REVIEWS

CANCER ORIGINS

An evolutionary perspective on field

cancerization
Kit Curtius, Nicholas A. Wright and Trevor A. Graham
Abstract | Tumorigenesis begins long before the growth of a clinically detectable lesion and,
indeed, even before any of the usual morphological correlates of pre-malignancy are
recognizable. Field cancerization, which is the replacement of the normal cell population by a
cancer-primed cell population that may show no morphological change, is now recognized to
underlie the development of many types of cancer, including the common carcinomas of the
lung, colon, skin, prostate and bladder. Field cancerization is the consequence of the evolution
of somatic cells in the body that results in cells that carry some but not all phenotypes required
for malignancy. Here, we review the evidence of field cancerization across organs and examine
the biological mechanisms that drive the evolutionary process that results in field creation.
We discuss the clinical implications, principally, how measurements of the cancerized field
could improve cancer risk prediction in patients with pre-malignant disease.

Cell lineage
A group of cells that share a
recent common ancestor cell;
also known as a clone.
Epimutations
Gains or losses of DNA
methylation or heritable
chromatin changes; such
changes are in contrast to
genetic changes to DNA
nucleotides.
Cancerized field
A collection of cells that have
gained some but not all the
phenotypic alterations
required for malignancy; in
general, the altered phenotype
will have been caused by an
underlying mutation.
Centre for Tumour Biology,
Barts Cancer Institute,
EC1M 6BQ London, UK.
Correspondence to T.A.G.
t.graham@qmul.ac.uk
doi:10.1038/nrc.2017.102
Published online 8 Dec 2017
Field cancerization is a paradigm for tumorigenesis itself.
Before the growth of a malignant lesion, a normal cell
­lineage can acquire pro-tumorigenic genetic mutations or
epimutations (hereafter collectively referred to as ‘muta-
tions’) that are positively selected for in the microenvi-
ronment of an otherwise healthy organ. Consequently, the
mutant lineage, also referred to as a ‘mutant clone’, can
grow to produce large patches, or fields, of cells that are
predisposed to eventually progress to a neoplasm (FIG. 1).
A cancerized field is a group of cells that are considered to
be further along an evolutionary path towards cancer 1.
The microenvironment surrounding the mutant cells may
also be abnormal and/or be altered by the cancerized field;
the interplay between the mutant cells and their micro­
environment determines which mutations are selected for.
The field may or may not exhibit morphological change
(for example, cancerized cells may look normal or could
exhibit dysplasia). Field cancerization is interchangeably
referred to in the literature as field effects or field defects.
There is evidence that cancerized field development,
with or without overt morphological change, occurs
across tissue types and throughout the body (TABLE 1).
The size of the cancerized field varies substantially,
and a cancerized area of tissue will have a microscopic
morphology classified as noncancerous, that is, normal,
­ yperplasia, metaplasia or dysplasia2. In a technical sense,
h the tissue causes the clonal expansion of some of these
the smallest cancerized field is a single lineage — even a
single cell — that has evolved towards cancer but has not
yet itself become malignant. Together, these data imply
that every cancer arises from a potentially detectable
­cancerized condition.
Pre-malignant diseases — morphologically discern­
ible conditions that increase cancer risk — are examples
of field cancerization with associated morphological
change. Pre-malignant disease is recognized in many dif-
ferent organs, perhaps most notably in Barrett oesopha-
gus (BE)3, ductal carcinoma in situ (DCIS) in the breast4
and prostatic intraepithelial neoplasia (PIN)5. These pre-­
malignant lesions all represent the growth of mutant lin-
eages that are putatively further on an evolutionary path
towards cancer. Because the removal of cancerized fields
is often associated with substantial risk of morbidity, or
even mortality (for instance, in BE6 or inflammatory bowel
disease (IBD)7), they are typically left in situ, and watchful
waiting (longitudinal surveillance typically with biopsy
sample c­ ollection) is performed8,9 (BOX 1).
Discerning the degree of cancerization of a field at
the genetic level requires making the important — but
challenging — distinction between mutant lineages and
cancerized lineages. The former may have many somatic
mutations that are inconsequential to tumorigenesis,
whereas the latter will have mutations that, in an appro-
priate microenvironmental context, increase the rate of,
or drive, cancer development. Mutations accrue steadily
throughout life in ageing tissues10, and the neutral com-
petition (drift) between the adult stem cells that maintains
mutants11; therefore we could broadly consider all mutant
cells to be cancerized, and then we would logically have to
believe that the entire body becomes increasingly cancer-
ized as it grows older. This definition is clearly not specific
enough to be useful.

NATURE REVIEWS | CANCER ADVANCE ONLINE PUBLICATION | 1

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

a b Mutagenic c
insult

Normal cell Cancerized cell Malignant cell

Figure 1 | Characterization and initiation of a cancerized field. a | A sporadic tumour is a lesionNature
that does not share
Reviews its
| Cancer
tumorigenic mutations with surrounding tissue. The mucosa surrounding the tumour is typically morphologically and
phenotypically normal. b | By contrast, a tumour growing in a cancerized field is one that began from a malignant cell that
initially shared tumorigenic mutations and concomitant altered phenotype with the surrounding tissue; these alterations
led to expansion of its ancestral lineage to form a field of cancer-primed cells. Within the field, the malignancy evolved
because of additional mutations (and phenotypic change) in an evolving and potentially cancer-promoting
microenvironmental context. The cancerized field of cells harbouring tumorigenic mutations may appear morphologically
abnormal (for example, dysplastic) or normal, but the cells in the cancerized field will have an altered phenotype — such
as increased survival — that underpins the expansion of the field. c | A cancerized field may be independently initiated by
many cells following a mutagenic insult. Such an insult may provide cells with an advantageous phenotype as a direct
consequence of DNA damage (for example, mutations caused by ultraviolet radiation) within a key gene and/or by
providing an enabling microenvironment (for example, microenvironmental changes induced by cigarette smoke).
These initiated cells, which occur in varying numbers, will clonally expand to create a (mostly) contiguous patch of
cancer-primed cells.

