Expression of microRNA-221 is progressively reduced in aggressive prostate cancer and metastasis and predicts clinical recurrence

Expression of microRNA-221 is progressively reduced inaggressive prostate cancer and metastasis and predictsclinical recurrence

Martin Spahn1, Susanne Kneitz2, Claus-Jurgen Scholz2, Nico Stenger1, Thomas Rudiger3, Philipp Strobel4,

Hubertus Riedmiller1 and Burkhard Kneitz1

1 Department of Urology and Paediatric Urology, University Medical School, University of Wurzburg, Wurzburg, Germany2Microarray Core Unit, IZKF (Interdisciplinary Center for Clinical Research), University of Wurzburg, Wurzburg, Germany3 Institute of Pathology, Community Hospital Karlsruhe, University Medical Center Mannheim, University of Heidelberg, Wurzburg, Germany4 Institute of Pathology, University Medical Center Mannheim, University of Heidelberg, Wurzburg, Germany

Emerging evidence shows that microRNAs (miR) are involved in the pathogenesis of a variety of cancers, including prostate

carcinoma (PCa). Little information is available regarding miR expression levels in lymph node metastasis of prostate cancer

or the potential of miRs as prognostic markers in this disease. Therefore, we analyzed the global expression of miRs in

benign, hyperplastic prostate tissue (BPH), primary PCa of a high risk group of PCa patients, and corresponding metastatic

tissues by microarray analysis. Consistent with the proposal that some miRs are oncomirs, we found aberrant expression of

several miRs, including the downregulation of miR-221, in PCa metastasis. Downregulation of miR-221 was negatively

correlated with the expression of the proto-oncogen c-kit in primary carcinoma. In a large study cohort, the prostate-specific

oncomir miR-221 was progressively downregulated in aggressive forms of PCa. Downregulation of miR-221 was associated

with clinicopathological parameters, including the Gleason score and the clinical recurrence during follow up. Kaplan–Meier

estimates and Cox proportional hazard models showed that miR-221 downregulation was linked to tumor progression and

recurrence in a high risk prostate cancer cohort. Our results showed that progressive miR-221 downregulation hallmarks

metastasis and presents a novel prognostic marker in high risk PCa. This suggests that miR-221 has potential as a diagnostic

marker and therapeutic target in PCa.

Prostate cancer is one of the most common visceral malig-nant neoplasms in men. It was estimated in 2006 that345,000 cases were newly diagnosed in the European commu-nity.1 The natural history of prostate carcinoma (PCa) variesfrom an indolent disease that might not cause symptomsduring a patient’s lifetime to a highly aggressive cancer thatmetastasises quickly and causes severe pain and untimelydeath. The marked disparity between these biological behav-iors is not currently understood and had an increasing socio-economic impact. Over the years, several potential prognostic

markers for PCa, including mRNA based gene expression sig-natures, have been identified.2–4 Unfortunately, both clinicalcriteria and molecular genetics approaches have had onlylimited success in patient stratification.

While low risk patients rarely develop tumor progressionor clinical recurrence after radical prostatectomy, patientswith high risk PCa had a 50% risk of progression at 5 years.5

Moreover we could show recently in a multicenter study onmore than 800 high risk PCa patients that 63% of thepatients experienced 15-years metastasis free survival afterradical prostatectomy (unpublished data). The main knownprognostic factors for treatment efficacy and survival, thePSA-value at diagnosis and the Gleason score,1,6,7 were notuseful in this high risk cohort as predictor for clinical pro-gression and mortality. Therefore new prognostic markers forhigh and low risk PCa patients are urgently needed to opti-mize and to individualize therapy strategies.

MicroRNAs (miR) are small (19–25-nt long), noncodingRNA strands that comprise a new class of regulatory mole-cules.8 Various alterations in miR expression were detected inmalignant tissues. Furthermore, functional studies of individ-ual miRs have shown that they appear to function in a tis-sue-specific manner, either as tumor suppressors or onco-genes (oncomiRs).9 miR expression profiles have been usedsuccessfully to classify various types of tumors by

Key words: microRNA, high risk prostate carcinoma, metastasis,

miR-221, prognosis

Additional Supporting Information may be found in the online

version of this article.

Grant sponsor: Interdisciplinary Center for Clinical Research

(IZKF), University of Wuerzburg

DOI: 10.1002/ijc.24715

History: Received 16 Jan 2009; Accepted 9 Jun 2009; Online

7 Jul 2009

Correspondence to: Burkhard Kneitz, Department of Urology and

Paediatric Urology, University Medical School, University of

Wurzburg, Oberdurrbacher Str. 8, D-97080 Wurzburg, Germany.

Fax: þ49-931-201-32719, E-mail: [email protected]

Early

Detection

andDiagn

osis

Int. J. Cancer: 127, 394–403 (2010) VC 2009 UICC

International Journal of Cancer

IJC

developmental lineage and state of differentiation10 andrecent studies have investigated the role of miRs in mediatingbreast cancer metastasis,11–13 These endeavors gave rise tothe notion that miR s might also be involved in the progres-sion and metastasis of PCa. Previous expression studies haveshown that there are characteristic miR signatures in PCa14.However, it is unknown whether these potential oncomiRsare relevant in the malignant progression and metastasis ofPCa.

