Original Research

Comparative Expression Analysis Reveals Relationships between SPINK1, TUBB3, and EZH2, and Prostate Cancer Molecular Biomarkers in the Cancer Genome Atlas (TCGA) Data

Carolina Saldana1,2, Myriam Kossai3, Guillaume Ploussard4,5, Salem Chouaib6 and Stéphane Terry6*
1Department of Medical Oncology, Henri Mondor Hospital (AP-HP), France
2University of Paris-Est, France
3Department of Pathology, Gustave Roussy Cancer Campus, France
4University Institute of Cancer Toulouse Oncopole, France
5Department of Urology, Clinique Saint Jean Languedoc, France
6INSERM U1186, Gustave Roussy Cancer Campus, France

*Corresponding author: Stéphane Terry, INSERM U1186, Team Labeled For The League Against Cancer, Gustave Roussy Cancer Campus, 114 Rue Edouard Vaillant, 94800 Villejuif, France

Published: 21 Oct, 2016
Cite this article as: Saldana C, Kossai M, Ploussard G, Chouaib S, Terry S. Comparative Expression Analysis Reveals Relationships between SPINK1, TUBB3, and EZH2, and Prostate Cancer Molecular Biomarkers in the Cancer Genome Atlas (TCGA) Data. Clin Oncol. 2016; 1: 1128.


Background: Despite the recent discovery of molecular subtypes in prostate cancer (PCa) expressing or not gene fusions involving E26 Transformation-Specific (ETS) transcription factors, including ERG (for v-ets avian erythroblastosis virus E26 oncogene homolog), little is known on molecular alterations associated, and cooperative events at play during initiation and progression of PCa.
Objective and methods: Using RNA-Seq data from The Cancer Genome Atlas (TCGA) collection of surgically managed primary prostate adenocarcinomas, we investigated the relations between gene expression of the candidate prognostic markers SPINK1, TUBB3 (class III beta-tubulin), EZH2, and known PCa molecular markers. 484 cases were included in the analysis.
Results: Clustering analysis consistently showed TUBB3 associating with EZH2, and SPINK1 with PTEN and TFF3, but not with ERG, ETV1, CHD1, AR or SPOP expression. Positive and negative correlations were found among these PCa markers. Notably, in tumors highly expressing SPINK1 or TUBB3, a subset of cases showed substantial EZH2 expression, while EZH2 expression was highly correlated with AURKA expression (r=0.7178; p <0.0001), an oncogenic target in cancer. Interestingly, we found that high expression of EZH2 was strongly associated with reduced SPOP expression (r=-0,455; p <0.0001). Moreover, tumors expressing SPINK1 and TUBB3 often appeared to have reduced expression of RB1, AR, and REST as possible signs of neuroendocrine differentiation.
Conclusions: Despite substantial heterogeneity among the PCa cases, the current study suggests that significant associations and overlaps exist between PCa molecular alterations and expression of candidate PCa prognostic markers. A better understanding of these alterations and their cooperative role should help refine PCa subtypes, identify aggressive subgroups among those, and improve PCa management and therapy response.