Dysplasia
A feature of tissues in which
cells have an abnormal
morphology or arrangement,
usually regarded as being an
unequivocal neoplastic
alteration.
Hyperplasia
A feature of tissues in which an
increase in cell number occurs
without malignant change.
Metaplasia
A feature of tissues in which
the usual cells of a tissue are
replaced with a cell type that
morphologically resembles
another tissue type.
Inflammatory bowel disease
(IBD). A group of inflammatory
conditions of the bowel that
includes ulcerative colitis and
Crohn’s disease.
Sporadic tumour
A tumour that does not share
tumorigenic mutations and/or
other neoplastic changes in
phenotype with the
surrounding tissue.
Genetically mosaic
A feature of a group of cells
that is composed of two or
more clonal populations, each
with a different genotype.
Rather, we here propose that a cancerized field is
best described by its phenotypic properties. Our revised
definition of a cancerized field is a collection of cells
that have gained some but not all the phenotypic alter-
ations required for malignancy; in general, this altered
pheno­type will have been caused by underlying muta-
tion. These phenotypic changes could include proper-
ties such as an increased growth rate, decreased death
rate or increased immune evasion; consequently, the
pheno­typic changes need not be morphological changes.
These phenotypic changes will have been caused by
mutations in key cancer-associated genes that act either
autonomously or in tandem with an altered microenvi-
ronment. The field need not be monoclonal in origin.
Moreover, this definition avoids the challenge inherent
in p
­ rescribing importance to individual mutations.
We note that cancerized phenotypes may be subtle
and/or transient. For instance, mutation of TP53 (which
encodes p53) in the skin provides a survival a­ dvantage —
a cancerized phenotype — but this phenotype is evident
only in response to ultraviolet exposure12. An initial muta-
tion in the adenomatous polyposis coli (APC) tumour
suppressor gene in the colon is clearly a step towards
colorectal cancer (CRC), and cells within crypts with
mono-allelic APC mutation show increased methy­lation
pattern diversity that is indicative of an increased size of
the intestinal stem cell niche and/or increased stem cell
lineage survival13,14. Our phenotypic definition excludes
regions of tissue having DNA damage or mutations that
do not yet have a cancer-related pheno­type, excluding
even oncogene or tumour suppressor gene mutations that
do not cause a cancer-­related pheno­type until additional
mutations emerge. Nevertheless, we suspect that most,
if not all, cancer-­associated mutations will have pheno-
typic consequences in normal tissue — at least under
particular environmental ­conditions — and furthermore
that the altered pheno­type of the cancerized lineage will
be the mechanistic cause of its expansion. Consequently,
a cancer arising from a field passes through multiple
pheno­type states, whereas the lineage leading to a s­ poradic
tumour may not have had any p ­ henotypic changes before
cancer growth.
Inherited cancer risk — for example through the
inheritance of a defective copy of a tumour suppressor
gene — also leads to cells in the body being predisposed
to cancer. We reserve the term ‘field cancerization’ to
apply to only a somatic evolutionary process that pro-
duces cells that are ‘closer to’ cancer. Genetically mosaic dis-
ease, resulting from mutation and consequent pheno­typic
changes arising during development, is an i­ntermediate
example of field cancerization15.
In this Review, we discuss what combination of
geno­types and phenotypes constitutes a cancerized
field, how such fields are initiated and how they expand,
and end by considering the clinical implications of
field cancerization.
A brief history of field cancerization
Historical definition of a cancerized field. In 1953,
Slaughter et al. first proposed the concept of field can-
cerization in order to explain the statistical enrichment
of the multifocal (rather than isolated) oral squamous
carcinomas they found in 783 patients with cancer 16. The
authors hypothesized that the patches of microscopically
abnormal (but histologically benign) tissue surround-
ing the tumours suffered from the “preconditioning of
an area of epithelium to cancer growth by a carcino-
genic agent” (REF. 16). In the decades that followed, the
advent of molecular biology fostered genetic analyses
that revealed that fields of epithelial cells predisposed

2 | ADVANCE ONLINE PUBLICATION www.nature.com/nrc

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

to malignancy were, in some cases, composed of a clone
derived from a mutated cell2. In one striking example,
a patient was reported to have a clonal expansion of a
single TP53 mutation across both lungs17. Thus, the defi-
nition for field cancerization was revised by Braakhuis
et al. to be “the growth of a mutant clone to produce a
field of cells predisposed to subsequent tumour growth”
(REF. 1). To frame this previous idea from an overtly evo-
lutionary perspective, we need only note that the mutant
clone that expands to form the field is fitter (more likely
to produce surviving offspring) than the resident cell
population. We note that this previous definition implies
that the cancerized clone has an altered phenotype that
drives its expansion and thus is consistent with our
­current definition.
More recently, fields of epithelial cells carrying
somatic epimutations (abnormal DNA methylation)
have been noted, for example, hypermethylation of
various promoters in a Helicobacter pylori-infected
stomach18 or similar patterns of unusual DNA methyla-
tion in synchronous, spatially separate CRCs19. In such
cases, the high rate of de novo DNA methylation, the

