Molecular Biology Underlying the Clinical
`Heterogeneity of Prostate Cancer
`An Update
`
`A. Craig Mackinnon, MD, PhD; Benjamin C. Yan, MD, PhD; Loren J. Joseph, MD; Hikmat A. Al-Ahmadie, MD
`
`● Context.—Recent studies have uncovered a number of
`possible mechanisms by which prostate cancers can be-
`come resistant to systemic androgen deprivation, most in-
`volving androgen-independent reactivation of the andro-
`gen receptor. Genome-wide expression analysis with mi-
`croarrays has identified a wide array of genes that are dif-
`ferentially expressed in metastatic prostate cancers
`compared to primary nonrecurrent tumors. Recently, re-
`current gene fusions between TMPRSS2 and ETS family
`genes have been identified and extensively studied for their
`role in prostatic carcinoma.
`Objective.—To review the recent developments in the
`molecular biology of prostate cancer, including those per-
`taining to the androgen receptor and the newly identified
`TMPRSS2-related translocations.
`Data Sources.—Literature review and personal experi-
`ence.
`Conclusions.—Prostatic adenocarcinoma is a heteroge-
`
`Prostate cancer is the most common noncutaneous ma-
`
`lignancy in men and the second most common cause
`of cancer deaths, accounting for 186 320 new annual cases
`and 28 660 deaths in America and an annual incidence of
`679 023 cases and 221 002 deaths worldwide in 2008.1,2 The
`highest incidence of prostate cancer is found in the United
`States, Canada, and Scandinavia, and the lowest in China
`and other parts of Asia.3,4 Risk factors include advancing
`age, African American ethnicity, family history, and diet.5–9
`The increasingly widespread testing for serum levels of
`prostate-specific antigen (PSA) has allowed for the in-
`creasing detection of prostate cancers at earlier stages of
`development. As a result, prostatic adenocarcinoma has
`become a clinically heterogeneous entity, with some early
`carcinomas following an indolent clinical course, remain-
`ing confined to the prostate with little effect on overall
`lifespan, while other cases can lead to the development of
`lethal metastatic disease. Despite the recent advances in
`
`Accepted for publication February 5, 2009.
`From the Department of Pathology, University of Chicago, Chicago,
`Illinois. Dr Al-Ahmadie is now with the Department of Pathology, Me-
`morial Sloan-Kettering Cancer Center, New York, New York.
`The authors have no relevant financial interest in the products or
`companies described in this article.
`Reprints: Hikmat A. Al-Ahmadie, MD, Department of Pathology, Me-
`morial Sloan-Kettering Cancer Center, 1275 York Avenue, New York,
`NY 10065 (e-mail: alahmadh@mskcc.org).
`
`neous group of neoplasms with a broad spectrum of path-
`ologic and molecular characteristics and clinical behav-
`iors. Numerous mechanisms contribute to the develop-
`ment of resistance to androgen ablation therapy, resulting
`in ligand-independent reactivation of the androgen recep-
`tor,
`including amplification, mutation, phosphorylation,
`and activation of coreceptors. Multiple translocations of
`members of the ETS oncogene family are present in ap-
`proximately half of clinically localized prostate cancers.
`TMPRSS2:ERG gene rearrangement appears to be an early
`event in prostate cancer and is not observed in benign or
`hyperplastic
`prostatic
`epithelium. Duplication
`of
`TMPRSS2:ERG appears to predict a worse prognosis. The
`relationship between TMPRSS2:ERG gene rearrangement
`and other morphologic and prognostic parameters of pros-
`tate cancer is still unclear.
`(Arch Pathol Lab Med. 2009;133:1033–1040)
`
`treatment modalities, surgical, radiation, and hormonal
`therapies for prostate cancer are not without complica-
`tions, making the development of methods for distin-
`guishing indolent cancers from their aggressive counter-
`parts necessary to avoid excessive treatment that may lead
`to significant morbidity.10 Furthermore, current methods
`of treating advanced metastatic disease often prove to be
`insufficient in the long-term. Recent research has therefore
`sought to identify new molecular pathways by which in-
`vestigators can distinguish indolent prostate cancers from
`those that go on to pursue a more aggressive clinical
`course, as well as to discover new targets for the treatment
`of metastatic carcinomas.