Prostate cancer (PCa), as many other cancers, is characterized by extensive clinical and molecular heterogeneity [1]. Over the past 10 years, with the advent of high throughout methods, our understanding of the PCa genome has significantly changed, while revealing considerable intertumor (between tumors of the same type), and intra-tumor (within tumors, different subclones) heterogeneities [2-8]. Based on the molecular alterations identified, different molecular PCa subclasses or subtypes have emerged with the attempt to correlate those PCa subtypes to clinical features, disease progression and response to therapy. Approximately 50% of PCAs harbor a gene fusion between ERG, an ETS transcription factor, and an androgen-regulated gene (TMPRSS2 ~ 90%, SLC45A3, NDRG1, HERPUD1, or others <10%) [9,10]. ERG expression is routinely used as a surrogate marker of these alterations [11]. Trefoil factor 3 (TFF3) represents a highly specific molecular biomarker of cancer in the prostate. Detectable in 40% to 60% of PCa cases [12,13], it appeared to be inversely correlated with ERG expression in most instances [14] (Figure 1). Inactivation of tumor suppressor phosphatase and tensin homolog (PTEN) is also commonly found in PCa and could be associated with cancer progression [15,16]. Several studies have found that PTEN alterations are enriched in ERG-over expressing PCa. Moreover, ERG overexpression and alterations of PTEN could cooperate, leading to more aggressive disease [17,18]. Despite these significant advances, it remains a challenge to link this molecular classification with clinical features to improve prognostic estimation, and treatment decisions in routine clinical practice [12,13,19-27].
SPINK1 (previously referred to as TATI, or tumor-associated trypsin inhibitor) is expressed in various diseases including cancer [28]. Tomlins et al. [29] identified SPINK1 as a candidate marker for a group of PCa devoid of ETS gene fusions associated with aggressive disease features and adverse outcomes. In other studies, a correlation between SPINK1 expression and adverse prognosis was not observed [25,30,31]. Nevertheless, in a recent survey, the prognostic value of SPINK1 was confirmed in a well-annotated cohort [14]. We proposed that SPINK1 overexpression emerges from a subgroup of PCa with ERG negative/TFF3 (trefoil factor 3) positive pattern [14] (Figure 1).
It is to note that various experimental studies have shown that ERG, TFF3 and SPINK1 are all associated with increased cell motility and/or invasive behavior in PCa models supporting their role in PCa progression [29,32,33].
Elevated βIII-tubulin (encoded from TUBB3 gene) expression was previously identified as significantly associated with tumor aggressiveness in PCa patients with presumed localized disease [34]. In this study, βIII-tubulin expression was found to be an independent marker of disease recurrence after local treatment. Recently, Tsourlakis and colleagues examined a large European cohort and confirmed this finding [35]. Additionally, increased βIII-tubulin expression is associated with the emergence of Castrate Resistant PCa (CRPC) [36,37], and with lower survival for patients receiving docetaxel-based chemotherapy [34].
The polycomb group protein enhancer of zeste homolog 2 (EZH2) is known to be increased in metastatic PCas. Clinically localized PCa that express elevated levels of EZH2 show a poorer prognosis [38], suggesting that EZH2 has a potential role in disease progression and patient prognosis [39-41]. Moreover EZH2 expression is found elevated in Neuroendocrine PCa, a higly aggressive form of human PCa [42,43]. In the era of precision medicine [44,45], these findings underscore the potential utility of decrypting the relationships at play between SPINK1, TUBB3, EZH2 expression and other PCa molecular alterations in order to improve our definition of molecular PCa subclasses and find best therapeutic solutions for the management of patients.
In the present work, we examined publically available gene expression data from primary PCa cases of The Cancer Genome Atlas (TCGA), and studied relationships between gene expression of TUBB3, SPINK1, EZH2 and expression of other known molecular PCa biomarkers including ERG, ETV1, PTEN, CHD1, TFF3, MYC, RB1, MYC, AURKA, and SPOP [2,3,6,7,42, 46-49].

Figure 1

Another alt text

Figure 1
Schematic representation of the PCa molecular subtypes deriving from previous studies. Tumors without or with ETS fusions are shown on the left and right, respectively. In the ETS positive tumors, the vast majority harbor genomic rearrangements leading to ERG overexpression, thus representing an ERG positive PCa subtype. PTEN loss is commonly found in this subtype. ETS negative PCas generally express TFF3 as a molecular cancer biomarker, and a subgroup of these PCas express SPINK1. This subgroup may reflect a subset of disease with more aggressive behavior.


RNA-Seq gene expression analysis, clustering and statistical analysis
Human samples analyzed consisted of primary of prostate adenocarcinomas from The Cancer Genome Atlas (TCGA) project collection (http://cancergenome.nih.gov/). TCGA RNA-Seq expression data and sample information were accessed before June 2016 from cBioPortal [50] and the TCGA public access data (http:// tcga-data.nci.nih.gov/). Only cases with available expression data, and analyzed for mutational landscape were considered. The cohort consisted of men surgically managed for localized or locally advanced disease. Of note, about 16% (41 of 260 cases; NA for the remaining cases) also received adjuvant treatments consisting of hormone therapy, radiotherapy, or a combination of those. Available patient cohort characteristics are shown in Table 1 (n=484).
To explore expression levels and associations of the different genes, gene expression levels (RSEM) were subjected to correlation and unsupervised clustering analyses using Cluster and TreeView softwares after transforming the RSEM into Log2 (RSEM+1). Genes analyzed included ERG, ETV1, PTEN, CHD1, TFF3, MYC, SPOP, SPINK1, EZH2, TUBB3, RB1, and AURKA. Pearson's coefficient was determined to assess correlations between expression levels. For differential expression analysis, an unpaired t test or a nonparametric Mann-Whitney U test was applied as appropriate. All p values were two-sided and values of p <0.05 were considered statistically significant.

