2 EUROPEAN UROLOGY SUPPLEMEN TS XXX (2017) XXX–XXX
of biologically significant PCA would help further refine the
current clinical and pathologic input based management
algorithms. Such biomarkers will enhance confidence in
accurately assigning patients with insignificant PCA catego-
rized as very low risk in National Comprehensive Cancer
Network guidelines (http://www.nccn.org/professionals/
physician_gls/f_guidelines.asp) to active surveillance ap-
proach that is coupled with active monitoring by molecular
imaging and ongoing look out for genomic signatures of
biologic progression using tissue samples.
2. Pathogenesis of prostate cancer: genetic risk and
environmental factors
The interplay between genetic determinants of risk and
environmental factors is responsible for the long-observed
variation in PCA incidence among geographic populations.
The higher incidence of PCA in African-Americans compared
with Asian-Americans is thought to be only partially driven
by genetic predisposition [11]—the fact that such risk is
modulated upon migration in a given ethnic group points to
environmental and lifestyle factors as additional contribut-
ing determinants of risk [4,5,12,13].
2.1. Inherited genetic risk factors
Twins and family pedigree studies have long supported
genetic predisposition as a risk factor for PCA
[11,14,15]. Men with a first degree relative affected by
the disease are at twice the risk for developing PCA. The risk
is even higher (> 4-fold) if a relative was diagnosed with
PCA at a younger age (< 60 yr) [16,17].
PCA is now recognized as one of the most heritable
cancer types driven by numerous inherited germline
genetic variants of risk ranging from those that are more
common with a weak risk effect to the rare variants that are
associated with a stronger risk for PCA. While earlier linkage
analysis studies pointed to numerous chromosomal loci of
association, such claims have failed to be consistently
validated [16,18–25]. The same is true for the initial
suggestion that inflammatory and infection response gene
loci (ELAC2, RANSEL, and MSR1) were associated with higher
risk of PCA [26,27]. In the largest linkage study performed
by the international consortium of PCA genetics only one
locus (22q) stood out [28]. Subsequent studies also pointed
to 8q24 [29–31] as a region harboring genetic risk variants.
In the last decade, the application of Genome Wide
Association Studies that enables assessment of millions of
single nucleotide polymorphisms (SNPs) in a given indi-
vidual for disease risk association to PCA [32] has allowed
for the detection of far more common germline genetic
variants with only low-to-moderate penetrance. To date,
over 280 SNPs associations with PCA risk have been
established by 30 such studies (A Catalog of Published
Genome-Wide Association Studies (https://www.ebi.ac.uk/
gwas/search?query=Prostate%20cancer#association). Col-
lectively, these SNPs may account for up to one-third of
PCA familial risk. Regions such 8q24 region [33,34]
containing SNPs exerting a modifier effect on neighboring
Please
Pleasecite
citethis
thisarticle
articleininpress
pressas:
as:Netto
NettoGJ,
GJ,etetal.
al.Prostate
ProstateCancer:
Cancer:An
Eur
EurUrol
UrolSuppl
Suppl(2017),
(2017),https://doi.org/10.1016/j.eursup.2017.10.001
https://doi.org/10.1016/j.eursup.2017.10.001
MYC oncogene, a recognized player in PCA pathogenesis
[31] and SNP variants located on chromosomes 19 q13 that
harbor kallikrines KLK2 and KLK3 (PSA) genes are of
particular interest [35,36]. Although each susceptibility
SNP allele individually carries only a small risk, multiplica-
tively, a subject SNP profile can be deduced from risk
algorithms models that will identify individuals in the
upper 1% risk tier with approximately 5-fold the risk of the
general population [37,38]. It is such an approach that have
the potential to refine the best groups of men to be targeted
for screening and prevention strategies and help address
the current concerns of overdiagnosis and overtreatment of
PCA [8,9].
The detection of the more rare but strongly penetrant
germline variants (> 5% frequencies) requires exhaustive
case/control direct sequencing that was only achievable
with more recently propagated massively parallel next
generation sequencing technologies. Given their rarity, such
variants account for only a minute fraction of PCA incidence
but importantly impart a high risk in carrier individuals for
early onset (5—7-fold) and at times more aggressive PCA.
Most established among these are germline mutations in
the BRCA 2 tumor suppressor gene [39] and HOXB13 (G84E)
[40,41]. A 5% germline carrier frequency for HOXB13 (G84E)
mutation was shown in families of PCA of mostly European
descent [41]. Men with the BRCA2 germline mutation were
found to be at 5-fold the risk for PCA (7-fold the risk of early
disease) in the Breast Cancer Linkage Consortium study
[42]. The evidence for BRCA1 [43] and other DNA repair
genes such as PALB2, CHECK2, BRIP1, and NBS1 remains less
robust [44–47].
