Investigation of the molecular mechanisms underlying metastasis in prostate cancer by gene expression profiling.

The present study aimed to screen potential genes associated with metastatic prostate cancer (PCa), in order to improve the understanding of the mechanisms underlying PCa metastasis. The GSE3325 microarray dataset, which was downloaded from the Gene Expression Omnibus database, consists of seven clinically localized PCa samples, six hormone-refractory metastatic PCa samples and six benign prostate tissue samples. The Linear Models for Microarray Data package was used to identify differentially-expressed genes (DEGs) and a hierarchical cluster analysis for DEGs was performed with the pheatmap package. Furthermore, potential functions for the DEGs were predicted by a functional enrichment analysis. Subsequently, microRNAs (miRNAs) potentially involved in the regulation of PCa metastasis were identified by WebGestalt software, and the miRNA-DEG regulatory network was visualized using Cytoscape. In addition, a pathway enrichment analysis for DEGs in the regulatory network was performed. A total of 306 and 2,073 genes were differentially expressed in the clinically localized PCa and the metastatic PCa groups, respectively, as compared with the benign prostate group, of which 174 were differentially expressed in both groups. A number of the DEGs, including CAMK2D and SH3BP4, were significantly enriched in the cell cycle, and others, such as MAF, were associated with the regulation of cell proliferation. Furthermore, some DEGs (CAMK2D and PCDH17) were observed to be regulated by miR-30, whereas others (ADCY2, MAF, SH3BP4 and PCDH17) were modulated by miR-182. Additionally, ADCY2 and CAMK2D were distinctly enriched in the calcium signaling pathway. The present study identified novel DEGs, including ADCY2, CAMK2D, MAF, SH3BP4 and PCDH17, that may be involved in the metastasis of PCa.

Introduction

Prostate cancer (PCa) is the most common cancer among European and American men, and accounts for 27% (233,000) of cancer incidences in men in the USA (1). It has a high mortality rate as a result of its high propensity for metastasis (2,3). PCa has been shown to preferentially metastasize to the bone marrow stroma of the axial skeleton (4); however, the precise mechanism underlying PCa metastasis is currently unclear. Therefore, the identification of specific metastasis biomarkers and novel diagnostic targets is required in order to improve the prognosis and treatment of the disease.

Previous studies have made considerable progress in identifying the key regulators in the PCa metastatic process. E-cadherin, which is attached to the actin cytoskeleton via intracellular catenin, has been implicated in the process of PCa metastasis; in primary PCa, reduced E-cadherin expression was associated with bone metastasis and a poor prognosis (5). In addition, the expression of the DLC1 tumor-suppressor gene in metastatic PCa cells has been shown to upregulate the expression of E-cadherin, resulting in the suppression of highly metastatic PCa cell invasion by inhibiting the activity of RhoA-GTP and RhoC-GTP (6). The activation of Rho GTPases is dependent on the downstream Ras protein, which has a major influence on cell signaling (7). Members of the Rho GTPase family are involved in cancer cell motility by regulating actin dynamics and controlling morphological changes (8). A previous study demonstrated that the suppression of the farnesyl and geranyl-geranyl prenylation pathways markedly reduced the migration and motility of PCa cells by inhibiting Ras prenylation and concurrent Rho activation (9). Furthermore, activation of the phosphoinositide 3-kinase/protein kinase B (AKT) signaling pathway has been more frequently observed in resistant and metastatic PCa compared with primary PCa, and thus targeting this signaling pathway may improve the outcome of patients with aggressive PCa (10). Previous studies have reported various genes able to promote PCa tumorigenesis and metastasis, including CCL2 (11), SERPINB5 (12), SRC (13), TMPRSS2-ERG gene fusion and PCA3 (14). In addition, microRNAs (miRNAs), which are considered to be important regulators of gene expression, have been associated with the development of metastatic PCa. For instance, miR-203 (15), miR-16 (16), miR-205 (17), miR-24 (18), miR-29a (19) and miR-145 (16) have all been implicated in PCa metastasis.