`Previous cytogenetic and molecular studies11–21 had
`shown that prostatic adenocarcinomas tend to incur fre-
`quent and consistent losses of specific chromosomal loci,
`including chromosomes 8p, 10q, 13q, and 17p, and, less
`commonly, 6q, 7q, 16q, and 18q, as well as gains in chro-
`mosomes 7 and 8q. More recently, investigators compared
`the gene expression profiles of primary nonrecurrent pros-
`tatic adenocarcinomas and metastatic prostate cancers by
`using microarrays.22 Genes that were more highly ex-
`pressed in metastatic carcinomas included those involved
`in DNA synthesis and repair, mitosis, and cell cycle reg-
`ulation, such as RFC5, TOP2A, RFC4, and MAD2L1, which
`have previously been shown to be highly expressed in
`proliferating cells.23 Other differentially expressed genes
`
`Arch Pathol Lab Med—Vol 133, July 2009
`
`Updates in Prostate Cancer—Mackinnon et al 1033
`
`002004
`
`AVENTIS EXHIBIT 2025
`Mylan v. Aventis, IPR2016-00712
`
`

`
`included those involved in signal transduction, transcrip-
`tional regulation, chromatin modification, RNA process-
`ing, protein synthesis, posttranslational protein modifica-
`tion, cell adhesion, cell migration, cytoskeletal regulatory
`elements, extracellular matrix proteins, biosynthetic en-
`zymes, and transport proteins.22 Several unclassified genes
`with unknown functions were also found to be differen-
`tially expressed.22
`
`PROSTATIC ADENOCARCINOMA RESISTANT TO
`ANDROGEN ABLATION THERAPY
`Prostate cancer affects 1 of 9 men older than 65 years.24
`Age correlates with a decrease in the ratio of androgens
`to estrogens in men, suggesting that a physiologic change
`in hormonal status may contribute to the progression of
`preneoplastic lesions to adenocarcinoma.25–27 Androgen
`ablation therapy is the most common systemic treatment
`for metastatic disease; it prevents testosterone production
`by the testes and thereby causes tumor regression during
`the short-term by depleting androgen-dependent carci-
`noma cells.28,29 Androgen deprivation can be achieved sur-
`gically or by medical castration, which can be performed
`by the administration of estrogens and gonadotropin-re-
`leasing agonists and antagonists, and has been shown to
`be effective in treating advanced and metastatic disease in
`several large clinical trials.30–33 However, most prostatic ad-
`enocarcinomas become refractory to androgen ablation.34
`Experiments with animal models of prostatic adenocarci-
`nomas,34,35 such as Dunning R-3327-H rat prostate carci-
`nomas and the transgenic adenocarcinoma mouse prostate
`model, have shown that androgen therapy ultimately se-
`lects for androgen-independent carcinoma cells during the
`long-term, leading to the development of highly aggres-
`sive, androgen-resistant tumors. Despite this progression
`to more aggressive disease, it is still advised that patients
`with hormone-resistant cancers continue to receive andro-
`gen ablation therapy.36
`Aberrant activation of the androgen receptor (AR) can
`result from gene amplification, mutation, phosphorylation,
`activation of coregulators, or androgen-independent acti-
`vation. Most cases of prostatic carcinoma resistant to an-
`drogen ablation therapy demonstrate activation of AR by
`one of these mechanisms.37–39 The AR gene is either mu-
`tated or amplified in 20% to 30% of androgen-resistant
`prostate carcinomas.40–42 Further, 20% of hormone-resis-
`tant carcinomas contain gene amplifications as compared
`to just 2% of hormone-sensitive tumors, suggesting that
`aberrant activation in response to low levels of androgens
`or other ligands may underlie the progression to aggres-
`sive disease that is refractory to androgen ablation thera-
`py.43 Specific mutations, such as the T877A and H874Y
`substitutions, confer increased sensitivity to AR for steroid
`hormones such as progesterone, 17␤-estradiol, or hydroxy-
`flutamide in prostate cancer cell lines and xenografts.44,45
`Mutant AR containing the E231G substitution has also
`been shown to predispose transgenic mice to the devel-
`opment of prostatic intraepithelial neoplasia (PIN), ade-
`nocarcinoma, and metastases.46 Experiments have shown
`that AR hyperactivity results in the formation of prostatic
`neoplasms: overexpression of AR leads to the develop-
`ment of
`focal PIN, whereas AR overexpression in
`LAPC-4 prostate cancer cells and xenografts results in a
`transition from androgen-sensitive disease to androgen-
`resistant cancer.47,48 Phosphorylation of AR at different
`specific serine residues may cause stabilization of the pro-
`
`tein against proteolytic degradation or induce transcrip-
`tional activation of the receptor protein.49
`Inappropriate AR hyperactivity may also be caused by
`activation of coregulators. Recurrent CWR22 tumors were