Finally, molecular risk studies support increased sus-
ceptibility for PCA in Lynch syndrome. A recent meta-
analysis of 12 risk studies showed a 2.28-fold (95%
confidence interval [CI]: 1.37–3.19) increased risk of PCA
for all men from mismatch repair genes mutation-carrying
families. In another study, the relative risk was greatest for
MSH2 carriers (5.8, 95% CI: 2.6–20.9) where PCA was the
first or only diagnosed tumor in 37% of carriers [48–50].
Several multinational consortia (PCA Association Group
to Investigate Cancer Associated Alterations in the Genome,
http://practical.ccge.medschl.cam.ac.uk/; International
Consortium for PCA Genetics, http://www.icpcg.org/
?q=content/about-icpcg; Elucidating Loci Involved in PCA
Susceptibility, http://epi.grants.cancer.gov/gameon/
personnel.html#ellipse) are investigating the important
clinical impact of genetic susceptibility to PCA in order to
address the potential screening, risk management guide-
lines and functional and treatment implications of the
growing list of identified germline genetic variants.
2.2. Environmental risk factors
A body of evidence points to glandular epithelial cell injury
resulting from chronic exposure to dietary carcinogens,
estrogens, or oxidants as a trigger for a chronic inflamma-
tory milieu that set the stage for cancer development
[12,51–55]. Epidemiologic and animal model studies
[56,57] strongly support the dietary intake of charred red
AnUpdate
Updateon
onMolecular
MolecularPathology
Pathologywith
withClinical
ClinicalImplications.
Implications.
 EUROPEAN UROLOGY SUPPLEMEN TS XXX (2017) XXX–XXX
Fig. 1 – Etiological factors implicated in prostate cancer development. Chronic inflammation triggered by environmental and lifestyle exposures leads
to persistent prostate epithelial cell damage. Inherited genetic predisposition also plays a determining factor in promoting oncogenesis.
PIA = proliferative inflammatory atrophy; PIN = prostatic intraepithelial neoplasia; SNP = single-nucleotide polymorphism; STD = sexually transmitted
disease.
meats and animal fats as risk factors for PCA through
formation of heterocyclic aromatic amine (eg, 2-amino-1-
methyl-6-phenylimidazopyrine) and polycyclic aromatic
hydrocarbon carcinogens [58–60]. Likewise, animal data
link estrogen exposure to prostate epithelial cell damage
and inflammation potentially through induction of autoim-
munity [61,62]. Less robust evidence implicates sexually
transmitted infectious agents (eg, trichomonas, chlamydia,
and gonorrhea) as potential initiators of chronic inflamma-
tion of the prostate (Fig. 1) [63–66]. Epithelial damage and
ensuing inflammation is the common pathogenic link
between all the above environmental carcinogens and
PCA development. The persistent oxidative stress exerted
upon prostatic epithelial cells is counteracted by a genome
damage defense and cell survival response through
expression of a and p class glutathione S-transferases,
cyclooxygenase-2, and others mediators [54,67–69]. Subse-
quently, epigenetic silencing of hundreds of genes including
the crucial caretaker gene GSTP1 take hold in the damaged
epithelium and persists throughout subsequent cancer
progression phases. Such changes are detectable in the
earliest histologic manifestation of injury response, namely
proliferative inflammatory atrophy that seems to share
many of the somatic genetic and epigenetic alterations that
are exhibited by prostatic intraepithelial neoplasm and PCA
[70–72].
3. Somatic molecular alterations in PCA
3.1. Genetic alterations
PCA is a complex and heterogeneous disease. Prostate
tumor development and progression involves alterations in
numerous genetic pathways. As our understanding of these
pathways and key driver molecular alterations evolved, an
Please
Pleasecite
citethis
thisarticle
articleininpress
pressas:
as:Netto
NettoGJ,
GJ,etetal.
al.Prostate
ProstateCancer:
Cancer:An
AnUpdate
Eur
EurUrol
UrolSuppl
Suppl(2017),
(2017),https://doi.org/10.1016/j.eursup.2017.10.001
https://doi.org/10.1016/j.eursup.2017.10.001
3
extensive list of biomarkers has been evaluated in the last
2 decades for their potential role in predicting disease
outcome and as therapeutic targets [4,5,12,72–75]. They
include markers of proliferation index (ki67) [76–79],
tumor suppression genes (eg, p53, p21, p27, NKX3.1, PTEN,
retinoblastoma gene Rb), oncogenes (eg, Bcl2, c-myc, EZH2,
and HER2/neu), adhesion molecules (CD44, E-Cadherin),
PI3K/akt/mTOR pathway members [80,81], apoptosis reg-
ulators (eg, survivin and transforming growth factor b1),
androgen receptor status, and prostate tissue lineage
specific markers (PSA, prostatic-specific acid phosphatase,
and prostate-specific membrane antigen).