Table 1 | Examples of field cancerization in human tissues

Cancer type Pre-malignant diagnosis Evidence Clinical importance Comments Refs
(recommended action)
Colon carcinoma in IBD, dysplasia Mutations, aberrant Yes (resection of all Seen in both ulcerative colitis 32,33,
patients with IBD methylation, altered affected area necessary) and Crohn’s disease; can involve 137,138
proteomics whole colon
Colon carcinoma in Adenoma Mutations, aberrant Unknown (removal) Importance unclear; unknown 139–146
non-IBD patients methylation fraction progress
Gastric carcinoma Intestinal metaplasia Mutations, aberrant Yes; occurs in intestinal Not all metaplasias progress 18,66,
methylation metaplasia, a known 147–151
pre-malignant state
(surveillance)
Oesophageal Squamous dysplasia Mutations, aberrant Yes; high possibility of Smoking-associated 152–158
squamous methylation, altered metachronous lesions
carcinoma proteomics (surveillance)
Oesophageal BE, dysplasia Mutations, altered Yes (surveillance) Early evidence implicated the 111,
adenocarcinoma metabolomics development of a cancerized 159–161
field, but later studies showed
carcinoma developed focally
with no selective sweeps
Non-small-cell Dysplasia, carcinoma in situ Mutations Yes (review after Smoking-associated; can be 17,46,
lung squamous resection important) very widespread; amenable to 162–166
carcinoma screening
Non-small-cell lung AAH, AIS Mutations Possible (none) A continuum exists between the 167–171
adenocarcinoma precursor lesions AAH and AIS
and adenocarcinoma
Small-cell lung Diffuse idiopathic Mutations Possible (none) Smoking-associated; can be 167–170,
carcinoma pulmonary neuroendocrine very widespread 172
cell hyperplasia
HNSCC (oral, Leukoplakia, erythroplakia, Mutations Yes; high possibility of These are squamous carcinomas 1,16,
oropharynx, oral submucous fibrosis, metachronous lesions of the oral and pharyngeal 173–179
hypopharynx and lichen planus, dysplasia (surveillance and mucosa
larynx carcinoma) resection)
Breast carcinoma DCIS, ductal hyperplasia Mutations, aberrant Yes; possibility of Explains the presence of 180–184
methylation metachronous lesions multiple areas of DCIS
(resection)
Prostate carcinoma PIN Mutations, altered Possible (surveillance) Multifocal carcinomas of 185–187
proteomics, aberrant prostate are common
methylation
Bladder carcinoma Intestinal metaplasia, Mutations Yes (surveillance, Multifocal tumours are common 188–190
dysplasia, carcinoma in situ resection and therapy)
Skin carcinoma Actinic keratosis, Bowen Mutations Yes (surveillance and Common in face and scalp; 22,
disease treatment) the whole field is treated 191–194
Melanoma Acquired nevi, dysplastic Mutations Yes (surveillance and Suspect lesions are resected 195
nevi (atypical nevi) and treatment)
congenital nevi
Blood cancer None Mutations Yes (surveillance) Clones seen in a substantial 92–94
percentage of newborn babies
AAH, atypical adenomatous hyperplasia; AIS, adenocarcinoma in situ; BE, Barrett oesophagus; DCIS, ductal carcinoma in situ; HNSCC, head and neck squamous
cell carcinoma; IBD, inflammatory bowel disease; PIN, prostatic intraepithelial neoplasia.

NATURE REVIEWS | CANCER ADVANCE ONLINE PUBLICATION | 3

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Box 1 | Pre-malignant disease: a resource for cancer evolution studies
Fortuitously for the research community, the clinical practice of watchful waiting for
pre-malignant lesions leads to extremely well-characterized longitudinal tissue
catalogues that facilitate the temporal study of clonal evolution in situ in humans. Such
study can reveal the pace and pattern of cancerized clone evolution and/or phenotype
generation and spread109,134, and this information can be directly leveraged to optimize
biomarker development and screening regimens135. The exciting multi-institution
Pre-Cancer Genome Atlas (PCGA) aims to compile a comprehensive characterization
of the molecular alterations in pre-malignant lesions as well as, importantly, the
corresponding changes in the microenvironment and cellular phenotypes136. It is also
crucial that the PCGA contains longitudinal data so that the context-specific nature
of cancer evolution and the concomitant intricacies of the genotype–phenotype
mapping, together with the critical rates of cancer evolution, can be fully elucidated.
potential for demethylation and the likelihood of con-
vergent epi­genetic evolution mean that the clonality of
the e­ pigenetic fields is open to question. Instead, the can-
cerized field is likely to have formed because of exposure
to a prevalent mutagen and promoter of clonal expan-
sion (for example, in the case of the H. pylori-infected
stomach) and/or subsequent convergent evolution of
the epigenome (for example, in the case of synchronous
CRC). Thus, field cancerization can occur because of
multiple independent clonal expansions. Nevertheless,
mutational diversity that exists within the cancerized
field is likely a substrate for natural selection, and so over
time, it is likely that the fittest clone (for example, that
with the most adaptive phenotype) will come to dom-
inate the field. Genetic drift will also shape the clonal
composition of the field20. Related to this, others have
suggested that exposures that have pleiotropic effects on
cell phenotypes and microenvironmental inter­actions
across many organs — such as smoking — should be
thought of as causing a body-wide ‘aetiological field
effect’ (REF. 21) that is clearly not clonal in origin.
Alterations of stromal cells can also promote field
cancerization. In the skin, epigenetic modification of
fibroblasts is induced by ultraviolet exposure, leading to
the production of diffusible growth factors, inflammatory
cytokines and matrix-remodelling enzymes that together
predispose to subsequent epithelial tumours22. In this
particular example, the resulting epithelial tumours
show recurrent genetic alterations, implying that the can-
cerized stromal compartment alone is unable to cause
a tumour but provides selective pressure for particular
epithelial genotype–phenotype combinations. Because
the stroma does not transform into a tumour itself, here,
we do not consider the stroma itself to be cancerized.
Field cancerization driver mutations. In cancer genom-
ics, a distinction is commonly made between driver
mutations, which confer growth or survival advantages
on tumour cells (within the appropriate microenvi-
ronment) and passenger (neutral) mutations, which
passively accumulate in cell lineages23–25. Borrowing
that nomenclature, here, we term the mutations that
are functionally important for the phenotypic changes
Driver mutations
Mutations that are positively
that underpin the cancerized field ‘field cancerization
selected for and are implicated drivers’. We note that the phenotypic consequences
in tumorigenesis. of a driver mutation may be context dependent, so
simply having a field cancerization driver may not be
sufficient to cause a phenotypic change in the current
­microenvironmental condition.
In skin (squamous epithelium), putative field cancer-
ization driver mutations are found in genes, including
Notch family members and TP53. Mutations in these
genes are detected at high frequency in apparently normal
sun-exposed skin cells26. Mouse models show that Notch
family mutants are clonally selected for in oesopha­
geal squamous cell epithelium owing to their ability to
inhibit differentiation27. TP53 mutation has been long
recognized to be present in nondysplastic skin, and its
frequency increases with age28 and causes an increase in
the stochastic growth rate of the clone, but only in the
presence of chronic ultraviolet exposure12. Similarly, lin-
eage tracing in mice has shown that Hedgehog pathway
mutations are positively selected for in skin because they
inhibit differentiation29. Remarkably, the density of clon-
ally expanded putative driver alterations in skin was esti-
mated at ~140 driver mutations per square centimetre26
— for example, more driver mutations can be found in
a small patch of morphologically normal skin than in a
skin basal cell carcinoma30.
In the inflamed intestine, mouse models show that
Trp53 mutation is a field cancerization driver because it
increases stem cell replacement in a chronic inflammatory
setting31. Interestingly, in the absence of inflammation, the
Trp53‑mutant clones have no advantage31, indicating
the interplay between mutant lineage and microenviron-
mental context in determining selective advantage of the
phenotype induced by a field cancerization driver muta-
tion. Correspondingly, patients with IBD show clonal
expansions of TP53‑mutants in nondysplastic bowel32,33,
and we are unaware of reports of clonal expansion of
TP53‑mutants in the colon mucosa in patients without
IBD. Patients with IBD also show clonal expansion of
­KRAS-mutant and/or aneuploid cells in the absence
of dysplasia32,33.
To determine precisely which mutations are field can-
cerization drivers requires in vivo lineage tracing, pref-
erably in both model systems and human tissues. Such
model systems can reveal the mechanism that provides a
clone with a selective advantage (as per the above exam-
ples). Nevertheless, we note that the exact phenotypic
consequences of many putative cancer driver mutations
remain elusive, and their common occurrence in cancers
may in fact be because the driver is selected for before
overt cancer growth, that is, during field cancerization.
In cancers, broadly speaking, a mutation will be called a
driver if it occurs more frequently across tumours than
is expected from the mutation rate alone34 — the excess
frequency is then a consequence of positive selection for
the mutation. However, this analysis does not show when
during the tumorigenic process the driver mutation was
positively selected for or if it is still under positive selec-
tion within the tumour at the time it is observed. In other
words, the so‑called cancer driver may not actually drive
cancer growth per se but rather had previously driven the
growth of an ancestral lineage. Recognizing that driver
mutation selection is inherently context-dependent and,
moreover, that the context changes over time provides a