`found to harbor overexpressed transcriptional intermedi-
`ary factor 2 and steroid receptor coactivator 1, which in-
`creased AR transactivation at physiologic androgen con-
`centrations.37 Other coregulators of AR function include
`ARA70, p300, CBP, Tip60, ARA55, ARA54, gelsolin, Stat3,
`and RAC3.50–53 The Foxa1 and Foxa2 proteins are tran-
`scription factors belonging to the forkhead box A (Foxa)
`superfamily (also known as hepatocyte nuclear factor 3
`proteins) that are essential for endodermal development
`and are involved in respiratory, intestinal, and hepatic
`gene expression.54–60 Although Foxa1 was found to be ex-
`pressed in prostatic carcinomas of different grades, Foxa2
`stimulates transactivation of the PSA promoter in an an-
`drogen- and AR-independent manner and has been iden-
`tified in small cell carcinomas and high-grade adenocar-
`cinomas of the prostate, suggesting that Foxa2 regulation
`of gene expression may contribute to progression of pros-
`tatic carcinomas to a more aggressive and androgen-in-
`dependent state.58
`Genome-wide expression analyses61 have identified
`genes that are differentially expressed in prostate cancers
`from patients who had received the gonadotropin-releas-
`ing agonist goserelin and AR antagonist flutamide for 3
`months. Hierarchical clustering algorithms that analyzed
`gene expression profiles classified the specimens accord-
`ing to treatment status, suggesting that distinct transcrip-
`tional programs are activated in prostate carcinomas in
`response to androgen therapy. The genes that were more
`highly expressed in carcinomas treated with androgen ab-
`lation agents included those encoding AR and steroid bio-
`synthetic enzymes, as well as a suite of genes that have
`been previously shown to be targets of AR or have been
`implicated as being regulated by it, including the gene
`encoding PSA (kallikrein-related peptidase 3), KLK3, and
`KLK2, as well as DBI, FASN, IL6, SERPINB5, TGFBR3,
`TMPRSS2, TUBA1, HOXC6, TRG , and FOLH1.61–66 Other
`upregulated genes may represent secondary, indirect ef-
`fects of androgen ablation that occur later than reactiva-
`tion of AR. To identify only those genes that are subject
`to transcriptional regulation by AR, gene expression pro-
`files of LNCaP human prostatic adenocarcinoma cells
`were examined after androgen withdrawal.61 Approxi-
`mately 25% of the genes differentially expressed in carci-
`nomas after chronic androgen ablation therapy also
`showed an altered transcript level in the carcinoma cell
`line. Finally, comparison of the gene expression profiles of
`androgen-resistant cancers to those of cancers that had not
`developed resistance demonstrated that prostatic adeno-
`carcinomas resistant to androgen ablation therapy had
`gene expression profiles more similar to those of untreat-
`ed, androgen-dependent tumors than of cancers under
`conditions of androgen deprivation. This finding suggests
`a reversal in the gene expression profile of androgen-re-
`fractory cancers that is caused by androgen deprivation
`therapy, possibly by ligand-independent reactivation of
`AR, a mechanism that has been proposed by several au-
`thors.67–70 Furthermore, a unique set of genes was ex-
`pressed in androgen-resistant prostatic carcinomas that
`was not expressed in primary androgen-dependent tu-
`mors or in other metastatic carcinomas.61
`Activation of AR can be highlighted by immunohisto-
`
`1034 Arch Pathol Lab Med—Vol 133, July 2009
`
`Updates in Prostate Cancer—Mackinnon et al
`
`

`
`chemistry as a strong nuclear expression in androgen-re-
`sistant prostate cancers.61 Moreover, reactivation of AR
`gene was not due to gene amplification in most cases, as
`it was shown by FISH analysis that only a minority of the
`androgen-resistant carcinomas studied contained ampli-
`fied AR genes. The human prostatic adenocarcinoma xe-
`nograft CWR22, which is propagated in nude mice, reca-
`pitulates the properties of in vivo prostate cancers, with
`an initial period of androgen-dependent proliferation fol-
`lowed by persistent growth several months after androgen
`deprivation.71,72 Androgen receptor protein from a re-
`lapsed CWR22 carcinoma has a half-life that is 2 to 4 times
`that of AR from LNCaP cells, demonstrating that recurrent
`tumors have hyperstabilized AR as compared to andro-
`gen-dependent neoplasms.73 The increased expression,
`greater stability, and nuclear localization of AR in recur-
`rent prostate cancers resistant to androgen deprivation
`correlated with hypersensitivity to low levels of androgens
`in these tumors; androgen ablation–resistant prostate can-
`cers required a significantly much lower concentration of
`dihydrotestosterone than that required by androgen-de-
`pendent tumors for stimulation of proliferative activity.