Varambally et al (20) performed an integrative genomic and proteomic analysis of benign prostate and metastatic PCa; they reported 48–64% concordance between protein and transcript levels and demonstrated that proteomic alterations between metastatic and clinically localized PCa, which map concordantly to gene transcripts, can serve as predictors of clinical outcome in PCa as well as other solid tumors. However, to the best of our knowledge, the potential miRNAs involved in metastatic PCa, and the interactions of differentially-expressed genes (DEGs) targeted by miRNAs, have yet to be investigated. Therefore, the present study aimed to further elucidate the molecular mechanisms underlying the metastasis of PCa by analyzing the microarray data of benign prostate, clinically localized and metastatic PCa deposited by Varambally et al (20) in the Gene Expression Omnibus (GEO) database. Initially a hierarchical cluster analysis for DEGs was performed, followed by a Gene Ontology (GO) functional enrichment analysis. Furthermore, potential miRNAs in metastatic PCa were identified and a miRNA-DEG regulatory network was constructed. Finally, a pathway enrichment analysis for DEGs in the regulatory network was performed. The results of this bioinformatics analysis may shed light on the molecular mechanisms underlying the metastasis of PCa and provide novel diagnostic biomarkers.

Materials and methods

Affymetrix microarray data

The GSE3325 gene expression profile data (20) was downloaded from the GEO (http://www.ncbi.nlm.nih.gov/geo/) and was based on the GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array platform. A total of 19 human prostate tissue samples were available for further analysis, including seven clinically localized PCa samples, six hormone-refractory metastatic PCa samples and six benign prostate tissue samples.

CEL and probe annotation files were downloaded from GEO, and the gene expression data for all samples were preprocessed via Robust Multichip Averaging background correction, quantile normalization and probe summarization (21) in the affy software package (version 1.34.0; http://bioconductor.org/packages/release/bioc/html/affy.html), as described previously (22).

DEGs screening

The Linear Models for Microarray Data package of R (https://bioconductor.org/packages/release/bioc/html/limma.html) was used to identify genes that were differentially expressed in the primary PCa and metastatic PCa groups, as compared with the benign prostate group, as described previously (23). The raw P-value was adjusted according to the false discovery rate (FDR) using the Benjamin and Hochberg method (24). Only genes with a cut-off criteria of |log2fold change| >1 and FDR<0.01 were considered to be differentially expressed.

Hierarchical cluster analysis for DEGs

Hierarchical clustering is a common method used to determine clusters of similar data points in a multidimensional space (25). The pheatmap package (version 1.0.2; https://cran.r-project.org/web/packages/pheatmap/index.html) was used to perform hierarchical clustering of the DEGs via joint between-within distances, as described previously (26). Expression values from multiple clones or probe sets mapping to the same Unigene Cluster ID were averaged.

GO functional enrichment analysis for DEGs

The Database for Annotation, Visualization and Integrated Discovery (DAVID; https://david.ncifcrf.gov/) provides a comprehensive set of novel and powerful tools for assigning biological meaning to a set of genes (27). FDR<0.05 was used as the cut-off criterion for GO functional enrichment analysis by DAVID.

Integrated miRNA-DEG regulatory network construction

The common miRNAs in Gene set B, as predicted by the databases of miRecords (http://c1.accurascience.com/miRecords/), TarBase (http://diana.imis.athena-innovation.gr/DianaTools/index.php?r=tarbase/index) and TargetScan (http://www.targetscan.org/), were selected using WEB-based GEne SeT AnaLysis Toolkit software (update 2013; http://bioinfo.vanderbilt.edu/webgestalt/), and P<0.05 was used as the cut-off criterion. Subsequently, the Search Tool for the Retrieval of Interacting Genes (http://string-db.org/) was used to analyze the interactions between the DEGs targeted by miRNAs by calculating their combined score; a score of >0.4 was set as the cut-off criterion. Finally, the integrated miRNA-DEG regulatory network was visualized using Cytoscape (http://cytoscape.org/).

Pathway enrichment analysis for DEGs in the regulatory network

Pathway enrichment analysis was conducted as described previously (28) to identify significant metabolic pathways for the DEGs. P<0.05 was used as the cut-off criterion for the Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis using DAVID.

Results

Identification of DEGs

Based on the cut-off criteria, 2,727 DEGs were identified for the clinically localized PCa and metastatic PCa groups, of which 306 were differentially expressed in the clinically localized PCa group only (Gene set A). A total of 2,073 genes were differentially expressed in the metastatic PCa group only (Gene set B) and 174 genes were differentially expressed in both groups (Gene set C; Fig. 1), as compared with the benign prostate group.