4 | ADVANCE ONLINE PUBLICATION www.nature.com/nrc

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

Epistasis
The interaction of multiple
genes that leads to the
development of a phenotypic
trait.
Polyclonal
An attribute of a lesion that is
derived from two or more
clones as opposed to a
monoclonal origin.
convenient explanation for the high frequency at which
prominent cancer-related genes are found in cancerized
fields35. We also note that in some instances, the same
phenotype will be positively selected for in both the
cancerized field and in a cancer itself: for example, func-
tional TP53 mutations appear to drive cancerization of
the inflamed intestine31, likely because they provide a sur-
vival and/or growth advantage, and these same pheno­
types are clearly also important for subsequent cancer
development 36. Hence, a field cancerization driver may
also be a cancer driver.
Progression to cancer likely involves the accumula-
tion of multiple field cancerization driver mutations37.
Phenotypic change is caused by either individual
high-impact mutations or epistasis among a synergis-
tically acting group of mutations, each of which would
be insufficient to cause a change to a cancerous pheno­
type without the others. Alternatively, progression may
be caused by sudden large-scale mutational events that
simultaneously alter large parts of the genome and
elicit large phenotypic change: whole-genome doubling
appears to apply a transformative role in BE, for exam-
ple38,39. If mutations accrue gradually, then conceivably,
the field cancerization driver burden can be used to
identify patients at risk of progression, whereas predict-
ing the likelihood of such ‘catastrophic’ events presents a
larger challenge.
Mapping genotype to phenotype. Nuances in cell mor-
phology and behaviour mean that cells within a cancer-
ized field have phenotypes that theoretically occupy an
infinite phenotype space, where position in phenotype
space is determined partly by the underlying genotype
(FIG. 2). Understanding the genotype–phenotype map is
critical for understanding the biology and evolution of
cancerized fields. The example of TP53 and other driver
genes given above shows that the genotype–phenotype
map is many‑to‑many (there is one or more genotype
that underlies each phenotype, but a genotype may incur
different phenotypes on the basis of nongenetic effects);
hence, the notion of a ‘genetic blueprint’ that entirely
determines the biology of a cancerized field is overly
simplistic. Pathological diagnosis categorizes the mor-
phological features of pre-malignant disease into a small
number of groups (for example, grade) — see TABLE 1.
However, we note that the underlying multivariate nature
of genotypes and phenotypes means that such classifi-
cations are in fact continuous and fluid. This continuity
perhaps explains some of the evident challenges in reli­
ably identifying particular pre-malignant states, such as
a low-grade dysplasia in BE40,41.
Initiation of the cancerized field
Cancerized fields may be induced by ‘unlucky muta-
tions’42 that occur owing to natural DNA replication
errors during ageing 43 or by mutagenic insult (FIG. 1c).
Immediately following the mutagenic insult, the mutated
field will be composed of many genetically distinct
clones, but over time, we expect that the clones with the
fittest phenotypes will come to dominate the cancer-
ized field44. If the mutagenic insult is ongoing, perhaps
because it is caused by a persistent environmental factor
such as tobacco carcinogens, diet or infections21, then
new clones will be continually generated, and the field
will continue to appear genetically mosaic.
For example, mutant cancerized fields, often bear-
ing mutations in TP53, are seen in the lung and air-
ways, and these mutations are associated with smoking
exposure45,46. In the cervix, high-risk strains of human
papillomavirus impair normal p53 function, a condition
that is likely permissive for the development of a field of
preinvasive cervical intraepithelial neoplasia47. H. pylori
infection is associated with field cancerization via intes-
tinal metaplasia in the stomach48, but recent whole-­
genome sequencing on 100 gastric tumours found no
association with mutational spectrum and H. pylori
infection49; this suggests that infection induces intestinal
metaplasia fields via a nongenetic route, perhaps by pro-
viding a selective pressure for atrophic gastritis-­resistant
clones that bear epigenetic changes. This intriguing
result highlights the need for further study to elucidate
the role(s) of H. pylori infection in the formation of
cancerized clones in the stomach. In the oesophagus,
gastroesophageal reflux disease (GERD) increases the
risk of oesophageal adenocarcinoma50, and a mutational
signature thought to be characteristic of acid exposure
is observed in the genomes of these cancers51, implying
that the cancer arose from a field mutationally affected
by GERD.
Accelerated ageing — the increase in the rate of events
attributable to ageing — is implicated in tumorigenesis
in many tissues. Indeed, the majority of high-frequency
mutations found in a cancer (of which there are typically
very many) is thought to accrue before the initiation of
tumour growth52, and it seems likely that these would
often be within a cancerized field. In patients with ulcer-
ative colitis, mutational signature analysis53, CpG island
hypermethylation54 and telomere shortening 55 have indi-
cated accelerated ageing in the colon. Utilizing epigenetic
drift as a molecular clock shows that the ages of BE tissue
vary dramatically between patients of similar chronologi-
cal age56. Moreover, just as chronological age is recognized
as one of the strongest predictors of cancer risk, biological
tissue age (that is, the time a tissue has had to accrue addi-
tional stochastic alterations) contributes to field canceriza-
tion, and particular attention should be given to exploring
its roles in neoplastic progression57.
It is noteworthy that some early neoplastic lesions,
such as adenomas in the colon58 and nevi in the skin59,
are derived from multiple independent clones — that is,
the neoplastic lesions are polyclonal in origin2. While the
mechanism that causes this polyclonality is unknown,
a localized field defect provides a compelling explana-
tion: an initial mutant clone may produce some kind of
field (through aberrant signalling, for example) that alters
the behaviour of surrounding stromal cells, creating an
environment that in some way promotes mutations in
neighbouring cells60. Indeed, in the colon, transformed
intestinal crypts have profound effects on neighbouring
wild-type crypts61,62. Physical compression of intestinal
mucosa induces WNT pathway signalling, and so the
signalling induced by an initial expanded mutant clone