`Several of the genes that were found by the microarray
`study to be more highly transcribed in androgen ablation–
`resistant tumors encoded biosynthetic enzymes involved
`in the synthesis of cholesterol, including HMG-CoA syn-
`thase, squalene synthase, lanosterol synthase, and squa-
`lene monooxygenase, the rate-limiting enzyme in sterol
`synthesis.61,74 Androgen ablation–resistant tumors were
`shown to be more strongly immunoreactive for squalene
`monooxygenase than were androgen-dependent tumors.61
`The increased production of steroid biosynthetic enzymes
`in resistant tumors suggests that one mechanism by which
`these carcinomas overcome androgen deprivation is by
`compensatory synthesis of androgens, with consequently
`increased AR activity. Recurrent prostatic carcinomas con-
`sistently exhibit decreased expression of the tumor sup-
`pressor gene PTEN (phosphatase and tensin homolog),
`which carries loss-of-function mutations in advanced pros-
`tate cancers.75 The PTEN protein dephosphorylates phos-
`phatidylinositol-3,4,5-trisphosphate (PIP3), resulting in in-
`hibition of the Akt (protein kinase B) cell survival signal-
`ing pathway.76
`Besides reactivation of AR and loss of PTEN tumor sup-
`pressor activity, other mechanisms for the development of
`hormone-resistant prostate cancers have been proposed.
`Aberrant overexpression or amplification of the HER2/neu
`gene has been identified in prostatic carcinomas77 and el-
`evated serum levels of the HER2/neu extracellular domain
`were found in androgen ablation–refractory prostate can-
`cers.78 Overexpression of the HER2/neu (ERBB2, CD340)
`receptor tyrosine kinase was capable of rescuing LNCaP
`cells from the antiproliferative effect of androgen depri-
`vation and also shortened the latency period for tumor
`formation in castrated mice with severe combined im-
`munodeficiency by 50%.77 Furthermore, HER2/neu can en-
`hance by 15-fold the expression of the AR target PSA in
`the absence of androgens in LAPC-4 cells.77 HER2/neu can
`also activate MAP kinase and PIP3/Akt signaling cas-
`cades, culminating in the phosphorylation of serines 213
`and 791 of AR.79,80 Constitutive Akt activity led to in-
`creased neoplastic growth in LNCaP xenografts.81 Other
`growth factors that may contribute to the development of
`hormone-resistant cancers include insulin-like growth fac-
`
`tor 1, epidermal growth factor, keratinocyte growth factor,
`and factors secreted by neuroendocrine cells.82–84
`The conserved basic helix-loop-helix transcription factor
`TWIST has been shown to be highly expressed in the ma-
`jority (90%) of prostate cancers and only in a minority
`(6.7%) of benign prostatic hyperplasia cases.85 TWIST ex-
`pression levels were also found to be proportional to Glea-
`son grade, and higher levels correlated with metastasis.85
`Experiments with DU145 and PC-3 androgen-resistant
`prostatic adenocarcinoma cell lines85 found that down-reg-
`ulation of TWIST expression by RNA interference led to
`suppression of invasiveness and a reduction in E-cadherin
`expression, as well as loss of the morphologic and molec-
`ular changes that signify the epithelial-mesenchymal tran-
`sition.
