC25A. In this way it functions as a negative feedback mechanism against E2F1 induced cell proliferation [28]. In other studies using cell lines originating from prostate, breast and lung cancer the miR-449 cluster has also been associated with cell cycle arrest and a role as tumor suppressor although through different mechanisms [29‐31]. However a different functional role of miR-449 cluster is the interaction with LEF-1 a known effector of the WNT pathway [32]. In the cell line hBM-MSC the miR-449 cluster has been shown to directly inhibit the LEF-1 gene and thus the WNT pathway. Although WNT pathway inhibition in general inhibits cancer growth, including PC, [33‐35] other effects have been observed that relate to PC progression. Firstly LEF-1 inhibition has been shown to inhibit osteoblast differentiation and result in reduced bone density, a phenotype that would facilitate the establishment of PC cells in bone [36]. Secondly studies looking at the interaction between WNT signaling and androgen receptor function reveals a complex interaction with some evidence that WNT pathway signaling enhances androgen receptor signaling other evidence inhibiting androgen signaling [37]. The miR-449 cluster has also been identified targeting the NOTCH pathway thus altering cell differentiation [38]. Whereas NOTCH pathway signaling again in general is shown to induce EMT and cancer progression, NOTCH signaling is also linked to inhibiting bone formation. In this manner and like the inhibition of LEF-1, the mechanism of action for miR-449b overexpression in aggressive PC could be an increased propensity to metastasize to bone. Further studies are needed to investigate the functional role of miR-449 in normal and malignant prostate cells.

The current study identified 31 miRNAs to be associated with biochemical recurrence of prostate cancer after radical prostatectomy. The highest ranked miRNA, miR-449b, was further validated in an independent cohort of patients and shown to be an independent predictor of recurrence after radical prostatectomy.