In this study, we identified miRs, including miR-221, thatwere related to metastasis by comparing miR expression pat-terns in primary carcinoma tissue and in corresponding meta-static tissue. We further demonstrated that the downregulationof miR-221 was associated with tumor progression, poor prog-nosis and the development of metastasis. These results suggestthat mir-221 is a novel prognostic indicator in high risk PCaand might be a potential target for diagnosis and therapy.

Material and MethodsPatients and samples

This study was approved by the local ethical committee (no.59/04) and all patients provided written, informed consent.Samples were paraffin-embedded tissue specimens from ei-ther radical prostatectomy, lymph node metastasis (regionswith >90% cancerous tissue were used) or prostate adeno-mectomy (BPH) (regions with >80% adenoid tissue wereused). Patients had received operations in the Departmentsof Urology, at the Community Hospital of Karlsruhe and theUniversity Hospital of Wurzburg. All patients were recruitedfrom a well characterized group of high risk PCa.15 Allpatients were staged preoperatively with DRE, an abdomino-pelvic computed tomography (CT) scan and bone scan. Clin-ical node positive disease was not considered as exclusion cri-teria. None of the patients had received neoadjuvanthormonal-, radiation- or chemotherapy. Lymph node metas-tasis and prostate specimens (whole mount sections, 4 mmintervals) were staged and graded according to the 2002TNM classification and the Gleason grading system by onesenior pathologist (P.S.) (Table 1). Follow-up was performedevery 3 months for the first 2 years after surgery, every 6months in the following 3 years, and annually thereafter. Bio-chemical progression (BP) was defined as PSA �0.2 ng/mlon 2 consecutive follow-up visits. Clinical progression wasdefined either as histologically proven local recurrence or dis-tant metastasis confirmed by CT or bone scan. Overall sur-vival was defined as time from RRP to death of any cause,cancer specific survival as the time from RRP to death attrib-uted to PCa or complications of the disease. BPH sampleswere derived from prostate adenomectomy specimens. Allpatients had normal PSA levels before surgery and carcinomawas excluded by histopathology.

Microarray analysis

A set of 665 miRs (Probe Set 1564V2 mirVana, Applied Bio-systems) was spotted inhouse on SCHOTT NexterionTM

HiSense E microarray slides in quadruplicate. A completelisting of the probes in the probe set can be found at:www.ambion.com/techlib/resources/miR_array/index.html.

Slide processing was performed according to the AppliedBiosystems mirVanaTM manuals. For RNA purification fromformalin-fixed paraffin embedded (FFPE) material the Pure-Link FFPE Total RNA Isolation Kit (Invitrogen) was used incombination with the RiboMinus Concentration Module(Invitrogen). All steps were performed according to the rec-ommendations of Invitrogen. Each target was hybridized to aseparate array using the NcodeTM Rapid miRNA LabelingSystem, (Invitrogen).

RNA extraction and reverse transcription

Total RNA was extracted from the PCa and normal epithelialtissues with a Total RNA Extraction Kit (Applied Biosys-tems). The RNA concentration was determined with a

Table 1. Clinical demographics of patients with prostate carcinoma(N ¼ 92)

Clinical/pathologic features n (%)1

Age, years (range) 67 (52–78)

Follow up, months (st.dev) 74 (636.9)

MiR-221 expression2

reduced 90 (97.8%)

normal 2 (2.2%)

Clinical failure/clinical recurrence3

No 72 (78.2%)

Yes 20 (21.8%)

Preoperative PSA (ng/ml)

Mean (range) 38.4 (20–156)

Gleason score

6 2 (2.2%)

7 37 (40.2%)

8 7 (7.6%)

9 33 (35.9%)

10 13 (14.1%)

Pathological tumor stage

pT2 13 (14.1%)

pT3a 22 (23.9%)

pT3b 40 (43.5%)

pT4 17 (18.5%)

Lymph node positive

Positive 54 (58.7%)

Negative 38 (41.3%)

1Percentages indicate available patient data. 2Reduced miR-221expression is defined by greater than two-fold reductions in expressioncompared to the median BPH expression (n ¼ 9). 3Clinical recurrenceis the clinical failure after prostatectomy defined either ashistologically proven local recurrence or distant metastasis confirmedby a CT or bone scan that had the date of failure.

Early

Detection

andDiagn

osis

Spahn et al. 395


Bioanalyzer (Agilent). cDNA was synthesized from totalRNA with stem-loop reverse transcription primers accordingto the TaqMan miR Assay protocol (PE Applied Biosystems).

QRT-PCR

Mature miR expression was quantified in tissue samples withTaqManR miR assay kits and an Applied Biosystems7,900HT system. We followed the protocol provided in themanufacture’s instructions (Applied Biosystems). The expres-sion of snRNA RNU6b and RNU43 was used for normaliza-tion. Relative miRexpression was calculated with the compar-ative DCt-method (DCt sample ¼ Ct sample � Ct RNU6b; DCt BPH

¼ Ct BPH � Ct RNU6b). Fold changes in miR expressionbetween samples and controls were determined by the 2�DDc

t

method described previously.16 mRNA analysis of c-kit andp27Kip1 expression was performed according to standardqRT-PCR procedures. The expression of both GAPDH andb-Actin was used for normalization. All primer sequences are

available under request. Mean Ct was always determinedfrom triplicate PCRs.