The role of p53 expression in predicting prognosis in PCA
has been extensively studied. Brewster et al [82] found p53
expression and Gleason score in needle biopsy to be
independent predictors of biochemical recurrence follow-
ing radical prostatectomy (RP) [82]. Retrospective studies
evaluating prostatectomy specimens also found p53 to be of
prognostic significance independent of grade, stage, margin
status [83–85]. The later has been further illustrated in
more recent genome wide studies [86]. A prognostic role of
p27 in predicting progression following prostatectomy have
also been shown while less robust evidence exists for the
prognostic role of p21 [87] and NKX3.1 [88,89]. Several
studies have supported a prognostic role for Bcl2 [82,84] and
myc oncogenes [90] as potential adjuncts to histologic
prognostic parameters. With the exception of tumor
suppressor gene PTEN and myc oncogene, none of the
above individual biomarkers have transitioned into clinical
use.
Assessment of loss of PTEN alone or in combination with
ERG fusion have recently gained significant momentum
in routine practice as an ancillary test to help better
delineate indolent lower grade disease that can be managed
by active surveillance. The following sections further
Updateon
onMolecular
MolecularPathology
Pathologywith
withClinical
ClinicalImplications.
Implications.
4 EUROPEAN UROLOGY SUPPLEMEN TS XXX (2017) XXX–XXX
expand on the important role of ETS gene fusions and
mammalian target of rapamycin (mTOR) pathway in PCA.
3.2. ETS gene fusions
In 2005, Tomlins et al [91,92] identified a recurrent
chromosomal rearrangement in over one-half of their
analyzed PCA cases. The rearrangements lead predomi-
nantly to the fusion of the androgen-responsive promoter
elements of the TMPRSS2 gene (21q22) to one of three
members of the ETS transcription factors family members
ERG, ETV1, and ETV4 located at chromosomes 21q22, 7p21,
and 17q21, respectively. Although the prognostic role of
assessing TMPRSS2-ETS rearrangements in PCA tissue
samples, as a standalone marker, was subsequently not
substantiated in well-designed large cohort studies [93,94],
the discovery had great implications in terms of furthering
our understanding of PCA pathogenesis and provided a new
marker for molecular diagnosis in PCA [95–99]. Further-
more, as shown below, recent genome wide studies have
repointed to the prognostic significance of TMPRSS2-ERG
fusion status as part of genomic signatures that could
stratify disease aggressiveness. The potential diagnostic and
prognostic role of detecting TMPRSS2-ERG fusion in post-
prostate massage urine samples is also very promising
[100–102].
The development of commercial anti-ERG monoclonal
antibodies facilitated the use immunohistochemistry (IHC)
for the evaluation of ERG protein expression as a simpler, less
costly, surrogate approach to fluorescent in situ hybridiza-
tion (FISH) for detection of TMPRSS2-ERG fusion. Several
groups were able to demonstrate a strong correlation
between ERG overexpression by IHC and ERG fusion status
with over 86% sensitivity and specificity rates [103,104].
In certain settings, ERG immunostaining could be of
utility in establishing the diagnosis of carcinoma when a
pathologist is faced with a small focus of atypical glands on
needle biopsy. In such cases, positive ERG staining in
combination with other immunomarkers such as racemase
(AMACR), p63, and PTEN could be very helpful. As discussed
below, a prognostic role for ERG IHC expression has also
recently evolved in combination with PTEN.
3.3. PTEN loss
The PI3K/mTOR pathway plays an important role in cell
growth, proliferation, and oncogenesis in PCA [105–107].
PTEN is the master negative regulator of the pathway.
Several well-designed retrospective studies have revealed
that loss of PTEN tumor suppressor gene activity, and the
ensuing mTOR pathway activation, is associated with poor
prognosis in PCA.
PTEN loss in PCA tissue samples can be evaluated either
by FISH to detect deletion of PTEN gene allele(s) in tumor
cells or more recently by IHC (Fig. 2). Lotan et al [108]
compared PTEN assessment by IHC and FISH in the largest
existing RP cohort with clinical follow up (13 665 patients).