Table 1

Another alt text

Table 1
Clinico-pathological characteristics of the TCGA studied cohort (n=484).

Figure 2

Another alt text

Figure 2
Hierarchical cluster analysis and heatmap generated using SPINK1, ERG, TFF3, ETV1, AR, CHD1, PTEN, SPOP, ERG, and EZH2 expressions across TCGA prostate adenocarcinomas (n=484). In the heatmap, each column represents a different case, and each row represents a marker. blue to red: lowest to highest expression. PCa cases are ordered with respect to expression of SPINK1 (A), ERG (B), TUBB3 (C), or EZH2 (D).


In the group of ETS negative PCa, SPINK1 expression is positively correlated with TFF3 and PTEN expression levels
Expression data were retrieved from TCGA collection, tumor samples ordered by SPINK1 expression, and clustering analysis was performed for gene expression of PCa molecular biomarkers including ERG, TFF3, ETV1, AR, CHD1, PTEN, SPOP, ERG, and EZH2 (Figure 2a). Expectedly, SPINK1 expression clustered with TFF3 expression, and seems inversely correlated with ERG and ETV1 expression. SPINK1 and TFF3 also clustered with PTEN expression. PTEN expression appeared to be especially elevated in a number of cases expressing high levels of SPINK1. By contrast, it was reduced in the cases expressing high levels of ERG, as also evidenced by an additional heatmap with tumor classified with respect to ERG expression (Figure 2b).
To substantiate these results, we computed correlation scores between SPINK1 expression and each molecular marker (Table 2). SPINK1 and TFF3 were significantly correlated (r=0.36, p <0.0001) and both inversely correlated with ERG, CHD1, and AR (Table 2). Moreover, ERG negatively correlated with PTEN (r=-0.2911; p <0.0001). This likely reflects an enrichment of PTEN deletion in ERG+PCa subtype as described previously [17,51,52].
A subset of tumors expressing SPINK1 concomitantly expresses high levels of EZH2
Interestingly, we also noted in the heatmap a relative enrichment of EZH2 expression in tumors expressing high levels of SPINK1 (Figure 2a), and based on the correlation analysis, EZH2 expression was positively correlated with SPINK1 expression (r=0.1554, p=0.0006). This data suggests that at least a subset of SPINK1 expressing tumors also expresses high levels of EZH2. Importantly, there was however no correlation between EZH2 and TFF3 (Table 2), neither with ERG. We previously described the presence of SPINK1 expression characterizes an aggressive subtype in the group of ERG-/TFF3+PCa tumors [14]. Our observation here suggests that EZH2 overexpression preferentially arises from ERG-/TFF3+/ SPINK1+PCas rather than in ERG-/TFF3+/SPINK1-PCas, which could coincide with more aggressive forms of the disease.
A subset of tumors expressing TUBB3 concomitantly expresses high levels of EZH2