Statistical and bioinformatics analysis

Spot intensities from scanned slides were quantified usingScanAlyze Software (M. Eisen, LBNL, CA). Data were ana-lyzed with different R packages from the Bioconductorproject (www.bioconductor.org). Resulting signal intensitieswere normalized by variance stabilization.17 Differentiallyexpressed genes were selected from the microarray data bylimma (Linear Models for Microarray Analysis) packageimplementing the empirical Bayes linear modelling approachdescribed by Smyth et al.,18,19 The output is a table of thetop-ranked genes determined from the linear model fitincluding a gene list, ratio on the log (base2) scale, averagegene intensities, moderated t-statistics, adjusted p value(FDR) and log odds. Correspondence analysis of microarraydata to visualize associations between genes and

Figure 1. (a) Hierarchical cluster analysis of 4 primary prostate carcinomas (PCa-1–4), 4 corresponding metastasis (met1–4), and 4 benign

prostatic hyperplasias (BPH). A tree was generated that clearly separates BPH from the carcinomas. Pairs of primary carcinomas and

corresponding metastases cluster with each other. (b) Correspondence analysis of 4 primary prostate carcinoma (PCa-1–4, green), 4

corresponding metastasis (met1–4, red), and 4 benign prostatic hyperplasias (BPH, blue). The graph represents a low-dimensional

projection to display associations between individual patients and distances between miRs (grey letters). Shorter distances between

patients or miRs indicate higher similarities. There is a clear separation between benign and malign cases on the main axis (x-axis). Malign

cases cluster with the patient of origin rather than to the tissue of origin. [Color figure can be viewed in the online issue, which is available

at www.interscience.wiley.com.]

Early

Detection

andDiagn

osis

396 miR-221 in high risk prostate cancer


hybridizations was done as described.20 Analysis of the RT-PCR data was performed with the prediction analysis formicroarrays (PAMR) package as described previously.21 Dif-ferent groups were compared with a Welch Two Sample t-test. Correlation between c-kit or p27kip1 mRNA expressionand expression of miR-221 is given as Spearman rank corre-lation coefficient. Survival analysis was assessed as the timefrom radical prostatectomy to clinical recurrence, and wasperformed with survival analysis software (Therneau et al.,1990). MiR-221 data were dichotomized with the intermedianmedian, which was determined by calculating the medianfrom the combined group medians representing an intuitivethreshold for group separation. The associations between fac-tors and clinical recurrence were determined by Cox regres-sion with uni- and multivariate models. p values <0.05 wereconsidered significant.

ResultsMicroarray detection of miRs differentially expressed in

primary PCas and metastasis

We compared the miR expression in 4 pairs of primary carci-noma and metastasis tissues vs. 4 nonmalignant prostate tis-sues (BPH) by hybridization of miRs on microarrays. A PCa-

specific miR signature was identified by the expression of 48downregulated and 18 upregulated miRs that were signifi-cantly different in carcinoma compared with BPH tissues(adj. p < 0.01, expression ratio >5; Supporting InformationTable 1). Hierarchical clustering of the microarray data gen-erated a tree with a clear distinction between 2 major groups:miRs related to PCa/metastasis and miRs related to BPH(Fig. 1a and Supporting Information Fig. 1). We confirmed arobust separation of BPHs from carcinoma with correspon-dence analysis. The clustering of all 4 metastasis samplesshowed more similarity to their corresponding primary tu-mor than to each other (Fig. 1). The expression of miRs inpairs of PCa versus metastasis was analyzed to identify miRsthat were differentially expressed in the primary PCa and itscorresponding synchronous lymph node metastasis. Statisticalanalysis yielded a list of miRs that were differentiallyexpressed (adj. p < 0.05, expression ratio >1.5) in BPH ver-sus PCa (143 up, 169 down), in BPH versus metastasis (161up and 196 down), and in PCa versus metastasis (8 up and 6down). We found that miR-221 was strongly downregulatedin PCa and metastasis compared with non malignant controltissue (Fig. 2a). Hierarchical clustering was performed basedon the 14 miRs that were differentially expressed between

Figure 2. (a) A Venn diagram showing relationships between human miRs that were differentially expressed in BPH versus primary

carcinoma (PCa), BPH versus metastasis (met), and primary carcinoma versus metastasis (FDR adj. p value <0.05, expression ratio >2, n ¼4 in all groups). Circles include the numbers of up or downregulated miRs for each pairwise comparison. Common miRs between different

comparisons are shown in the intersections; red ¼ upregulated, green ¼ downregulated miRs. The list of up or downregulated miRs in

primary carcinomas vs. metastases is shown (lower arrow points to the relevant box of miR identities). * indicates ambi-miRs. (b) Cluster

dendrogram using average linkage cluster analysis with Eisen’s correlation metric shows human miRs that exhibited a >1.5-fold increase or

decrease in expression (adj. p < 0.01) in 4 prostate carcinomas vs. 4 corresponding metastases. * indicates ambi-miRs.

Early

Detection

andDiagn

osis

Spahn et al. 397


PCa and metastasis. This generated a tree with clear distinc-tions between PCa, metastases and BPHs (Fig. 2b).