NATURE REVIEWS | CANCER ADVANCE ONLINE PUBLICATION | 5

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

can result from mechanical cues rather than juxtacrine
or paracrine signalling 63. Thus, the multitude of inter­
actions between clones is important in the establishment
of cancerized fields.
Mechanisms of clone growth
Human epithelia are organized into two broadly dif-
ferent types, glandular and squamous, and we discuss
how clones grow in each epithelium type below. Tissue
architecture constrains evolution by limiting the ability
of mutant clones to expand and in so doing lengthens the
waiting time to cancer 64.
Glandular epithelium. In order for a clone to grow in
glandular tissue, such as intestine and BE mucosa, it
must first undergo niche succession, whereby a mutant
stem cell in the gland replaces all other stem cells (and
so subsequently, all differentiated cells in the gland
will also be mutant); second, the repopulated mutant
gland must produce a field of mutant glands by gland
fission13,65–67 (FIG. 3). The unit of selection in glandular
tissue therefore switches between individual stem cells
within a gland to the meta-population of the gland
as a whole (the differentiated cells in the gland are
the somatic lineage derived from the germ stem cells

SMAD4
TP53
Genotype CDKN2A
dimension 2 CDKN2A
TP53
CDKN2A
N

Genotype– Genotype
phenotype dimension 1
map

M
Phenotype C
dimension 2
D

N H

Phenotype dimension 1

b WGD c
TP53 TP53
Genotype
dimension 2
TP53

Genotype dimension 1
Epigenetic and
microenvironmental
influences

End End
Phenotype
dimension 2

Start Start

Phenotype dimension 1
Nature Reviews | Cancer

6 | ADVANCE ONLINE PUBLICATION www.nature.com/nrc

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

in the gland). Measurement of gland division rates and
the impact of tumorigenic mutations upon the rate are
generally lacking. In the healthy colon, gland fission
rates are very low — once every few decades13 — and so
it is clear that a field cancerization driver mutation that
clonally expands to form a large patch of glands can do
so only by causing a large increase in the fission rate.
Squamous epithelium. Epithelial cells within squa-
mous tissue, such as epidermis, cervical, squamous oral
and oesophageal squamous mucosa, expand by basal
replacement of neighbouring stem cells12,27,68,69. During
normal homeostasis, basal cells proliferate, producing
differentiated progeny that go on to form the superfi-
cial layers of the epithelium and also new basal cells
that compete neutrally to grow a patch by the lateral
replacement of one progenitor cell by another 70 (FIG. 3).
Tumorigenic mutations, such as in TP53 and Notch
family genes, tilt the competition in favour of the
­expansion of mutant clones12,27 (detailed above).
Localized clonal expansion is the canonical mech­
anism of mutant field formation, but there may also be
noncanonical methods. Long-range stem cell replace-
ment occurs in the Drosophila ovaries71, and similar
within-organ metastasis could occur in the human body:
distant engraftment of genetically labelled organoids
in the intestine provides a proof‑of‑concept of such a
mechanism72. High-resolution lineage tracing studies73,74
where there is ongoing frequent genetic labelling that
provides finer resolution of individual somatic lineages
are required to reveal the frequency of long-range stem
cell migration that is possible in both glandular tissue
and squamous tissue.
◀ Figure 2 | Evolutionary trajectories of cancerized clones. a | Common evolutionary
pathways of pretumour progression in Barrett oesophagus (BE)37. The genotype–
phenotype map guides the evolutionary path of cells in the two abstract infinite spaces
of the genotype and phenotype. Progression from normal to a field-cancerized state
and eventually to cancer is underpinned by changes in genotype; these changes are
due to random somatic mutation (black filled arrows to pink points) that is then
projected by the genotype–phenotype map, which incorporates cells’ genetic
mutations, epimutations, and microenvironmental context to produce a specified
phenotype (dashed lines to orange points). Natural selection acts in the phenotype
space to shift the predominant phenotype of the clone population to one of higher
(contextual) fitness (dashed lines with unfilled arrowheads to purple points). Pink ovals
represent regions of similar phenotype (N, normal tissue; M, metaplasia; H, hyperplasia;
D, dysplasia; C, cancer). The genotype of fit parents is deterministically preserved back
in genotype space (dashed lines to blue points). The diagram highlights the overlap in
defined phenotypic regions — for example, some genotype–phenotype combinations
could be histologically classified as either metaplastic or dysplastic. b | Punctuated
evolution. Here, in contrast to the gradual phenotypic evolution of part a, the
ostensibly normal clone carries a TP53 mutation, which then undergoes
whole-genome doubling (WGD) and causes a large change in both genotype and
phenotype, as has been seen in BE. c | Small changes in genotype space may also lead
to large changes in phenotype space, possibly owing to rare variants having large
phenotypic effects, epistatic effects between accumulated mutations and/or other
epigenetic or microenvironmental effects. For example, a TP53 inactivating mutation
may occur in a morphologically normal clone but not be selected for until subsequent
small events accumulate to promote large phenotypic changes. In summary, the
degree of phenotypic change need not be proportional to the degree of underlying
genetic change. CDKN2A, cyclin-dependent kinase inhibitor 2A; SMAD4, mothers
against decapentaplegic homologue 4.
What is selected for? Broadly speaking, phenotypes
associ­ated with increased growth and/or increased
survival will be positively selected for. In squamous
lesions, cells that are best able to adhere to the basement