High concordance between IHC and FISH was documented.
Ninety-three percent of tumors with intact PTEN IHC
Please
Pleasecite
citethis
thisarticle
articleininpress
pressas:
as:Netto
NettoGJ,
GJ,etetal.
al.Prostate
ProstateCancer:
Cancer:An
Eur
EurUrol
UrolSuppl
Suppl(2017),
(2017),https://doi.org/10.1016/j.eursup.2017.10.001
https://doi.org/10.1016/j.eursup.2017.10.001
showed absence of PTEN gene deletion and 66% of cases
with PTEN protein loss by IHC showed PTEN gene deletion
by FISH. Among cases with intact PTEN by IHC, homozygous
(p = 0.04) but not heterozygous (p = 0.10) PTEN gene
deletion was weakly associated with reduced recurrence-
free survival. The data support the utility of PTEN IHC and
PTEN FISH as complementary screening tools for PTEN loss
in PCA [108].
The same group has previously shown IHC screening for
PTEN loss to be particularly useful to identify cases with
heterogeneous PTEN gene deletion especially when present
in a subset of tumor glands. Mutations, small insertions, or
deletions, and/or epigenetic or microRNA-mediated mech-
anisms may lead to PTEN protein loss in tumors with
normal or hemizygous PTEN gene copy number [109].
In a large nested case control tissue microarray, we
initially demonstrated loss of PTEN immunoexpression to
be an independent predictor of biochemical recurrence
(BCR) following RP [110]. Our group subsequently demon-
strated [111] the prognostic role of assessing PTEN
alteration in a surgical cohort of high-risk PCA patients
where its loss was also predictive of decreased time to
metastatic spread. Assessing PTEN in a large cohort of PCA
(over 4750), Krohn et al [112] demonstrated that biallelic
PTEN inactivation, by either homozygous deletion or
deletion of one allele and mutation of the other, occurs in
most PTEN-defective cancers and characterizes a particu-
larly aggressive subset of metastatic and hormone-refrac-
tory PCA. PTEN deletions were present in 20% of PCA (8%
heterozygous and 12% homozygous). PTEN deletions were
associated with early BCR, advanced tumor stage, high
Gleason grade, presence of lymph node metastasis,
hormone-refractory disease, presence of ERG gene fusion,
and nuclear p53 accumulation. The prognostic impact of
PTEN deletion was seen in both ERG fusion-positive and ERG
fusion-negative tumors [112] (Fig. 3). An evaluation of PTEN
and ERG in 1044 incident PCA cases in the Health
Professionals Follow-up Study and Physicians’ Health Study
showed PTEN loss to be associated with lethal progression
of disease; however, this was true only in tumors that were
also negative for ERG fusion [113].
As mentioned above, assessment of PTEN loss is gaining
an important role as an adjunct biomarker to stratify
patients with pathologically low risk disease (Grade Group
1 and Grade group 2; Gleason grade 3 + 3 = 6 and 3 + 4 = 7,
respectively) that are in consideration for active surveil-
lance. Loss of PTEN on needle biopsies in such patients
would argue for a definitive therapy over active surveil-
lance. This is based on multiple studies showing PTEN
loss on needle biopsy to be associated with a higher
likelihood of tumor upgrade and the presence of nonorgan
confined disease on RP [114,115]. In one such study, Guedes
et al [116] showed PTEN loss in a Gleason score
3 + 4 = 7 biopsy to be independently associated with an
increased risk of nonorgan confined disease at prostatec-
tomy (52% vs 27%).
The mTOR pathway is also a potential target of therapy in
PCA. Several rapamycin analogs have been assessed as
potential therapeutic agents for PCA [107,117,118].
AnUpdate
Updateon
onMolecular
MolecularPathology
Pathologywith
withClinical
ClinicalImplications.
Implications.
 EUROPEAN UROLOGY SUPPLEMEN TS XXX (2017) XXX–XXX 5

Fig. 2 – Phosphatase and tensin homolog (PTEN) immunohistochemistry in prostate cancer. (A) PTEN intact in tumor cells ‘‘T’’, note equivalent staining
in adjacent benign glands labeled ‘‘B’’. Lower frame: higher magnification inset shows positive ‘‘P’’ staining tumor glands. (B) PTEN homogeneous loss
in tumor glands ‘‘T’’, with intact staining in nearby benign glands ‘‘B’’ and stroma. Lower frame: higher magnification inset shows negative ‘‘N’’ staining
tumor glands. (C) PTEN heterogeneous loss, with staining loss in some but not all tumor cells ‘‘T’’. Higher magnification inset below shows positive
‘‘P’’staining and negative ‘‘N’’ staining tumor glands. (D) PTEN uninterpretable staining. PTEN stain is weaker but not lost in tumor glands and absence
of background benign glands for comparison makes it difficult to interpret. Lower frame: higher magnification inset shown below. Adapted with
permission from Lotan et al [108].