Validation of microarray data by real-time RT-PCR analysis

of miR-221, -29a, -16 and -125b in primary carcinoma

and metastasis

To confirm our microarray data, RT-PCR was performed toanalyze the expression of the most significantly regulatedmiRs, including miR-221, -125b, -29a and -16. The qRT-PCRanalysis using 9 BPH, 12 PCa and 12 corresponding metasta-ses confirmed that the overall expression levels of miR-221, -125b, -29a and -16 were downregulated in primary PCa andmetastasis samples (p < 0.01 for all 4 miRs) (Fig. 3a). Com-parison of BPH samples and non cancerous peripheral zoneprostatic tissue showed no difference in expression of all

4 miRs (Supporting Information Fig. 2), indicating that thesemiRs are not dysregulated in hyperplastic prostatic tissue.The expression levels of miR-125b, -29a and -16 were alsocomparable between PCa and metastasis samples; however,the overall expression level of miR-221 was downregulated (p< 0.01) in metastasis compared with its primary carcinoma,as expected from the microarray data. As shown in Figure3b, the expression of miR-221 was downregulated in 11 of 12metastasis samples in a direct comparison with the primarytumor.

Downregulation of miR-221 is negatively correlated to

mRNA over expression of c-kit in primary carcinoma

To assess the regulation of c-kit in primary PCa by miR-221we analyzed the expression of c-kit and p27Kip1 in groups of

Figure 3. Real-Time PCR assays in benign hyperplastic prostate tissue, primary carcinoma (PCa) and metastasis (met). (a) Box-and-whisker

plots of selected miRs predicted to be aberrantly regulated in prostate cancer. Relative expression of indicated miRs in 9 BPHs (blue), 12

primary carcinomas (green), and 12 corresponding lymph node metastasis (red) was analyzed by qRT-PCR. (b) Downregulation of miR-221

expression in metastasis. Expression levels of miR-221 in 12 pairs of carcinoma and metastases were analyzed by qRT-PCR and calculated

by the DCt-method. The graph shows the log fold change of mir-221 in metastasis (M1-M12) as compared with the value obtained for the

corresponding carcinoma. The samples are shown in ascending order based on log fold change. (c) Correlation of miR-221 and c-kit

expression in primary prostate cancer. Relative expression of miR-221 and c-kit was analyzed in 9 BPH (blue), 18 carcinoma samples with

high miR-221 expression (PCa A; green) and 18 carcinoma samples with low expression of miR-221 (PCa B, yellow) by qRT-PCR. Box- and

whisker plots of miR-221 and c-kit show significant increased expression of c-kit in PCa B samples. In (a) and (c): Median expression levels

are indicated with black horizontal bars. The relative expression level in BPHs was arbitrarily set as 0. * indicates p < 0.001. p values were

calculated using the Welch 2 sample t-test.

Early

Detection

andDiagn

osis



PCa samples, characterized by high or low miR-221 expres-sion. We could not find any correlation between mir-221expression and the level of mRNA expression of p27Kip1(data not shown). However, tumor samples, that were charac-terized by low miR-221 expression, had a noticeably highexpression of c-kit mRNA levels (Fig. 3c). Comparing thevalues of c-kit mRNA expression and miR-221 expression aninverse Spearman Rank correlation of q ¼ �0.61 wasevident.

Downregulation of miR-221 is associated with tumor

progression and recurrence

On the basis of the result that miR-221 was progressivelydownregulated in metastasis, we postulated that dysregulationof miR-221 might also be associated with clinicopathologicalfeatures or prognosis of PCa in patients. To test this hypothe-sis, we used qRT-PCR to analyze miR-221 expression in a

large study cohort consisting of 92 patients. The clinical demo-graphics of the study cohort are summarized in Table 1. Weanalyzed the expression in each carcinoma sample and com-pared them to the median expression in 9 BPH samples. Withthe DD Ct method, we determined that miR-221 expression in90 of 92 (98%) carcinomas was more than 2-fold lower thanthe median expression in BPH (Table 1). Next we assessedwhether the expression of miR-221 was related to the Gleasonscore (Gleason score �7 vs. �8), the pathologic stage (pT2 vs.pT3 vs. pT4), or the clinical recurrence of the carcinoma. Wefound that miR-221 was downregulated in samples frompatients that had high Gleason scores, advanced stage tumorsand clinical recurrences (Fig. 4a).

miR-221 as risk factor in PCa

To determine whether miR-221 could serve as a prognosticindicator for clinical recurrence, we divided the sample

Figure 4. (a) Box- and wisker-plots show the expression of miR-221 in mild and aggressive prostate carcinomas. Relative miR-221

expression was analyzed in 92 prostate carcinoma samples by the DCt method using qRT-PCR. Subsequently, the samples were divided

into subgroups of tumor severity based on the Gleason score (left plot), the pathological tumor stage (middle plot), or clinical progression

(right plot). Specifics of the different subgroups are summarized in Table 1. Significant reductions in the median expression levels (black

bars) between subgroups are marked by * (p < 0.01). p values were calculated with the Welch 2 sample t-test; (b) Kaplan–Meier analysis

of patients with prostate cancer. Patients were dichotomized by miR-221 expression (miR-221 high ¼ DCt � 0.77; miR-221 low ¼ DCt <

0.77) or by Gleason score (GS � 7 or GS > 7). The survival curves were generated using the Bioconductor package for survival. Low miR-

221 expression or high Gleason scores were significantly associated with earlier clinical recurrence of the tumor (log-rank p < 0.01).