Figure 1.

Venn diagram for the differentially expressed genes in the clinically localized and metastatic prostate cancer groups. Gene set A represents the genes only differentially expressed in the clinically localized prostate cancer group; Gene set B represents the genes only differentially expressed in the metastatic prostate cancer group; Gene set C represents the genes differentially expressed in both groups.

Hierarchical cluster analysis

An unsupervised hierarchical cluster analysis of the data revealed that the DEGs could be used to accurately classify prostate samples as benign, clinically localized prostate cancer or metastatic disease (Fig. 2).

Figure 2.

Hierarchical cluster analysis for the genes differentially expressed between the (A) metastatic prostate cancer and the benign prostate groups, and (B) the metastatic and clinically localized prostate cancer groups. Each row represents a single gene; each column represents a tissue sample. M represents the metastatic prostate cancer group; N represents the benign prostate group; P represents the clinically localized prostate cancer group. The gradual color change from orange to blue represents the changing process from upregulation to downregulation.

GO functional enrichment analysis for Gene sets A, B and C

In Gene set A, DLX2, DLX1, HOXD10 and HOXD11 DEGs were associated with proximal/distal pattern formation (FDR=3.55E-04), whereas RBP4, PDE3B and PPARG were implicated in the response to insulin (FDR=7.8400), homeostatic processes (FDR=9.6200), chemical homeostasis (FDR=0.0019) and responses to peptide hormones (FDR=0.0023) and organic substances (FDR=0.0029) (Table I).

Table I.

Enriched terms for Gene sets A, B and C.