membrane will have increased survival because the lin-
eage will not migrate and differentiate, whereas faster-­
proliferating basal cells have more opportunities to
replace neighbouring cells. Similarly, in glands, upregu­
lation of crypt fission (for example, with clonal expan-
sion) and increased ability to survive e­ nvironmental
insult are traits expected to be under selection.
In the noninflamed colon, KRAS mutation accel-
erates the rate of crypt fission and is therefore selected
for 75. The cycles of wounding and repair that occur in
the colonic mucosa of individuals with IBD are expected
to provide a strong selective pressure, first for clones that
can survive the damage and then for fast-proliferating
clones that can rapidly heal the wound76. This selection
within the cancerized field provides a ready explanation
of the genetic differences between sporadic and colitis-­
associated CRC53. BE cells produce mucins that provide
protection against insult from gastric acid reflux; hence,
BE may be an adaptive response that is selected for in
response to reflux 77. Indeed, mathematical modelling
of oesophageal adenocarcinoma progression suggests
that age-dependent GERD symptoms affect not only BE
onset but also proliferation rates of high-grade dysplasia
due to acid-induced injury 78. Together, these examples
show how field cancerization represents an evolution-
ary trade-off between wound repair and cancer devel-
opment: faster-healing wounds may come at the cost of
field cancerization.
The mutation burden of apparently healthy tissues
increases with age10 — and consequently, aged epithelia
are a patchwork of different clones. Clonal competition,
or other interactions between these clones, may have an
important influence on progression to cancer 79. Clonal
competition in general slows adaptation, so it may have
a cancer-suppressive effect 64,80.
Role of the tissue microenvironment
Stromal reprogramming is increasingly recognized to
play a critical role in tumorigenesis81. In general, the stro-
mal microenvironment is a key regulator of self-­renewal
in the epithelium82, and so aberrant changes in the stroma
can enable field cancerization of the epithelium.
Aberrant stroma (measured by fibroblast phenotypes
and immune cell population composition) is indeed
found in pre-malignant disease. For example, patients
with IBD are at higher risk of CRC, and the severity of
inflammation is an independent predictor of cancer
risk83. In BE, the stromal compartment shows consist-
ent changes in gene expression that also have prognostic
value, and moreover, these changes are somewhat similar
among pre-malignant conditions along the gastrointesti-
nal tract 84. In skin, loss of Notch family signalling in the
mesenchymal cells of transgenic mouse models induces
epithelial tumorigenesis, demonstrating how stromal
changes enable field cancerization22. In the breast, trans-
genic manipulation and cancer-associated fibroblasts
have revealed a critical role for the stromal lineage in

NATURE REVIEWS | CANCER ADVANCE ONLINE PUBLICATION | 7

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Stem cell
Somatic alteration Wild-type cell Mutated cell

b c d

Progeny cells Basal cell Mutated cell

Figure 3 | Morphological mechanisms and examples of clonal expansion. a | Stem cell niche succession and fission in
glandular tissue. A single stem cell in the niche acquires a somatic alteration (blue) and may take Nature
over theReviews
entire crypt
| Cancer
population (‘niche succession’) either via neutral genetic drift or via selection, by which the cell with an advantageous
mutation outcompetes the other cell types (brown) to be the source for all future somatic cells in the crypt. The
monoclonally converted crypt then bifurcates, beginning at the niche compartment and producing two mutated crypts.
b | Expansion of a clone in the human colon. A histological en face slide depicts an entirely cytochrome c oxidase
(COX)-deficient crypt (blue) undergoing fission in the colon of a person with familial adenomatous polyposis syndrome.
The two resulting daughter crypts both are COX-deficient and are surrounded by COX-wild-type crypts (brown).
COX-deficiency is a somatically acquired event that usually results from an underlying somatic mutation in the
mitochondrial DNA; consequently, COX-deficiency is a natural heritable mark that facilitates lineage tracing in human
tissues. c | Transverse section from a biopsy sample of a patient with inflammatory bowel disease stained for COX activity
showing two crypts that are part of a single COX-deficient clone. In glandular tissues such as those found in the stomach,
Barrett oesophagus and the colon, clones such as this may continue to grow into a large patch of mutant glands.
d | COX-deficient human epithelial clone expanding in the human airway. Immunofluorescence staining for COX (green)
is shown with counterstains for the mitochondrial marker porin (red) and the nuclear marker 4ʹ,6‑diamidino‑2‑phenylin-
dole (DAPI; blue). COX-deficient cells stain red; COX-proficient cells stain green. A clone of COX-deficient cells has
expanded to form a small mutated patch of cells. The few COX-deficient cells that are separate from the main patch are
likely part of the same clone as the main patch but became separated as the clone grew. e | Expansion of a clone in
squamous epithelium. A single basal cell acquires a somatic alteration (dark red) that confers an advantageous
phenotype, leading the mutant clone to laterally replace neighbouring basal cells (dark green). These basal cells then
produce more superficial progeny cells (lighter red). This process, which occurs in tissues such as the lung, oesophagus
and skin, leads to a patch of clonally derived (and phenotypically altered) cells. Parts b and c are courtesy of Stuart
McDonald, Barts Cancer Institute, Queen Mary University of London, UK. Part d is courtesy of Samuel Janes, University
College London, UK.