Early

Detection

andDiagn

osis

Spahn et al. 399


population into groups with either low or high miR-221expression levels, with an intermedian cut-off of DCt ¼ 0.77(i.e., the 39th and 71st percentiles in patients without andwith recurrences, respectively). Kaplan–Meier estimates pre-dicted a significant difference between groups in rates of re-currence- free survival. The 10 year survival rate was 87.2%in patients with high miR-221 expression and 53.5% in thosewith low miR-221 expression. Cox regression showed thatdichotomized and continuous miR-221 expression levels, PSAlevels, dichotomized pathologic stage and the biomodal Glea-son score were univariately significant for the prediction ofclinical recurrence; in contrast, the age of the patient at thetime point of radical prostatectomy was not a significant fac-tor (Table 2). Kaplan–Meier estimates of cancer-free timeand recurrence-free survival confirmed that a Gleason score�7 and high miR-221 levels were associated with a goodprognosis for the patient (Fig. 4b). When these variables wereconsidered together in a multivariate proportional hazardsmodel, the miR-221 expression level (p ¼ 0.032), the Gleasonscore (p ¼ 0.05), and the pathological stage (p ¼ 0.01) werestill significant for prediction of clinical recurrence. Thus,miR-221 downregulation was significantly associated with tu-mor progression and clinical recurrence.

DiscussionAn ever growing number of articles have been publishedwithin the past few years that describe a link between theexpression of miR and human cancer, including PCa.22 How-ever, despite the large body of work that has been publishedto date, only limited information is available regarding theexpression levels of specific miRs in relation to the aggres-

siveness or the potential of miR expression as a prognosticmarker in PCa.14

Therefore, we initially analyzed miR expression patternsin 4 pairs of primary tumors and the corresponding lymphnode metastases relative to non cancerous prostate tissue.Our microarray comparison of global miR expressiondetected a PCa-specific miR signature represented by a set ofmiRs. This finding is consistent with previous studies on miRexpression in PCa. One or more studies have shown thatmost of the highly significant dysregulated miRs (e.g., miR-145, miR -205, miR-16, miR-29a and let-7c) were also foundin PCa miR profiles.23–26 In this study, the hierarchical clus-ter analysis of micro array data showed that metastasestended to cluster with their corresponding primary tumorrather than with each other (and vice versa). This result isconsistent with a clonal development of metastasis from theprimary tumor.27 It shows that miR-expression is not sub-stantially altered in metastasising PCa. In a greater numberof samples (12 pairs of metastasis and primary carcinoma),we clearly confirmed that 3 tumorigenic miRs (miR- 16,-125b and -29a) were expressed at comparable levels in theprimary tumor and the corresponding metastasis.

Despite the very similar expression signature in metastasisand the primary tumor, we found a metastasis-specific miRprofile characterized by 14 differentially expressed miRs (8up-, 6 downregulated). In a cluster analysis, these metastasis-specific miRs could successfully separate all 4 metastasesfrom the primary carcinoma. Thus, these miRs comprise anovel miR expression profile for PCa metastases. Recently, ithas been shown that miR s promoted tumor invasion andmetastasis in breast cancer; this suggested that the dysregula-tion of specific miRs provided a selective advantage in breast

Table 2. Cox proportional hazards regression for time to clinical recurrence

Variable n

Univariate Multivariate

HR (95%Cl) p-value HR (95% Cl) p-value

mIR-221

mIR-221 (conditions) 92 0.47 (0.28–0.80) 0.0055 0.525 (0.29–0.95) 0.032

mIR-221 Dc � 0.776 49 1.00

mIR-221 Dc < 0.776 (dichotomized) 43 2.68 (1.04–6.93) 0.042 NA NA

Pre-PSA

pre-PSA (conditions) 1.01 (1–1.03) 0.019 NA NA

Gleason Score

GS � 7 32 1.00

GS � 8 60 3.26 (1.35–7.86) 0.009 2.5 (1–6.3) 0.05

TNM stage

T2/T3 75 1.00

T4 17 3.02 (1.25–7.31) 0.014 3.24 (1.32–7.93) 0.01

Age (years)

�66 44 1.00

>66 48 1.66 (0.70–3.91) 0.25 NA NA

Early

Detection

andDiagn

osis



cancer metastasis11–13 Notably, we found no overlap betweenthe metastasis-specific miRs of PCa and those found in breastcancer. These findings suggest that the dysregulation of miRsin metastasis depends on the cellular microenvironment.

Currently, little is known about these PCa metastasis-spe-cific miRs except mir-221. Interestingly, only miR-221 wasdownregulated in comparisons between the primary tumorvs. metastasis, BPH and the primary tumor, and BPH vs. me-tastasis. We confirmed a progressive downregulation of miR-221 in metastasis by direct comparison of miR-221 expres-sion in 12 independent pairs of tumor and metastatic tissues.It has previously been shown that miR-221 and miR-222were over-expressed in tumors of the colon, breast, pancreasand in glioblastoma, papillary thyroid carcinoma and chroniclymphocytic leukaemia.28–32 Furthermore, the upregulation ofmiR-221 in chronic lymphocytic leukemia was found to beassociated with a poor prognosis.33 In contrast to the upregu-lation of miR-221 described for different types of cancer, weshowed that 98% of the analyzed PCas were characterized bya downregulation of miR-221. Strong downregulation ofmiR-221 in PCa was clearly confirmed by 4 other miRNAexpression studies of PCa.23–25 Therefore, it might be surpris-ing that the miR-221 cluster has been found to regulate thep27Kip1 tumor suppressor in different types of tumor celllines, including prostate cells.34–36 In recent studies it has fur-ther been shown that downregulation of p27Kip1 caused byover-expression of miR-221 is critically involved in the main-tenance of androgen independency of PCa cell lines and thatinhibition of miR-221 expression impairs the growth of xeno-grafts in a murin tumor model. In contrast, the finding thatmiR-221 downregulation is a very common event in PCaimplies that miR-221 mediated inhibition of p27Kip1 mightplay no role in the development or progression of primaryPCa. This suggestion is supported by our observation thatmiR-221 expression is not correlated to the mRNA levels ofp27Kip1 in primary PCa. The opposite regulation of miR-221in different cancer types seems also paradoxical, but it mightconsistent with the observation that miR-mediated regulationof mRNA target molecules may be determined by the specificcellular microenvironment.26 It is also widely accepted thatthere is not a ‘‘one to one’’ relationship between miR s andtarget mRNAs and that the relationship strongly depends onan orchestral, dynamic regulation.37 Therefore we suggestthat the discrepancy between miR-221 down regulation inprimary PCa samples and miR-221 mediated downregulationof p27Kip1 in PCa cell lines might be depend on molecularchanges, which occur while a primary tumor adapts to cellculture conditions. Currently we do not precisely know howmiRs are regulated or processed in tumors and how manymiR-221 targets exist, that might be involved in developmentof primary PCa. To understand the function of miR-221 andto solve the observed discrepancies in the role of miR-221 forthe development of PCa it seems necessary to analyze molec-ular changes of components in the miRNA processing ma-chinery and of additional potential targets of miR-221 in PCa