Category	Term	No. of genes	FDR	Genes
Gene set A	GO:0009954~proximal/distal pattern formation	5	3.5500	DLX2, DLX1, GREM1, HOXD10, HOXD11
	GO:0032868~response to insulin stimulus	8	7.8400	RBP4, EIF4EBP1, FADS1, PPARG, PDE3B, STXBP4, GAL, VLDLR
	GO:0042592~homeostatic process	24	9.6200	RBP4, SLC12A2, PPARG, F2RL1, PRDX4, PDE3B, CACNG2, ITPR3, PPARGC1A, MUC6…
	GO:0001501~skeletal system development	14	0.0011	RBP4, HOXD10, HOXD11, MSX2, DLX2, DLX1, COL9A2, BCL2, CLEC3A, NAB1…
	GO:0048878~chemical homeostasis	18	0.0019	RBP4, F2RL1, NOX1, PPARG, PDE3B, PPARGC1A, PRKCB, CCL11, MALL, ATP7B…
	GO:0021877~forebrain neuron fate commitment	3	0.0022	DLX2, DLX1, LHX6
	GO:0043434~response to peptide hormone stimulus	9	0.0023	RBP4, EIF4EBP1, FADS1, BCL2, PPARG, PDE3B, STXBP4, GAL, VLDLR
	GO:0010033~response to organic substance	22	0.0029	RBP4, ADCY1, GNRH1, FADS1, LOC646626, PPARG, PTGS1, PDE3B, COLEC12, STXBP4…
	GO:0009725~response to hormone stimulus	14	0.0038	RBP4, ADCY1, GNRH1, FADS1, PTGS1, PPARG, PDE3B, STXBP4, GAL, EIF4EBP1…
	GO:0034637~cellular carbohydrate biosynthetic process	6	60.0038	RBP4, ISYNA1, UAP1, GNE, PPARGC1A, ACN9
Gene set B	GO:0022402~cell cycle process	102	5.2300	PRC1, ZAK, AIF1, BTRC, CDCA8, CDC6, CENPF, PTTG1, AURKB, TGFB2…
	GO:0051726~regulation of cell cycle	68	1.1100	E2F2, PTGS2, ZAK, FAM175A, PKMYT1, PDCD4, PTEN, GTSE1, TGFB2, MYC…
	GO:0007049~cell cycle	128	1.8200	ZAK, PRC1, AIF1, BTRC, PKMYT1, RBM7, AURKA, AURKB, PTTG1, TGFB2…
	GO:0051301~cell division	61	4.6100	PRC1, PTTG1, CCNE1, CDCA2, CDC6, CABLES2, CDCA5, CCNA2, ASPM, CDK1…
	GO:0022403~cell cycle phase	78	4.7700	E2F1, PRC1, PKMYT1, RBM7, AURKA, AURKB, PTTG1, GTSE1, CCNE1, CDCA8…
	GO:0010035~response to inorganic substance	47	6.4600	CAV1, GCLC, PTGS2, PDGFA, SNCA, TPM1, PTEN, KCNMB1, FOS, GSN…
	GO:0007346~regulation of mitotic cell cycle	38	0.0011	CAV2, HOXA13, PML, PKMYT1, ASNS, ANLN, ZNF655, RCC1, SCRIB, MYC…
	GO:0042127~regulation of cell proliferation	126	0.0012	HRAS, CD38, IL6ST, PDGFA, TP63, MAF, TGFB3, STRN, PNP, TGFB2…
	GO:0030030~cell projection organization	70	0.0015	CAV2, HOXA13, PML, PKMYT1, ANLN, ZNF655, RCC1, SCRIB, GTSE1, MYC…
	GO:0000278~mitotic cell cycle	70	0.0018	E2F1, PRC1, BTRC, PKMYT1, AURKA, AURKB, PTTG1, GTSE1, CCNE1, NDE1…
Gene set C	GO:0022402~cell cycle process	21	3.3000	MKI67, DLGAP5, SGOL1, NUSAP1, BIRC5, PBK, CDKN3, CCNB1, CENPA, CAMK2D…
	GO:0022403~cell cycle phase	18	3.5700	MKI67, DLGAP5, SGOL1, NUSAP1, TTK, BIRC5, PBK, CCNB1, CAMK2D, ID4…
	GO:0000278~mitotic cell cycle	17	4.0700	DLGAP5, SGOL1, NUSAP1, TTK, BIRC5, PBK, CDKN3, CCNB1, CENPA, CAMK2D…
	GO:0000279~M phase	15	2.7400	MKI67, DLGAP5, SGOL1, NUSAP1, TTK, BIRC5, PBK, UBE2C, CCNB1, KIF2C…
	GO:0007049~cell cycle	22	1.1800	DLGAP5, SGOL1, NUSAP1, BIRC5, PBK, CCNB1, SH3BP4, KIF2C, CENPA, CAMK2D…
	GO:0000280~nuclear division	11	4.5000	CCNB1, KIF2C, CCNB2, DLGAP5, SGOL1, NUSAP1, BIRC5, PBK, UBE2C, ERCC6L…
	GO:0007067~mitosis	11	4.5000	CCNB1, KIF2C, CCNB2, DLGAP5, SGOL1, NUSAP1, BIRC5, PBK, UBE2C, ERCC6L…
	GO:0000087~M phase of mitotic cell cycle	11	5.2300	CCNB1, KIF2C, CCNB2, DLGAP5, SGOL1, NUSAP1, BIRC5, PBK, UBE2C, ERCC6L…
	GO:0048285~organelle fission	11	6.3000	CCNB1, KIF2C, CCNB2, DLGAP5, SGOL1, NUSAP1, BIRC5, PBK, UBE2C, ERCC6L…
	GO:0007346~regulation of mitotic cell cycle	9	9.6800	DLGAP5, CAMK2D, NUSAP1, TTK, BIRC5, AFAP1L2, GAS1, UBE2C, MYC

Gene set A represents the genes only differentially expressed in the clinically localized prostate cancer group; Gene set B represents the genes only differentially expressed in the metastatic prostate cancer group; Gene set C represents the genes differentially expressed in both groups. FDR, false discovery rate.

In Gene set B, the DEGs were predominantly associated with the cell cycle: PRC1, ZAK, PTTG1, TGFB2, CDCA8, CDC6 and CENPF were associated with the cell cycle process (FDR=5.2300); PRC1, PTTG1, CCNE1, CDCA2 and CDC6 were involved in cell division (FDR=4.6100); and HRAS, CD38, IL6ST, PDGFA, TP63, MAF and TGFB3 were associated with the regulation of cell proliferation (FDR=0.0012) (Table I).