We then investigated how these markers could cluster with TUBB3 (encoding for Class III β-tubulin), another candidate marker for aggressive PCa disease [34], also assumed to be an early marker for reduced AR signaling [36] and NE differentiation [37,53]. A Heatmap of the same set of genes in tumors classified by TUBB3 expression revealed a marked enrichment of EZH2 in tumors overexpressing TUBB3 (Figure 2c), further highlighted by a positive correlation coefficient (r=0.32; p <0001; Table 2). The analysis also revealed TUBB3 and PTEN expression patterns inversely correlated (r=-0.3159; p <0001), with AR and SPOP also following this trend (r=-0.26 and -0.20, respectively; p<0001). Intriguingly, SPINK1 expression did not appear to be associated with TUBB3 expression. SPINK1 expression was correlated with PTEN expression, when TUBB3 expression anticorrelated with PTEN expression. Moreover, TUBB3 expression seemed to be more closely associated with EZH2 than SPINK1. This indicates that SPINK1 or TUBB3 expressions may be mutually exclusive under some circumstances; thus representing two distinct subsets of disease. Another possibility might be that a large proportion of TUBB3 expressing tumors is confined to the ERG+ / PTENlow PCa subtype, that is negative for SPINK1, while a smaller fraction is linked to ETS–TFF3+SPINK1 PCa subtype, and both can exhibit an enriched expression of EZH2.
EZH2 expression is associated with AURKA expression and SPOP alterations
We then generated a heatmap classifying tumors with respect to EZH2 expression (Figure 2d), in conjunction with correlation analysis. This denoted a striking positive correlation between EZH2 and AURKA levels (r=0.7178; p <0.0001), and an inverse correlation between EZH2 and SPOP levels (r=-0.455; p <0001; Table 2). The connection between EZH2 expression and downregulated levels of SPOP is intriguing, especially considering recent work by Barbieri and colleagues who identified inactivating mutation in SPOP gene as the most common point mutation in PCa [6]. Further work by this group revealed that this mutation occurs predominantly in the group of ERG rearranged PCa tumors [6], and is concomitant with deletions at 5q21 CDH1 locus [47]. We then sought to determine whether mutation in SPOP gene, is associated with varying levels of SPINK1, EZH2 and TUBB3 (Figure 3). When considering all patients, SPOP mutant cases had elevated expression of EZH2, SPINK1 and TFF3, but reduction in expression of ERG and CHD1. There were no significant changes noted for TUBB3 and SPOP expression levels.
SPINK1, TUBB3 and EZH2 are associated with various NE features
Previous studies have reported associations of TUBB3 and EZH2 overexpression with aggressive features in localized PCa, or NE features in castrate resistant CRPC tumors [34-36,42]. We asked weather SPINK1, TUBB3, EZH2 could also associate with NE features in this cohort of locally managed PCa tumors. Heatmaps were generated with respect to TUBB3, EZH2, or SPINK1 expression, and their distribution was studied as above among a panel of NE markers, putative drivers and suppressors of NE phenotype, (NE suppressors (RB1, REST, AR); NE drivers (AURKA, SRRM4, MYCN); NE markers (SYP (synpatophysin), ENO2 (NSE), CHGA (chromogranin A), CHGB (chromogranin B)) (Figure 4). Correlation coefficients were determined as above to assess associations between variables. Notably, EZH2 highly correlated with AURKA gene expression, but with the exception of TUBB3, showed no or negative correlation with other NE components. By contrast, tumors with high expression of TUBB3, and SPINK1 to a lesser extent, more frequently exhibited NE features, as judged by anti-correlations with REST, AR and RB1, and positive correlation with SYP (Table 3).