samples and cancer cell lines. It has been shown that miR-221 downregulation is involved in the growth of erythroleu-kemic cells via inhibition of the c-kit oncogene.38 In ourstudy we could show that downregulation of miR-221corre-lates with upregulation of c-kit on mRNA level. However, c-kit is not or only weekly expressed in prostate epithelial cells.Yet a study on c-kit expression in bone metastasis underlinedthe potential relevance of c-kit as metastasis specific onco-gene in PCa.39 But further research is necessary to determinewhether miR-221 mediated regulation of c-Kit plays a criticalrole in carcinogenesis or malignant progression.

Our results show that progressive reduction of miR-221expression is a prostate specific and very frequent event inmetastasis. In recent years, studies have revealed that the sig-nalling steps in the invasion-metastasis cascade are linked tothe dysregulation of genes or gene products.40 Detection ofthese gene aberrations in the primary tumor is often associ-ated with a poor prognosis for the patient. Thus, we postu-lated that a reduction in miR-221 expression might also beassociated with clinico-pathological parameters in the pri-mary PCa. To test this hypothesis, we analyzed miR-221expression in a large PCa study group. Most of the studieson radical prostatectomy included patients that presentedwith localized, intermediate, and low risk PCa following theradical prostatectomy.41,42 It is difficult to identify biologicallyrelevant parameters that are significantly associated with tu-mor aggressiveness and progression in these patient cohorts,because these series represent only limited numbers of localtumor recurrence and metastatic disease. To acquire a statis-tically satisfying number of high grade primary tumors, weused radical prostatectomy specimens from a clinically well-defined, high risk group of patients with PCa.15 A high pro-portion of the patients had locally advanced, high grade dis-ease and lymph node metastases. In these tissue specimens,we detected a progressive downregulation of miR-221 thatwas associated with high risk carcinoma and a poor progno-sis, based on the Gleason Score, the tumor stage, or the rateof clinical recurrence. In these tissue specimens, we detecteda progressive downregulation of miR-221 that was associatedwith poor prognosis, based on the Gleason Score, the tumorstage, or the rate of clinical recurrence. The sample popula-tion was then divided into groups with high or low miR-221expression. Kaplan–Meier estimates predicted a significantlylower survival rate in patients that had samples with lowmiR-221 expression compared with those with high miR-221expression. The Cox proportional hazard analysis for time toclinical recurrence demonstrated the predictive power (i.e.,large hazard ratio) of miR-221 expression as a continuousand dichotomized variable. In fact, in uni- and multivariateanalysis of the high risk study cohort used, high miR-221expression as a continuous variable had higher predictivepower than the Gleason Score, which is known as one ofthe strongest conventional predictors of tumor recurrence.6

Because patients with high risk PCa had a 37% risk of clini-cal recurrence at 15 years after radical prostatectomy

Early

Detection

andDiagn

osis

Spahn et al. 401


(unpublished data), the detection of a powerful prognostica-tor in high risk carcinoma patients might be helpful for thedecision of additional therapy strategies in these patients.Our findings demonstrate a high predictive power and inde-pendence of miR-221 as a prognostic indicator in high riskPCa. The role of mir-221 as prognostic indicator in a moreconsecutive series of PCa patients after radical prostatectomyhas to be analyzed in the future using an enlarged studycohort.

To our knowledge, our study is the first to date that identi-fied a single miR that was associated with clinical outcomesand metastasis of PCa. Informative biomarkers are urgently

needed to guide clinical interventions for PCa. Our studyreports the progressive downregulation of miR-221 in primaryPCa and metastasis and provides strong evidence that miR-221can be used as a novel prognostic indicator in high risk PCa.Further research is needed to analyze the underlying mecha-nisms of miR-221 mediated tumorigenesis and further supportthe utility of miR-221 as a potential diagnostic marker andtherapeutic target for high risk PCa and metastasis.

AcknowledgementsThe authors thank Ms. S. Muller, Ms. B. Dexler and Ms K. Borschert forexcellent technical assistance.

References

1. Ferlay J, Autier P, Boniol M, Heanue M,Colombet M, Boyle P. Estimates of thecancer incidence and mortality in Europein 2006. Ann Oncol 2007;18:581–92.