3.4. Genomic classifications of PCA
In one of the earliest genomic studies, using gene expression
profiling (complementary DNA microarrays), Lapointe et al
[2] were able to identify three subclasses of PCA based on
their distinct patterns of gene expression. The authors
subsequently used array-based comparative genomic hy-
bridization [119] to identify recurrent copy number genetic
alterations that corresponded to three prognostically distinct
groups: (1) deletions at 5q21 and 6q15 group associated with
favorable outcome group, (2) a 8p21 (NKX3-1) and 21q22
(resulting in TMPRSS2-ERG fusion) deletion group, and (3)
8q24 (MYC) and 16p13 gains, and loss at 10q23 (PTEN) and
16q23 groups correlating with metastatic disease and
aggressive outcome. More recent integrated genome-wide
assessment of PCA has further defined various chromosomal
rearrangements and copy number gains and losses, including
ETS gene family fusions, PTEN loss, and androgen receptor
(AR) amplification, that drive PCA development and progres-
sion to lethal, metastatic castration-resistant PCA (CRPC).
Taylor et al [120], using DNA copy number, messenger
RNA (mRNA) expression, and focused exon resequencing
were able to stratify clusters of low- and high-risk disease
beyond the ability of risk stratification by traditional
pathologic prognostic parameters such as Gleason grade.
Six clusters of PCA are identified by unsupervised hierar-
chical clustering with distinct risk for BCR. Markert et al [86]
evaluated mRNA microarray signature profiles in a large
Swedish watchful-waiting cohort of PCA patients and found
mRNA stem-like signatures in combination with p53 and
PTEN inactivation to be associated with very poor survival
outcome. TMPRSS2-ERG fusion group had an intermediate
survival outcome compared with the remaining groups
with a more favorable outcome. The findings were validated
in an independent clinical cohort at Memorial Sloan-
Kettering Cancer Center.
In one of the first studies on whole genome sequencing
(WGS) of PCA, Berger et al [121] analyzed the entire genome
of seven PCA samples obtained from warm autopsy
procedures. The daunting effort brought to light the
occurrence of complex chains of balanced (ie, copy-neutral)
rearrangements within or adjacent to known cancer genes
that on average include 90 rearrangements per genome. This
process of complex rearrangements, termed ‘‘chromoplexy,’’
was further confirmed in a subsequent larger WGS study
performed by the same investigators on 57 PCA samples
[122]. Distinctive patterns of chromoplexy appeared in (ETS
neg
CHD1del) tumors compared with (ETS pos CHD1wt) PCA
(Fig. 4). Tumors with a deletion of CHD1 demonstrated an
excess of intrachromosomal chained rearrangements and
gene deletions clustered in one or two chromosome(s) with
breakpoints concentrated in GC-poor nonexpressed DNA
sites. In contrast, in ETS pos tumors, many single chromoplexy
events joined DNA from dispersed regions of six or more
chromosomes. In the latter group, the chain rearrangements
primarily involved breakpoints enriched near open (actively
transcribed) chromatin, AR, and ERG DNA binding sites. The
latter suggests a link between chromatin and transcriptional
regulation and the genesis of genomic aberrations. Among
the genes that were most frequently disrupted by the
rearrangements are adhesion molecule gene CADM2 and not
surprisingly PTEN and MAGI2, a gene that encodes for the
PTEN interacting protein.
The sequence of oncogenic events in PCA progression has
been further delineated with WGS studies. Genome-wide
germline SNPs coverage permitted the identification of DNA

Please
Pleasecite
citethis
thisarticle
articleininpress
pressas:
as:Netto
NettoGJ,
GJ,etetal.
al.Prostate
ProstateCancer:
Cancer:An
AnUpdate
Updateon
onMolecular
MolecularPathology
Pathologywith
withClinical
ClinicalImplications.
Implications.
Eur
EurUrol
UrolSuppl
Suppl(2017),
(2017),https://doi.org/10.1016/j.eursup.2017.10.001
https://doi.org/10.1016/j.eursup.2017.10.001