Table 2

Another alt text

Table 2

Figure 3

Another alt text

Figure 3
The effect of SPOP mutation on expression of SPINK1, TFF3, ERG, TUBB3, EZH2, CHD1, PTEN, and SPOP accross 484 PCa samples. Box plots showing the Median, 25th to 75th percentiles. Lower and upper bars correspond to the minimum and maximum values, respectively.

Table 3

Another alt text

Table 3

Figure 4

Another alt text

Figure 4
Hierarchical cluster analysis and heatmap generated using expression levels of SPINK1, TUBB3, EZH2, and NE components AR, SYP, RB1, CHGA, CHGB, ENO2, MYCN, AURKA, REST, SRRM4 across TCGA prostate adenocarcinomas (n=484). In the heatmap, each column represents a different case, and each row represents a marker. blue to red: lowest to highest expression. PCa cases are ordered with respect to expression of TUBB3 (A), EZH2 (B), or SPINK1 (C).


We previously proposed SPINK1 and βIII-tubulin expressions as independent prognosticators of disease recurrence in PCa patients primarily managed by prostatectomy [14,34]. Together with EZH2, another potential marker of PCa aggressiveness, these genes may be directly involved in progression, metastatic spread and/or therapy resistance of PCa. One interesting open question regarding these genes is to what extent these genes cooperate or overlap with other known molecular alterations recently characterized in PCa and defining PCa subclasses [1-3,6,7,42,46-49, 54].
In this work, by exploring RNAseq data from the TCGA prostate adenocarciomas, we confirmed on a large series of primary PCas that SPINK1 positive tumors represent a molecular subgroup of PCa tumors strongly associated with TFF3 expression, and correlating negatively with ERG expression. We found that these tumors express PTEN more often, but less CHD1 or AR. Importantly, a subset of those cases seem to overexpress EZH2. Tumors highly expressing TUBB3 also frequently exhibited higher expression of EZH2. Correlation analyses also revealed that EZH2 expression was positively associated with AURKA expression, an oncogenic target in cancer, while it was negatively associated with SPOP expression, a new putative tumor suppressor in PCa that is frequently mutated [1,6,23]. It is tempting to speculate that a molecular link exists between SPOP alterations, EZH2 and AURKA expression. In line with this possibility, our preliminary data already indicate, that in the group of SPINK1 high expressing PCa cases, SPOP mutants displayed higher expression of EZH2 and AURKA compared to SPOP wild-type (data not shown).
Altogether these findings should help refine PCa molecular subtypes, and identify subgroups of aggressive PCa. A working model of the different subgroups is presented in Figure 5.
It remains unclear however weather SPOP alterations influences the group of TUBB3 high PCa. Aside from the apparent relations between TUBB3, EZH2 and AURKA, SPOP expression was only slightly reduced in TUBB3 high PCa, and we did not find relationship between TUBB3 expression and SPOP mutation status. We posit that in the groups of SPINK1 high or TUBB3 high PCa tumors, also characterized by high vs. low expression of PTEN, respectively, a subset of cases express significant levels of EZH2 accompanied by substantial AURKA expression which could evoke more aggressive features. A thorough assessment of such hypotheses will require further investigations on independent cohorts, and validation by various techniques including Fluorescence in Situ Hybridization (FISH) and immunohistochemistry-based approaches. Our data investigating NE features in this series indicates that EZH2 is not directly associated with NE differentiation in this disease stage. Hence, EZH2 is unlikely to be a driver of NE differentiation in primary tumors. However, its concurrent expression with SPINK1 or TUBB3 in some circumstances could be linked to the emergence of NE features. Of therapeutic relevance, many inhibitors directed against AURKA and EZH2 has been developed these recent years [55-57]. Thus, if this hypothesis is confirmed, this could provide a biological rationale for testing the effect of such new-targeted therapies to treat these PCa subgroups. In addition, because EZH2 and/or AURKA upregulation might represent two key events during the transformation of prostate adenocarcinoma towards Neuroendocrine PCa [42,43,58], one could consider targeting these components at an early stage in order to prevent NEPC development and its progression [43,59].

Figure 5

Another alt text

Figure 5
Schematic representation of the PCa molecular subgroups deriving from this study and from previous studies. Tumors without or with ETS fusions are shown on the left and right, respectively. In the ERG positive PCa subtype, PTEN loss is a commonly found, while it is relatively rare in ETS negative (TFF3+) subtype in which reduction of CHD1 and inactivating mutations in SPOP become more common features. In each PCa subtype, other alterations such as TUBB3, SPINK1, or EZH2 overexpression working in parallel or together (likely in association with additional related events such as AURKA upregulation and SPOP downregulation) may characterize subsets of disease with more aggressive behavior, resistant to therapies, or being able to proliferate or progress more rapidly to metastatic disease. Other important alterations, that are not shown here, are likely involved in initiation, progression, or differentiation of the disease. This includes, but not only, MYC amplification, ETV1 amplification/overexpression; mutations in TP53, CTNNB1, ATM, BRCA2, or FOXA1; deletion or reduction of NKX3.1, RB1, AR, REST.


We are thankful to many colleagues for continuous encouraging discussions and give particular thanks to Yves Allory, Francis Vacherot, Christophe Tournigand, Alexandre de la Taille, and Nathalie Nicolaiew. We apologize to the authors whose work could not be cited. We declare no competing financial interests.