`In summary, a major clinical challenge presented by
`prostate cancer is the treatment of androgen ablation–re-
`sistant carcinomas. Recent experimental evidence suggests
`that there are multiple avenues leading to the development
`of this aggressive form of prostatic carcinoma, which sub-
`vert the molecular mechanisms of the cell to reactivate AR,
`activate its targets, gain inappropriate HER2/neu activity,
`lose PTEN-mediated tumor suppression, or stimulate the
`epithelial-mesenchymal transition via TWIST. Other au-
`thors postulate the existence of androgen-resistant pros-
`tate cancer stem cells that contribute to the growth of ag-
`gressive tumors. Future molecular studies will help fur-
`ther elucidate the diverse signaling pathways underlying
`the pathogenesis of prostate cancers refractory to systemic
`hormonal deprivation and may lead to the development
`of multiple pharmacologic agents and therapeutic modal-
`ities that will halt progression of hormone-naı¨ve tumors.
`
`MOLECULAR BIOLOGY OF TMPRSS2:ERG
`TRANSLOCATION
`A significant role for the ETS gene family, which en-
`codes transcription factors in prostate cancer, was recently
`discovered by using a novel bioinformatics approach
`known as COPA (cancer outlier profile analysis)86 that
`identified the oncogenes ERG (21q22.2), ETV1 (7p21.2),
`ETV4 (17q21), and ETV5 (3q27) as very highly expressed
`in a subset of prostate cancers on the basis of a large set
`of microarray data.86–88 ERG, ETV1, ETV4, and ETV5 are
`members of the ETS family of transcription factors, which
`are characterized by an evolutionarily conserved, 85–ami-
`no acid DNA-binding domain that facilitates binding to
`purine-rich DNA with a GGAA/T core consensus se-
`quence.89 ETS proteins function cooperatively with other
`transcription factors in the regulation of a diversity of cel-
`lular functions including proliferation, differentiation, an-
`giogenesis, hematopoiesis, oncogenic transformation, and
`apoptosis.90 Importantly, translocations involving mem-
`bers of the ETS family have been identified in human leu-
`kemia and solid tumors.89 A mechanism underlying ETS
`overexpression in prostate cancer was established once it
`was recognized that
`the androgen-responsive gene
`TMPRSS2 (see below) is fused to the coding region of an
`ETS family member (for example ERG) as a result of gene
`rearrangement, which was also demonstrated directly by
`studies in vitro.86,91,92 The TMPRRSS2:ERG gene fusion is
`observed in greater than 90% of prostate cancers with
`ETS-family gene rearrangements,93 whereas TMPRSS2:
`ETV1, TMPRSS2:ETV4, and TMPRSS2:ETV5 rearrange-
`ments occur more rarely. Furthermore, ETV1, ETV4, and
`ETV5 have additional
`fusion partners other
`than
`
`Arch Pathol Lab Med—Vol 133, July 2009
`
`Updates in Prostate Cancer—Mackinnon et al 1035
`
`

`
`TMPRSS2, including SLC45A3, HERV-K㛮22q11.3, C15orf21,
`and HNRPA2B1.94
`TMPRSS2 is located at 21q22.2,95 and TMPRSS2 is pre-
`dominantly expressed in luminal epithelial prostate cells,
`with much lower expression in pancreas, kidney, lung, co-
`lon, and liver and no measurable expression in testes, ova-
`ry, placenta, spleen, thymus, circulating leukocytes, brain,
`heart, or skeletal muscle.96,97 TMPRSS2 and ERG are lo-
`cated 3Mb apart on chromosome 21q22.2-22.3 (Figure, A
`through E). The 5⬘ ends of both genes are orientated to-
`ward the telomere, and TMPRSS2 is positioned telomeri-
`cally relative to ERG. Interstitial deletion of the intervening
`intronic genomic DNA is the most common mechanism
`for fusion and is observed in 60% to 90% of TMPRSS2:
`ERG fusion-positive prostate cancers (see below). Regions
`of microhomology exist in the TMPRSS2 and ERG loci,
`suggesting that they might underlie rearrangement events
`during defective homologous recombination.98
`Seventeen different types of TMPRSS2:ERG fusion tran-
`scripts involving various regions of the TMPRSS2 and
`ERG genes have been identified 86,99–102; however, 8 of these
`transcripts are unlikely to result in the translation of func-
`tional ERG proteins due to the introduction of premature
`stop codons. Of the 9 predicted functional TMPRSS2:ERG
`fusion transcripts, 2 code for normal ERG proteins, 6 code
`for amino-terminal–truncated ERG proteins, and 1 codes
`for a bona fide TMPRSS2:ERG fusion protein.101 These
`studies demonstrate that expression of multiple fusion
`mRNAs is common, with TMPRSS2 exon 1 fused to ERG
`exon 4 being the most frequently expressed type of