8 | ADVANCE ONLINE PUBLICATION www.nature.com/nrc

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

modulating epithelial cell behaviour 85,86. Moreover,
in the colon, for example, genome-wide association
studies for CRC have identified bone morphogenetic
protein (BMP) family members that are regulated and
expressed by stromal cells as key modulators of cancer
development risk87.
Microenvironmental changes enable field can-
cerization in the stroma by altering the fitness effects
of m
­ utations in epithelial cells (reviewed in REF.  88)
and so promote expansion of cancerized lineages.
Microenvironmental factors provide selective pressures
for phenotype adaptation: within the cancerized field,
cells explore the adaptive landscape (via genetic muta-
tion or phenotypic plasticity), and the genotypes of
future generations reflect the phenotype that successfully
surmounted any microenvironmental challenges (FIG. 2).
From this theoretical ‘fitness landscapes’ perspective,
Gatenby and Gillies proposed a model that identified six
such micro­environmental barriers for the development of
a malignant phenotype: apoptosis with loss of basement
membrane contact, ­inadequate growth promotion, senes-
cence, hypoxia, acid­osis and ischaemia 89. Growth,
senescence and cell death are clearly also influenced by
cell-intrinsic factors (for example, oncogenic mutations)
but, nevertheless, can also be regulated by intercellular
signalling (that is, by the microenvironment)90. It is not
simply serendipitous that a cancerized clone is found in a
particular ecosystem — clonal adaptation can occur only
owing to the current selective microenvironment.
Clinical implications
The key clinical issue presented by field cancerization is
that of accurately determining the risk of cancer devel-
opment from a cancerized lesion, as putatively cancer-
ized fields are very common but few actually progress
to cancer. Thus, simply assuming that the detection of
a cancerized field is evidence of likely cancer develop-
ment carries the hazard of overdiagnosis of cancer risk,
and moreover, subsequent interventions to ameliorate the
risk in most cases constitute overtreatment (with associ-
ated risks of unnecessary morbidities)91. Take the blood,
for example: 1% of all newborn babies have cells with an
acute lymphoblastic leukaemia-associated mutation and
histopathologically identifiable precursor lesions that are
present at ~100 times the corresponding rates of clinical
cancers92, and moreover, such early genetic events can be
shared between monozygotic twins, indicating that the
pre-malignant clone formed in utero93. However, bone
marrow transplant to remove the cancerized field would
seem obscene, given that the intervention is unneces­
sary in 99% of patients. In adults, 10% of unselected
individuals over the age of 65 show clonal expansion of
cancer-associated mutations in their blood94. This appar-
ent field cancerization in the blood increases cancer risk
13‑fold94, in this case indicating that detection of a can-
cerized field is a biomarker of cancer risk. Nevertheless,
the absolute risk of haematological cancer development
in those patients with clonal haematopoiesis was still
found to be very small (~1% annual incidence rate)94, so
detection of the field itself will overdiagnose cancer risk
in the majority of people. Clearly, there is an acute need
for biomarkers that can distinguish cancerized fields that
will remain indolent for a long time from those that will
rapidly become malignant.
The pitfalls of pathology. Conventional pathologic grad-
ing remains the most common biomarker of cancer risk
in conditions associated with field cancerization (for
example, IBD and BE) and in morphologically discern-
ible pre-malignant disease (for example, DCIS or PIN),
but in general, the predictive value of pathological sta­
ging of pre-malignant disease is low. For instance, non­
dysplastic BE has been estimated to incur an ~0.25%
annual risk of progression to adenocarcinoma95, whereas
high-grade dysplasia carries a near 25‑fold increase in
risk at ~6% annual risk of progression96; however, this still
means that more than nine out of ten patients with BE
who are diagnosed with a dangerous high-grade lesion
will not progress to cancer in the next year. The annual
risk of squamous carcinoma of the skin in patients with
actinic keratosis ranges between 0.15 and 80%97, and pub-
lished estimates of annual cancer incidence in patients
with BE who have low-grade dysplasia range between
0.15 and 36%98. This tremendous variability is due to a
multitude of factors, including interobserver variability
in grading 99, sampling bias (higher-grade lesions may be
missed at biopsy 100) and patient cohort differences97,98.
Moreover, given that field cancerization can occur with-
out associated morphological change, evidence of evolu-
tion along a trajectory towards cancer cannot reliably be
assessed by pathology alone.
Molecular biomarkers. Molecular profiling for sensitive
detection of cancerized fields at high risk of developing
malignancy is highly desirable (BOX 2). As for molecular
prognostication of established cancer, the ideal is to find
a particular (small and easily assayable) set of molecular
features (mutation status, RNA expression and/or pro-
tein level) that predict those patients at risk of progres-
sion with high sensitivity and specificity. For example,
a number of different molecular panels show predictive
value for oesophageal adenocarcinoma in patients with
BE101–103 and for lung cancer in current or former smok-
ers104,105. In the lung, gene expression within the prox-
imal airway indicates distal cancer, perhaps as a direct
consequence of a cancerized airway 105. Unfortunately, the
studies using such panels typically find that the nega­tive
predictive value is low; the panels are only moderately
successful at predicting which patients will not get can-
cer in the future. While increasing the size of the bio-
marker panels may increase their predictive accuracy,
panel size alone will not address the complicating factors
of inter­patient heterogeneity (not all patients will follow
the same molecular path to cancer, so a rare dangerous
path may not be assayed by the biomarker panel) and
intra­patient heterogeneity (whereby a biopsy may fail to
sample a dangerous lesion).
Instead, we emphasize that biomarkers based on
quantification of the underlying evolutionary process
occurring within the cancerized field (rather than on
measurement of a molecular feature that may or may
not be an outcome of that process) may have pan-tissue

NATURE REVIEWS | CANCER ADVANCE ONLINE PUBLICATION | 9

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Box 2 | Deriving biomarkers for cancer risk prediction

Traditionally, tissue morphology has been used as the sole indicator that a patient has a certain risk of developing a
malignancy. More recently, the presence or absence of particular molecular features has been used as a biomarker.
However, measures of the evolution of the cancerized field itself, such as genetic diversity107, may also act as biomarkers
for the risk of cancer development. Measuring the field as a whole removes the risk of sampling bias. Serial sampling and
spatially extensive sampling in longitudinal analyses38,109 will continue to provide better insight into the pace of evolution
and thus aid prognostication.
Below, we provide a table of cancer biomarker types alongside a few of their benefits, pitfalls and examples.