In Gene set C, the DEGs were also predominantly associated with the cell cycle. DLGAP5, SGOL1, NUSAP1, PBK, BIRC5 and CCNB1 were associated with the cell cycle process (FDR=3.3000), M phase (FDR=2.7400), mitosis (FDR=4.5000) and organelle fission (FDR=6.3000), whereas SH3BP4, KIF2C, CCNB2, CENPA and CAMK2D were associated with the cell cycle only (FDR=1.1800) (Table I).

Analysis of the miRNA-DEG regulatory network

A total of 10 miRNAs were identified in Gene set B, including miR-374, miR-128, miR-182, miR-30, miR-302c and miR-524. Notably, miR-30 targeted the majority of the DEGs (11 DEGs, including CAMK2D, PCDH17, EDNRB, KCNJ3 and SOX4), and miR-182 targeted seven DEGs, including EDNRB, MAF, ADCY2, PCDH17, RET, SH3BP4 and BCL11A (Table II).

Table II.

Enriched microRNAs in Gene set B.

microRNA	P-value	Count	Genes targeted by microRNA
hsa_TATTATA, MIR-374	2.1100	10	RORB, HOMER1, KIF20A, SYBU, DACH1, GATA3, ARHGAP28, AFAP1L2, SOX4, SYT1
hsa_CACTGTG, MIR-128	0.0003	9	RORB, FBLN2, ADCY2, ACOT11, INSM1, SYT1, FOXQ1, MME, BCL11A
hsa_ATGCAGT, MIR-217	0.0003	6	STX1A, MAF, PCDH17, DACH1, EZH2, BCL11A
hsa_TGTTTAC, MIR-30	0.0005	11	SOBP, CAMK2D, COL13A1, SLC36A1, PCDH17, AFAP1L2, EDNRB, KCNJ3, SOX4, MATR3……
hsa_ACAACTT, MIR-382	0.0032	4	NDRG2, SYT1, MATR3, DACH1
hsa_ACTGCCT, MIR-34B	0.0032	6	INSM1, SOX4, MYC, ADCY2, PIEZO2, JAKMIP1
hsa_TTGCCAA, MIR-182	0.0036	7	EDNRB, MAF, ADCY2, PCDH17, RET, SH3BP4, BCL11A
hsa_CTTTGTA, MIR-524	0.0036	8	SOBP, CTHRC1, PCDH17, ECT2, ID4, RCAN2, HOXD13, SOX4
hsa_ATGTTAA, MIR-302C	0.0038	6	EDNRB, SALL3, MAF, MATR3, DACH1, BCL11A
hsa_TGCACTT, MIR-519	0.0038	8	SOBP, RORB, SYBU, NETO2, SOX4, SYT1, APCDD1, JAKMIP1

Count represents the number of differentially-expressed genes targeted by microRNA. Gene set B represents the genes only differentially expressed in the metastatic prostate cancer group.

The miRNA-DEG regulatory network in Fig. 3 contained 10 miRNAs and 43 corresponding DEGs. ADCY2 was regulated by miR-128, miR-34B and miR-182; EDNRB was regulated by miR-30, miR-182 and miR-302C; CAMK2D was regulated by miR-30; PCDH17 was modulated by miR-217, miR-30, miR-182 and miR-524; SH3BP4 was modulated by miR-182; and MAF interacted with miR-182, miR-302c and BCL11A.

Figure 3.

Regulatory network containing microRNAs and their corresponding differentially expressed genes for metastatic prostate cancer. Dark grey nodes represent upregulated genes; light grey nodes represent downregulated genes; and diamonds represent microRNAs.

Pathway enrichment analysis for the DEGs in the regulatory network

The DEGs in the regulatory network were enriched in two pathways, including the calcium signaling pathway (EDNRB, ADCY2 and CAMK2D), and thyroid cancer (RET and MYC; Table III).

Table III.

Enriched pathways for the differentially-expressed genes in the regulatory network.

Term	Description	Count	P-value	Genes
hsa04020	Calcium signaling pathway	3	0.02231	EDNRB, ADCY2, CAMK2D
hsa05216	Thyroid cancer	2	0.03927	RET, MYC

Discussion

The present study identified 306 and 2,073 genes that were differentially expressed in the clinically localized PCa group and the metastatic PCa group, respectively, as compared with the benign prostate group. Of these, 174 genes were differentially expressed in both the clinically localized PCa and metastatic PCa groups.