`TMPRSS2:ERG fusion. Alternative splicing of
`the
`TMPRSS2:ERG gene is proposed as the most likely basis
`for the multiple different types of fusion mRNAs ob-
`served.99,101
`
`DETECTION AND PREVALENCE OF TMPRSS2:ERG GENE
`FUSION IN PROSTATE CANCER
`Most TMPRSS2:ERG gene fusion events in patients with
`clinically localized prostate cancer (ie, patients identified
`through PSA screening who have potentially curable dis-
`ease by surgical resection) are characterized by using ei-
`ther fluorescence in situ hybridization (FISH) or quanti-
`tative reverse transcription–polymerase chain reaction
`(RT-PCR). One common FISH method uses break-apart
`probes that bind the 5⬘ (ie, green) and 3⬘ (ie, red) ends of
`the ERG gene (Figure). In normal prostate tissue, both of
`these probes hybridize to the ERG locus and generate ad-
`jacent green and red fluorescent signals in the nucleus. In
`contrast, prostate cancer cells harboring rearranged ERG
`demonstrate distinct, separate green and red fluorescent
`signals, as the probes are split because of change in chro-
`mosome structure. One advantage of this method is that
`it also reveals the mechanism underlying the rearrange-
`ment by detection of the commonly occurring 3-Mb inter-
`stitial deletion between TMPRSS2 and ERG. In such cases,
`this deletion manifests by the loss of the 5⬘ (green) signal
`in nuclei of malignant cells.
`The initial report86 identifying the TMPRSS2:ERG gene
`rearrangement demonstrated that 47% of clinically local-
`ized prostate cancers contain the TMPRSS2:ERG fusion,
`and two-thirds of these translocations are formed second-
`ary to interstitial deletion between the TMPRSS2 and ERG
`genes
`on
`chromosome
`21q22.2-22.3.
`Subsequent
`work101,103–106 confirms that the TMPRSS2:ERG rearrange-
`ment is present in approximately 50% of primary prostate
`
`cancer samples with interstitial deletion of the 5⬘ region
`of ERG occurring in 60% of the TMPRSS2:ERG-positive
`primary prostate cancers. Gene rearrangements involving
`ETV1 and ETV4 are rare, accounting for approximately
`2% of all observed gene alterations.
`
`MORPHOLOGY OF TMPRSS2:ERG PROSTATE CANCER
`Initially, morphologic analysis of prostate cancer cases107
`identified 5 features strongly associated with the presence
`of the TMPRSS2:ERG fusion, which included blue-tinged
`mucin, cribriform growth pattern, macronucleoli, intra-
`ductal tumor spread, and signet ring cell features. Of the
`cases demonstrating none of these features, 24% were
`TMPRSS2:ERG fusion-positive. Conversely, 55%, 86%, and
`93% of cases with 1, 2, or 3⫹ features, respectively, were
`TMPRSS2:ERG fusion-positive.
`The association of the TMPRSS2:ERG fusion with high-
`grade prostatic intraepithelial neoplasia (HGPIN) was ob-
`served in up to 21% of cases in several studies.94,108,109 Un-
`like localized prostate cancer, in which TMPRSS2:ERG fu-
`sion-positive prostate cancers typically demonstrate over-
`expression of ERG, overexpression of ERG is less constant
`in TMPRSS2:ERG fusion-positive HGPIN.109 TMPRSS2:
`ERG fusions were not present in benign prostatic epithe-
`lium or other lesions not associated with prostate cancer,
`specifically, benign prostatic hyperplasia and proliferative
`inflammatory atrophy. Furthermore, all TMPRSS2:ERG-
`positive HGPIN cases identified by FISH showed the same
`fusion pattern as the matching prostate cancer from the
`same patient, and no fusion-positive HGPIN cases asso-
`ciated with fusion-negative prostate cancer were identified
`by FISH,108,110 suggesting TMPRSS2:ERG may play a role
`in the progression from HGPIN to adenocarcinoma.92
`Based upon (1) the homogenous distribution of the fu-
`sion gene throughout the cancer, (2) the absence of de-
`tectable TMPRSS2:ERG fusion events in normal and hy-
`perplastic prostate tissue, (3) the finding that TMPRSS2:
`ERG-positive HGPIN lesions show the same fusion pat-
`tern with the matching prostate cancer, and (4) the fact
`that TMPRSS2:ERG fusion-positive HGPIN is never ob-
`served with fusion-negative matching prostate cancer, it
`can be suggested that TMPRSS2:ERG fusion is an early
`event in the development of prostate adenocarcinoma. In
`support of these findings, transgenic mice expressing the
`equivalent
`truncated ERG gene coded by human
`TMPRSS2:ERG develop mouse PIN (mPIN), along with
`loss of the p63-positive basal layer adjacent to mPIN foci.92
`These findings strongly suggest that TMPRSS2:ERG fu-
`sion HGPIN is a true precursor for TMPRSS2:ERG-posi-
`tive prostate cancer.