  1. Shoag J, Barbieri CE. Clinical variability and molecular heterogeneity in prostate cancer. Asian J Androl. 2016; 18: 543-548.
  2. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, et al. Integrative genomic profiling of human prostate cancer. Cancer Cell. 2010; 18: 11-22.
  3. Brenner JC, Chinnaiyan AM. Disruptive events in the life of prostate cancer. Cancer Cell. 2011; 19: 301-303.
  4. Pflueger D, Terry S, Sboner A, Habegger L, Esgueva R, Lin PC, et al. Discovery of non-ETS gene fusions in human prostate cancer using next-generation RNA sequencing. Genome Res. 2011; 21: 56-67.
  5. Berger MF, Lawrence MS, Demichelis F, Drier Y, Cibulskis K, Sivachenko AY, et al. The genomic complexity of primary human prostate cancer. Nature. 2011; 470: 214-220.
  6. Barbieri CE, Baca SC, Lawrence MS, Demichelis F, Blattner M, Theurillat JP, et al. Exome sequencing identifies recurrent SPOP, FOXA1 and MED12 mutations in prostate cancer. Nat Genet. 2012; 44: 685-689.
  7. Network TCGA. The Molecular Taxonomy of Primary Prostate Cancer. Cell. 2015; 163: 1011-1025.
  8. Svensson MA, LaFargue CJ, MacDonald TY, Pflueger D, Kitabayashi N, Santa-Cruz AM, et al. Testing mutual exclusivity of ETS rearranged prostate cancer. Lab Invest. 2011; 91: 404-412.
  9. Tomlins SA, Bjartell A, Chinnaiyan AM, Jenster G, Nam RK, Rubin MA, et al. ETS gene fusions in prostate cancer: from discovery to daily clinical practice. Eur Urol. 2009; 56: 275-286.
  10. Rubin MA, Maher CA, Chinnaiyan AM. Common gene rearrangements in prostate cancer. J Clin Oncol. 2011; 29: 3659-3668.
  11. Park K, Tomlins SA, Mudaliar KM, Chiu YL, Esgueva R, Mehra R, et al. Antibody-based detection of ERG rearrangement-positive prostate cancer. Neoplasia. 2010; 12: 590-598.
  12. Garraway IP, Seligson D, Said J, Horvath S, Reiter RE. Trefoil factor 3 is overexpressed in human prostate cancer. Prostate. 2004; 61: 209-214.
  13. Faith DA, Isaacs WB, Morgan JD, Fedor HL, Hicks JL, Mangold LA, et al. Trefoil factor 3 overexpression in prostatic carcinoma: prognostic importance using tissue microarrays. Prostate. 2004; 61: 215-227.
  14. Terry S, Nicolaiew N, Basset V, Semprez F, Soyeux P, Maille P, et al. Clinical value of ERG, TFF3, and SPINK1 for molecular subtyping of prostate cancer. Cancer. 2015; 121: 1422-1430.
  15. Saal LH, Johansson P, Holm K, Gruvberger-Saal SK, She QB, Maurer M, et al. Poor prognosis in carcinoma is associated with a gene expression signature of aberrant PTEN tumor suppressor pathway activity. Proc Natl Acad Sci U S A. 2007; 104: 7564-7569.
  16. Shen MM, Abate-Shen C. Molecular genetics of prostate cancer: new prospects for old challenges. Genes Dev. 2010; 24: 1967-2000.
  17. Carver BS, Tran J, Gopalan A, Chen Z, Shaikh S, Carracedo A, et al. Aberrant ERG expression cooperates with loss of PTEN to promote cancer progression in the prostate. Nat Genet. 2009; 41: 619-624.
  18. King JC, Xu J, Wongvipat J, Hieronymus H, Carver BS, Leung DH, et al. Cooperativity of TMPRSS2-ERG with PI3-kinase pathway activation in prostate oncogenesis. Nat Genet. 2009; 41: 524-526.
  19. Demichelis F, Fall K, Perner S, Andren O, Schmidt F, Setlur SR, et al. TMPRSS2: ERG gene fusion associated with lethal prostate cancer in a watchful waiting cohort. Oncogene. 2007; 26: 4596-4599.
  20. Saramaki OR, Harjula AE, Martikainen PM, Vessella RL, Tammela TL, Visakorpi T. TMPRSS2: ERG fusion identifies a subgroup of prostate cancers with a favorable prognosis. Clin Cancer Res. 2008; 14: 3395-400.