ADCY2, which encodes adenylate cyclase 2, and CAMK2D, which encodes calcium/calmodulin-dependent protein kinase II δ (29,30), were shown to be enriched in the calcium signaling pathway. Metastasis is the predominant cause of mortality in patients with PCa, and Ca2+ is a crucial regulator of cell migration (31). Elevated intracellular concentrations of Ca2+ may facilitate the metastasis of PCa by triggering the activation of the Akt signaling pathway and promoting PCa cell (PC3) attachment (32). CAMK2D encodes components of the Wnt/β-catenin-signaling pathway, the inhibition of which delays metastatic PCa cell cycle progression and proliferation (33). In the present study, CAMK2D was associated with the cell cycle, which is known to be a critical event in tumor growth and metastasis (34). Furthermore, CAMK2D was observed to be regulated by miR-30. As a tumor suppressor, miR-30 has been shown to be downregulated by oncogenic signals, such as hepatocyte growth factor and epidermal growth factor, in PCa samples (35), and overexpression of miR-30 in PCa cells was reported to suppress the epithelial-to-mesenchymal transition and inhibit cell migration and invasion (36).

ADCY2 was observed to be modulated by miR-182. A previous study demonstrated that ectopic expression of miR-182 in PC3 significantly reduced protein expression levels of GNA13, GNA13-3′-untranslated region (UTR)-reporter activity and extracorporeal invasion of these cells (37). In addition, aberrant overexpression of miR-182 was shown to promote the proliferation, increase the invasion, facilitate the G1/S cell cycle transition and reduce early apoptosis of PC3 cells; and, miR-182 was able to suppress the expression of the NDRG1 tumor suppressor gene by directly targeting the NDRG1 3′-UTR (38). Therefore, CAMK2D and ADCY2 may be involved in the metastasis of PCa via calcium signaling and regulation by miR-30 and miR-182, respectively.

MAF, which was also modulated by miR-182 in the present study, was associated with the regulation of cell proliferation. MAF acts as a macrophage-activating factor and is generated from a precursor protein termed the Gc protein (39). Deglycosylation of the Gc protein prevents its conversion to MAF, inhibiting macrophage activation and resulting in immunosuppression (40). In a previous study, patients with metastatic PCa were administered Gc protein with MAF precursor activity (100 ng/week), and were shown to have serum activity levels of Nagalase equivalent to those of healthy controls, thus suggesting that these patients were tumor-free (41). Furthermore, MAF expression has been associated with the receptor tyrosine kinase, platelet-derived growth factor receptor (PDGFR)-β status (42). In the miRNA-DEG regulatory network, MAF was also modulated by miR-302c, and it has been reported that miR-302c is downregulated in clinical PCa samples (43). In addition, MAF interacted with BCL11A, which was observed to be upregulated in PC3 holoclones (44). Therefore, MAF may have an important role in the metastasis of PCa by interacting with miR-182, miR-302c and BCL11A.

In the present study, the downregulated DEG SH3BP4 was shown to be associated with the cell cycle and was also regulated by miR-182. SH3BP4 encodes SH3-domain binding protein 4 (45). SH3 domains are found in a variety of proteins, including tyrosine kinases, such as Abl and Src, and are involved in endocytosis, intracellular sorting and the cell cycle (46). Another downregulated DEG PCDH17, which encodes protocadherin 17, was shown to interact with miR-182 and miR-30. PCDH17 methylation is a common tumor-specific event in PCa and has been associated with a shorter biochemical recurrence-free survival rate and a reduced overall survival rate of patients with PCa following a radical prostatectomy (47). Therefore, SH3BP4 and PCDH17 may be responsible for the metastasis of PCa via their interactions with miR-182 and/or miR-30. Furthermore, miR-374 was significantly enriched in Gene set B. Previous studies have reported that miR-374 is markedly downregulated in PCa (48,49). Furthermore, miR-374b, which is a subtype of miR-374, has been shown to be downregulated in prostate fluid or serum samples from prostate cancer patients, and thus may serve as a PCa biomarker in clinical diagnosis (50).

In conclusion, the present study identified numerous important DEGs, including ADCY2, CAMK2D, MAF, SH3BP4 and PCDH17, that may be involved in the metastasis of PCa. However, the results of the present study require validation by further experiments, and the molecular mechanisms underlying metastatic PCa require further investigation.