`
`TMPRSS2:ERG AND CLINICAL PROGNOSIS
`As a wide array of distinct TMPRSS2:ERG fusions have
`been identified, several studies have explored whether
`specific gene fusion isoforms correlate with aggressive
`clinical behavior. One group101 observed that expression of
`TMPRSS2:ERG fusion variants consisting of exons 1 and
`2 of TMPRSS2 fused to exon 4 of ERG (T2E4) and, to a
`lesser extent, exon 1 of TMPRSS2 juxtaposed to exons 2
`or 3 of ERG (T1E2/3) are associated with early recurrence
`and seminal vesicle invasion. Interestingly, these fusion
`isoforms all contain the native ATG translation initiation
`codon from either TMPRSS2 or ERG, raising the possibil-
`ity that increased efficiency of translation from native
`translation start codons may underlie the correlation be-
`
`1036 Arch Pathol Lab Med—Vol 133, July 2009
`
`Updates in Prostate Cancer—Mackinnon et al
`
`

`
`Model of TMPRSS2:ERG gene rearrangements and fluorescence in situ hybridization (FISH) patterns observed in prostate cancer. A, Physical map
`of the TMPRSS2 and ERG loci on 21q22.2-22.3. T and C orientate toward the telomeric and centromeric regions, respectively. 5⬘ERG (green) and
`3⬘ ERG (red) FISH break-apart probes are positioned above the chromosome relative to where they hybridize. The TMPRSS2 and ERG loci are
`separated by approximately 3 Mb. B, Chromosome structure (upper) and nuclear FISH pattern (lower) observed in a normal ERG locus. This pattern
`is observed in approximately 50% of all clinically localized prostate cancers. C, Separated 5⬘ ERG and 3⬘ ERG loci due to gene rearrangement.
`This pattern has been referred to as Esplit (see text). Note the retention of both the 5⬘ ERG and 3⬘ ERG FISH probe signals. D, Chromosome
`structure and nuclear FISH pattern observed with an interstitial deletion of the 5⬘ERG locus. Note the loss of the 5⬘FISH probe signal. This is the
`most common TMPRSS2:ERG gene rearrangement pattern observed in prostate cancer and has been referred to as 1Edel. E, Interstitial deletion of
`the 5⬘ ERG locus accompanied by a duplication of 3⬘ ERG sequence. Note the loss of the 5⬘ FISH probe and duplication of the 3⬘ FISH probe
`signals. This pattern of TMPRSS2:ERG gene rearrangement is associated with a worse clinical prognosis and has been referred to as 2⫹Edel.
`
`tween TMPRSS2:ERG fusion type and aggressive clinical
`course. Alternatively, it may represent altered biochemical
`properties of the TMPRSS2:ERG fusion protein.
`A comprehensive FISH analysis of TMPRSS2:ERG re-
`arrangement in 445 cases111 has demonstrated that pros-
`tate cancer in which the 5⬘ portion of ERG is deleted has
`significantly worse cause-specific and overall survival
`than prostate cancer in which ERG is either not disrupted
`(ie, normal EGR) or contains a balanced ERG translocation
`(ie, split EGR). When ERG-deleted prostate cancers were
`further analyzed, the authors observed that prostate can-
`cer with 2 or more copies of the 3⬘ ERG region showed
`much worse clinical behavior: the survival rate of patients
`with duplication of the TMPRSS2:ERG gene rearrange-
`ment was 25% at 8 years compared to 90% for patients
`with prostate cancer without ERG rearrangement. A sep-
`arate study examining 521 patients112 reported similar re-
`sults in which duplication of TMPRSS2:ERG was associ-
`
`Arch Pathol Lab Med—Vol 133, July 2009
`
`ated with higher clinical stage and aggressive disease.