2. Dhanasekaran SM, Barrette TR, Ghosh D,Shah R, Varambally S, Kurachi K, PientaKJ, Rubin MA, Chinnaiyan AM.Delineation of prognostic biomarkers inprostate cancer. Nature 2001;412:822–6.

3. Singh D, Febbo PG, Ross K, Jackson DG,Manola J, Ladd C, Tamayo P, RenshawAA, D’Amico AV, Richie JP, Lander ES,Loda M, et al. Gene expression correlatesof clinical prostate cancer behavior. CancerCell 2002;1:203–9.

4. Varambally S, Dhanasekaran SM, Zhou M,Barrette TR, Kumar-Sinha C, Sanda MG,Ghosh D, Pienta KJ, Sewalt RG, Otte AP,Rubin MA, Chinnaiyan AM. Thepolycomb group protein EZH2 is involvedin progression of prostate cancer. Nature2002;419:624–9.

5. D’Amico AV, Whittington R, MalkowiczSB, Fondurulia J, Chen MH, Kaplan I,Beard CJ, Tomaszewski JE, Renshaw AA,Wein A, Coleman CN. Pretreatmentnomogram for prostate-specific antigenrecurrence after radical prostatectomy orexternal-beam radiation therapy forclinically localized prostate cancer. J ClinOncol 1999;17: 168–72.

6. Berney DM, Fisher G, Kattan MW, OliverRT, Moller H, Fearn P, Eastham J,Scardino P, Cuzick J, Reuter VE, FosterCS. Major shifts in the treatment andprognosis of prostate cancer due tochanges in pathological diagnosis andgrading. BJU Int 2007;100:1240–4.

7. Rodriguez-Covarrubias F, Larre S, De LaTaille A, Abbou CC, Salomon L. Theoutcome of patients with pathologicalGleason score > or ¼ 8 prostate cancerafter radical prostatectomy. BJU Int 2008;101: 305–7.

8. Du T, Zamore PD. microPrimer: thebiogenesis and function of microRNA.Development 2005;132:4645–52.

9. Calin GA, Croce CM. MicroRNAsignatures in human cancers. Nat RevCancer 2006;6:857–66.

10. Lowery AJ, Miller N, McNeill RE, KerinMJ. MicroRNAs as prognostic indicatorsand therapeutic targets: potential effect onbreast cancer management. Clin Cancer Res2008;14:360–5.

11. Huang Q, Gumireddy K, Schrier M, leSage C, Nagel R, Nair S, Egan DA, Li A,Huang G, Klein-Szanto AJ, Gimotty PA,Katsaros D, et al. The microRNAs miR-373and miR-520c promote tumor invasionand metastasis. Nat Cell Biol 2008;10:202–10.

12. Ma L, Teruya-Feldstein J, Weinberg RA.Tumour invasion and metastasis initiatedby microRNA-10b in breast cancer. Nature2007;449: 682–8.

13. Tavazoie SF, Alarcon C, Oskarsson T,Padua D, Wang Q, Bos PD, Gerald WL,Massague J. Endogenous humanmicroRNAs that suppress breast cancermetastasis. Nature 2008;451:147–52.

14. Shi XB, Tepper CG, Devere White RW.MicroRNAs and prostate cancer. J Cell MolMed 2008;12:1456–65.

15. Spahn M, Bader P, Woehr M, Frohneberg,M. PCA with PSA > 20—is there a chanceof cure?Eur Urol 2006;5:212–3.

16. Schmittgen TD, Livak KJ. Analyzing real-time PCR data by the comparative C(T)method. Nat Protoc 2008;3:1101–8.

17. Huber W, von Heydebreck A, Sultmann H,Poustka A, Vingron M. Variancestabilization applied to microarray datacalibration and to the quantification ofdifferential expression. Bioinformatics 2002;18 (Suppl 1):S96–104.

18. Smyth GK, Michaud J, Scott HS. Use ofwithin-array replicate spots for assessingdifferential expression in microarrayexperiments. Bioinformatics 2005;21:2067–75.

19. Gentleman RC, Carey VJ, Bates DM,Bolstad B, Dettling M, Dudoit S, Ellis B,Gautier L, Ge Y, Gentry J, Hornik K,

Hothorn T, et al. Bioconductor: opensoftware development for computationalbiology and bioinformatics. Genome Biol2004;5:1–16.

20. Fellenberg K, Hauser NC, Brors B,Neutzner A, Hoheisel JD, Vingron M.Correspondence analysis applied tomicroarray data. Proc Natl Acad Sci USA2001;98:10781–6.

21. Tibshirani RJ, Efron B. Pre-validation andinference in microarrays. Stat Appl GenetMol Biol 2002;1:1–18.

22. Croce CM. Oncogenes and cancer. N EnglJ Med 2008;358:502–11.

23. Ambs S, Prueitt RL, Yi M, Hudson RS,Howe TM, Petrocca F, Wallace TA, LiuCG, Volinia S, Calin GA, Yfantis HG,Stephens RM, et al. Genomic profiling ofmicroRNA and messenger RNA revealsderegulated microRNA expression inprostate cancer. Cancer Res 2008;68:6162–70.

24. Porkka KP, Pfeiffer MJ, Waltering KK,Vessella RL, Tammela TL, Visakorpi T.MicroRNA expression profiling inprostate cancer. Cancer Res 2007;67:6130–5.