  21. Hoogland AM, Jenster G, van Weerden WM, Trapman J, van der Kwast T, Roobol MJ, et al. ERG immunohistochemistry is not predictive for PSA recurrence, local recurrence or overall survival after radical prostatectomy for prostate cancer. Mod Pathol. 2012; 25: 471-479.
  22. FitzGerald LM, Agalliu I, Johnson K, Miller MA, Kwon EM, Hurtado-Coll A, et al. Association of TMPRSS2-ERG gene fusion with clinical characteristics and outcomes: results from a population-based study of prostate cancer. BMC Cancer. 2008; 8: 230.
  23. Garcia-Flores M, Casanova-Salas I, Rubio-Briones J, Calatrava A, Dominguez-Escrig J, Rubio L, et al. Clinico-pathological significance of the molecular alterations of the SPOP gene in prostate cancer. Eur J Cancer. 2014; 50: 2994-3002.
  24. Blattner M, Lee DJ, O'Reilly C, Park K, MacDonald TY, Khani F, et al. SPOP mutations in prostate cancer across demographically diverse patient cohorts. Neoplasia. 2014; 16: 14-20.
  25. Grupp K, Diebel F, Sirma H, Simon R, Breitmeyer K, Steurer S, et al. SPINK1 expression is tightly linked to 6q15- and 5q21-deleted ERG-fusion negative prostate cancers but unrelated to PSA recurrence. Prostate. 2013; 73: 1690-1698.
  26. Leinonen KA, Saramaki OR, Furusato B, Kimura T, Takahashi H, Egawa S, et al. Loss of PTEN is associated with aggressive behavior in ERG-positive prostate cancer. Cancer Epidemiol Biomarkers Prev. 2013; 22: 2333-2344.
  27. Ahearn TU, Pettersson A, Ebot EM, Gerke T, Graff RE, Morais CL, et al. A Prospective Investigation of PTEN Loss and ERG Expression in Lethal Prostate Cancer. J Natl Cancer Inst. 2016; 108.
  28. Rasanen K, Itkonen O, Koistinen H, Stenman UH. Emerging Roles of SPINK1 in Cancer. Clin Chem. 2016; 62: 449-457.
  29. Tomlins SA, Rhodes DR, Yu J, Varambally S, Mehra R, Perner S, et al. The role of SPINK1 in ETS rearrangement-negative prostate cancers. Cancer Cell. 2008; 13: 519-528.
  30. Smith SC, Tomlins SA. Prostate cancer SubtyPINg biomarKers and outcome: is clarity emERGing? Clinical Cancer Research. 2014; 20: 4733-4736.
  31. Flavin R, Pettersson A, Hendrickson WK, Fiorentino M, Finn S, Kunz L, et al. SPINK1 protein expression and prostate cancer progression. Clinical Cancer Research. 2014; 20: 4904-4911.
  32. Tomlins SA, Laxman B, Varambally S, Cao X, Yu J, Helgeson BE, et al. Role of the TMPRSS2-ERG gene fusion in prostate cancer. Neoplasia. 2008; 10: 177-188.
  33. Rickman DS, Chen YB, Banerjee S, Pan Y, Yu J, Vuong T, et al. ERG cooperates with androgen receptor in regulating trefoil factor 3 in prostate cancer disease progression. Neoplasia. 2010; 12: 1031-1040.
  34. Ploussard G, Terry S, Maille P, Allory Y, Sirab N, Kheuang L, et al. Class III beta-tubulin expression predicts prostate tumor aggressiveness and patient response to docetaxel-based chemotherapy. Cancer Res. 2010; 70: 9253-9264.
  35. Tsourlakis MC, Weigand P, Grupp K, Kluth M, Steurer S, Schlomm T, et al. βIII-tubulin overexpression is an independent predictor of prostate cancer progression tightly linked to ERG fusion status and PTEN deletion. Am J Pathol. 2014; 184: 609-617.
  36. Terry S, Ploussard G, Allory Y, Nicolaiew N, Boissiere-Michot F, Maille P, et al. Increased expression of class III beta-tubulin in castration-resistant human prostate cancer. Br J Cancer. 2009; 101: 951-956.
  37. Terry S, Maille P, Baaddi H, Kheuang L, Soyeux P, Nicolaiew N, et al. Cross modulation between the androgen receptor axis and protocadherin-PC in mediating neuroendocrine transdifferentiation and therapeutic resistance of prostate cancer. Neoplasia. 2013; 15: 761-772.
  38. Varambally S, Dhanasekaran SM, Zhou M, Barrette TR, Kumar-Sinha C, Sanda MG, et al. The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature. 2002; 419: 624-629.