`Furthermore, analysis of 214 patients with prostate cancer
`suggested that multiple copies of TMPRSS2:ERG were as-
`sociated with greater prostate cancer–specific mortality, al-
`though this study113 was not statistically significant. Taken
`together, these results demonstrate that ERG gene copy
`number may provide useful prognostic information for
`patients with prostate cancer.
`The association of TMPRSS2:ERG gene rearrangement
`with Gleason score, aggressive disease, and prognosis is
`unclear, as multiple studies with conflicting findings have
`been described. A population-based study114 of 252 men
`followed up expectantly with low-stage (T1a-bNXM0)
`prostate cancer explored the risk of metastasis or prostate
`cancer–specific death based upon the presence of the
`TMPRSS2:ERG fusion. These authors determined that
`TMPRSS2:ERG fusion-positive prostate cancer is associ-
`ated with higher Gleason score (⬎7) than fusion-negative
`Updates in Prostate Cancer—Mackinnon et al 1037
`
`

`
`prostate cancer. Furthermore, cumulative incidence regres-
`sion analysis demonstrated a significant association be-
`tween TMPRSS2:ERG fusion-positive prostate cancer and
`metastases or disease-specific death. Another study115 re-
`ported similar findings. However, several other stud-
`ies98,110,116–118 failed to establish any correlation between
`TMPRSS2:ERG gene fusion status and Gleason score, tu-
`mor stage, or clinical outcomes. Lastly, using a xenograft
`model system, Hermans et al119 demonstrated that ad-
`vanced, AR-negative tumors did not express
`the
`TMPRSS2:ERG fusion transcript despite its presence in
`the genomic DNA, indicating that TMPRSS2:ERG is not
`involved in androgen-independent growth of these xeno-
`grafts. This finding suggests that the TMPRSS2:ERG fu-
`sion is important during the early, androgen-sensitive
`stage of
`tumor growth, but androgen-dependent
`TMPRSS2:ERG expression is bypassed and down-regulat-
`ed as tumor growth progresses and becomes androgen
`resistant.119
`In summary, the discovery of TMPRSS2 gene rearrange-
`ments helped broaden our understanding of the molecular
`pathology of prostate cancer, and the numerous studies
`that followed the initial report confirmed the high preva-
`lence of TMPRSS2:ETS gene alterations in prostate cancer,
`as well as advanced our understanding of androgen sig-
`naling during prostate cancer progression. Mouse and hu-
`man studies clearly demonstrate the occurrence of
`TMPRSS2:ETS gene fusion events in early HGPIN lesions;
`however, no evidence to date demonstrates a direct role
`for TMPRSS2:ETS fusion genes in the progression to ad-
`enocarcinoma, a fact suggesting a requirement for addi-
`tional genetic mutations in the course of prostate cancer
`development.
`
`References
`1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA Cancer J Clin.
`2008;58:71–96.
`2. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA
`Cancer J Clin. 2005;55:74–108.
`3. Gronberg H. Prostate cancer epidemiology. Lancet. 2003;361:859–864.
`4. Quinn M, Babb P. Patterns and trends in prostate cancer incidence, survival,
`prevalence and mortality—part I: international comparisons. BJU Int. 2002;90:
`162–173.
`5. American Cancer Society. Cancer Facts and Figures, 2003. Atlanta, GA:
`American Cancer Society; 2003.
`6. Bratt O. Hereditary prostate cancer: clinical aspects. J Urol. 2002;168:906–
`913.
`7. Carter BS, Beaty TH, Steinberg GD, Childs B, Walsh PC. Mendelian inher-
`itance of familial prostate cancer. Proc Natl Acad Sci U S A. 1992;89:3367–3371.
`8. Haas GP, Sakr WA. Epidemiology of prostate cancer. CA Cancer J Clin.
`1997;47:273–287.
`9. Steinberg GD, Carter BS, Beaty TH, Childs B, Walsh PC. Family history and
`the risk of prostate cancer. Prostate. 1990;17:337–347.
`10. Michaelson MD, Cotter SE, Gargollo PC, Zietman AL, Dahl DM, Smith
`MR. Management of complications of prostate