25. Ozen M, Creighton CJ, Ozdemir M,Ittmann M. Widespread deregulation ofmicroRNA expression in human prostatecancer. Oncogene 2008;27:1788–93.

26. Volinia S, Calin GA, Liu CG, Ambs S,Cimmino A, Petrocca F, Visone R, IorioM, Roldo C, Ferracin M, Prueitt RL,Yanaihara N, et al. A microRNAexpression signature of human solidtumors defines cancer gene targets. ProcNatl Acad Sci USA 2006;103:2257–61.

27. Fidler IJ. The pathogenesis of cancermetastasis: the ‘‘seed and soil’’ hypothesisrevisited. Nat Rev Cancer 2003;3:453–8.

28. Iorio MV, Ferracin M, Liu CG, VeroneseA, Spizzo R, Sabbioni S, Magri E, PedrialiM, Fabbri M, Campiglio M, Menard S,Palazzo JP, et al. MicroRNA geneexpression deregulation in human breastcancer. Cancer Res 2005;65:7065–70.

Early

Detection

andDiagn

osis



29. Lee EJ, Gusev Y, Jiang J, Nuovo GJ, LernerMR, Frankel WL, Morgan DL, Postier RG,Brackett DJ, Schmittgen TD. Expressionprofiling identifies microRNA signature inpancreatic cancer. Int J Cancer 2007;120:1046–54.

30. Takamizawa J, Konishi H, Yanagisawa K,Tomida S, Osada H, Endoh H, Harano T,Yatabe Y, Nagino M, Nimura Y,Mitsudomi T, Takahashi T. Reducedexpression of the let-7 microRNAs inhuman lung cancers in association withshortened postoperative survival. CancerRes 2004;64:3753–6.

31. Ciafre SA, Galardi S, Mangiola A, FerracinM, Liu CG, Sabatino G, Negrini M, MairaG, Croce CM, Farace MG. Extensivemodulation of a set of microRNAs inprimary glioblastoma. Biochem Biophys ResCommun 2005;334:1351–8.

32. He H, Jazdzewski K, Li W, LiyanarachchiS, Nagy R, Volinia S, Calin GA, Liu CG,Franssila K, Suster S, Kloos RT, Croce CM,et al. The role of microRNA genes inpapillary thyroid carcinoma. Proc NatlAcad Sci USA 2005;102:19075–80.

33. Calin GA, Ferracin M, Cimmino A, DiLeva G, Shimizu M, Wojcik SE, Iorio MV,Visone R, Sever NI, Fabbri M, Iuliano R,Palumbo T, et al. A MicroRNA signature

associated with prognosis and progressionin chronic lymphocytic leukemia. N Engl JMed 2005;353:1793–801.

34. Galardi S, Mercatelli N, Giorda E,Massalini S, Frajese GV, Ciafre SA, FaraceMG. miR-221 and miR-222 expressionaffects the proliferation potential of humanprostate carcinoma cell lines by targetingp27Kip1. J Biol Chem 2007;282:23716–24.

35. le Sage C, Nagel R, Egan DA, Schrier M,Mesman E, Mangiola A, Anile C, Maira G,Mercatelli N, Ciafre SA, Farace MG,Agami R. Regulation of the p27(Kip1)tumor suppressor by miR-221 and miR-222 promotes cancer cell proliferation.EMBO J 2007;26: 3699–708.

36. Visone R, Russo L, Pallante P, De MartinoI, Ferraro A, Leone V, Borbone E, PetroccaF, Alder H, Croce CM, Fusco A.MicroRNAs (miR)-221 and miR-222, bothoverexpressed in human thyroid papillarycarcinomas, regulate p27Kip1 protein levelsand cell cycle. Endocr Relat Cancer 2007;14:791–8.

37. Baek D, Villen J, Shin C, Camargo FD,Gygi SP, Bartel DP. The impact ofmicroRNAs on protein output. Nature2008;455:64–71.

38. Felli N, Fontana L, Pelosi E, Botta R, BonciD, Facchiano F, Liuzzi F, Lulli V, Morsilli

O, Santoro S, Valtieri M, Calin GA, et al.MicroRNAs 221 and 222 inhibit normalerythropoiesis and erythroleukemic cellgrowth via kit receptor down-modulation.Proc Natl Acad Sci USA 2005;102:18081–6.

39. Wiesner C, Nabha SM, Dos Santos EB,Yamamoto H, Meng H, Melchior SW,Bittinger F, Thuroff JW, Vessella RL, CherML, Bonfil RD. C-kit and its ligand stemcell factor: potential contribution toprostate cancer bone metastasis. Neoplasia2008;10:996–1003.

40. Gupta GP, Massague J. Cancer metastasis:building a framework. Cell 2006;127:679–95.

41. Chun FK, Graefen M, Zacharias M, HaeseA, Steuber T, Schlomm T, Walz J,Karakiewicz PI, Huland H. Anatomicradical retropubic prostatectomy-long-termrecurrence-free survival rates for localizedprostate cancer. World J Urol 2006;24:273–80.

42. Hernandez DJ, Nielsen ME, Han M,Trock BJ, Partin AW, Walsh PC, EpsteinJI. Natural history of pathologicallyorgan-confined (pT2). Gleasonscore 6 or less, prostate cancer afterradical prostatectomy. Urology 2008;72:172–6.

Early

Detection

andDiagn

osis

Spahn et al. 403


Expression of microRNA-221 is progressively reduced in aggressive prostate cancer and metastasis and predicts clinical recurrence

Documents