  39. Bachmann IM, Halvorsen OJ, Collett K, Stefansson IM, Straume O, Haukaas SA, et al. EZH2 expression is associated with high proliferation rate and aggressive tumor subgroups in cutaneous melanoma and cancers of the endometrium, prostate, and breast. J Clin Oncol. 2006; 24: 268-273.
  40. Bryant RJ, Cross NA, Eaton CL, Hamdy FC, Cunliffe VT. EZH2 promotes proliferation and invasiveness of prostate cancer cells. Prostate. 2007; 67: 547-556.
  41. Melling N, Thomsen E, Tsourlakis MC, Kluth M, Hube-Magg C, Minner S, et al. Overexpression of enhancer of zeste homolog 2 (EZH2) characterizes an aggressive subset of prostate cancers and predicts patient prognosis independently from pre- and postoperatively assessed clinicopathological parameters. Carcinogenesis. 2015; 36: 1333-1340.
  42. Beltran H, Rickman DS, Park K, Chae SS, Sboner A, MacDonald TY, et al. Molecular characterization of neuroendocrine prostate cancer and identification of new drug targets. Cancer Discov. 2011; 1: 487-495.
  43. Beltran H, Prandi D, Mosquera JM, Benelli M, Puca L, Cyrta J, et al. Divergent clonal evolution of castration-resistant neuroendocrine prostate cancer. Nat Med. 2016; 22: 298-305.
  44. Roychowdhury S, Chinnaiyan AM. Advancing precision medicine for prostate cancer through genomics. J Clin Oncol. 2013; 31: 1866-1873.
  45. Rubin MA. Toward a prostate cancer precision medicine. Urol Oncol. 2015; 33: 73-74.
  46. Koh CM, Bieberich CJ, Dang CV, Nelson WG, Yegnasubramanian S, De Marzo AM. MYC and Prostate Cancer. Genes Cancer. 2010; 1: 617-628.
  47. Baca SC, Prandi D, Lawrence MS, Mosquera JM, Romanel A, Drier Y, et al. Punctuated evolution of prostate cancer genomes. Cell. 2013; 153: 666-677.
  48. Baena E, Shao Z, Linn DE, Glass K, Hamblen MJ, Fujiwara Y, et al. ETV1 directs androgen metabolism and confers aggressive prostate cancer in targeted mice and patients. Genes Dev. 2013; 27: 683-698.
  49. Boysen G, Barbieri CE, Prandi D, Blattner M, Chae SS, Dahija A, et al. SPOP mutation leads to genomic instability in prostate cancer. Elife. 2015; 4.
  50. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012; 2: 401-404.
  51. Han B, Mehra R, Lonigro RJ, Wang L, Suleman K, Menon A, et al. Fluorescence in situ hybridization study shows association of PTEN deletion with ERG rearrangement during prostate cancer progression. Mod Pathol. 2009; 22: 1083-1093.
  52. Bismar TA, Yoshimoto M, Vollmer RT, Duan Q, Firszt M, Corcos J, et al. PTEN genomic deletion is an early event associated with ERG gene rearrangements in prostate cancer. BJU Int. 2011; 107: 477-485.
  53. Nordin A, Wang W, Welen K, Damber JE. Midkine is associated with neuroendocrine differentiation in castration-resistant prostate cancer. Prostate. 2013; 73: 657-667.
  54. Barbieri CE, Rubin MA. Genomic rearrangements in prostate cancer. Curr Opin Urol. 2015; 25: 71-76.
  55. Falchook GS, Bastida CC, Kurzrock R. Aurora Kinase Inhibitors in Oncology Clinical Trials: Current State of the Progress. Semin Oncol. 2015; 42: 832-848.
  56. Frankel AE, Liu X, Minna JD. Developing EZH2-Targeted Therapy for Lung Cancer. Cancer Discov. 2016; 6: 949-952.
  57. Lens SM, Voest EE, Medema RH. Shared and separate functions of polo-like kinases and aurora kinases in cancer. Nat Rev Cancer. 2010; 10: 825-841.
  58. Terry S, Beltran H. The many faces of neuroendocrine differentiation in prostate cancer progression. Front Oncol. 2014; 4: 60.
  59. Vlachostergios PJ, Papandreou CN. Targeting neuroendocrine prostate cancer: molecular and clinical perspectives. Front Oncol. 2015; 5: 6.