Biomarker Benefits Pitfalls Examples

Tissue Standardized histopathology Grey areas of pathological staging; Metaplasia, dysplasia,
morphology methods; large studies completed interobserver variability; missed carcinoma in situ
for validation lesions on exam
Molecular Some genetic defects have higher Single time-point data may not DNA mutations, gene
marker predictive value than morphology reveal event ordering; marker expression patterns,
(single or a (for example, TP53 biallelic sometimes missed in biopsies; protein levels, viral
panel) inactivation in Barrett oesophagus); panels do not typically acknowledge load
ever-decreasing costs marker–marker interactions
Evolutionary Markers of heterogeneity more Some markers require serial Genetic diversity,
measure ubiquitous than single genetic sampling to accurately validate rates clonal expansions,
changes; measures evolutionary of change; limited studies available change over time,
trajectory itself at this time biological tissue age

prognostic value that is robust to the problem of inter-
patient and intrapatient molecular heterogeneity 106.
One such ‘evolutionary biomarker’ is the degree of
genetic diversity within the cancerized field. Genetic
diversity is a proxy measure of field evolvability, as a
more diverse field is more likely to contain a lineage
that is preadapted to a new selective pressure107. Indeed,
genetic diversity has been shown to be prognostic in
two different cohorts of patients with BE, even when
using different molecular tools to quantify the d ­ iversity
present 38,108,109. The potential utility of clonal diversity as
a prognostic measure across disease types was under-
lined by a recent pan-cancer study 110. Similarly, genetic
instability, as another measure of evolvability, has prog-
nostic value in BE111 and potentially also in colitis112.
A second evolutionary measure is the size of clonal
expansions, as this provides both a proxy assay for the
size of the cancerized field and evidence for positive
selection of clones within the field. Detection of clonal
expansions has prognostic value in ulcerative colitis113
and BE111 and in the blood94. Similarly, changes in lesion
composition over time could have prognostic value as
a marker of active evolution of the field; quantification
of the mutation rate (a proxy measure of the rate at
which new adaptive lineages are produced) and rates
of clonal expansion (how quickly a new high-­cancer-
risk clone grows to a large size) may hold particular
prognostic value.
Optimizing screening schedules. The intervals between
cancer screens may be optimized using knowledge of the
biology of cancerized fields. If a field is rapidly evolv-
ing, then frequent screening may be required, whereas
evolutionarily quiescent fields could be surveyed less
often. For instance, in BE, longitudinal measurement of
the clonal composition has revealed that little evolution
occurs over time, and consequently, the level of genetic
diversity within the BE segment — as an indicator of
cancer risk — is also invariant over time109. Refinement
of screening schedules and associated prognostic bio-
markers requires longitudinal studies of the evolution
within cancerized fields. Mathematical modelling of the
underlying disease aetiology may further help to opti-
mize the precise timing of clinical screens to maximize
the probability that an individual is screened within a
window of opportunity for timely intervention; here,
evolutionary-minded approaches that explicitly model
clonal expansions show particular promise114–119.
Chemoprevention. Part of the mechanism of action
of chemopreventive therapies is likely to be through
their modulation of pretumour field dynamics.
Though direct evidence of this is generally lacking
— owing to the difficulty inherent in studying pre­
tumour clones — studies in BE provide a rare but ele-
gant example. Therein, the cancer-reducing effects of
non­steroidal anti-inflammatory drugs (NSAIDs) 120
may occur because they can reduce the mutation rate
and change the microenvironment such that certain
clones that are already established in the tissue will no
longer be selected for 121. Similarly, we hypothesize that
­5‑aminosalicylate anti-inflammatory drugs reduce the
risk of CRC in IBD122 through similar mechanisms,
and more generally, the chemopreventive effects of
aspirin123 are likely to function through the modulation
of pre­tumour field evolution. Proton-pump inhibitors
used by patients with BE may also influence the for-
mation of mutant clones, although the relationship, if
any, to ­cancer progression remains unclear 124. Further,
the cancer-­preventive effects of Bacillus Calmette–
Guérin therapy for patients with superficial bladder
cancer 125–127 may occur because it modulates clone–­
microenvironment inter­a ctions 128. Understanding
the mechanisms by which field cancerization driv-
ers are selected for is a promising route towards the
­development of new chemopreventive drugs.

10 | ADVANCE ONLINE PUBLICATION www.nature.com/nrc

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

Metachronous tumours
Primary tumours arising
sequentially in time.
Surgical interventions. The prospect of field canceriza-
tion within a patient undergoing cancer resection has
implications for deciding surgical margins — should an
entire cancerized field be removed along with the overt
lesion? If field cancerization is commonly found in a cer-
tain disease, it is logical to think that patients will be at
risk of metachronous tumours if the cancerized field is not
removed along with the cancer itself 32,114,129–131. In colitis,
there has been a recent tendency for endoscopic (limited)
resection of dysplastic pretumour lesions, which may
underpin a slight increase in cancer incidence rates100,132.
As seen in the treatment of DCIS with breast-conserving
surgery rather than mastectomy 133, choosing optimal
resection margins requires balancing the competing risks
of morbidities associated with more radical surgery with
the increased risk of rapid recurrence from a cancerized
field left in situ. The development of an imaging modality
to visualize the cancerized field during surgery has the
potential to be revolutionary.
Summary and conclusions
The balance of evidence suggests that some tissues in our
bodies inevitably, and unfortunately, become cancerized
as we age: mutations accrue throughout life, and some
will be unlucky mutations that drive the formation and
phenotypic evolution of a field of mutant cells that is
‘one step closer’ to malignancy. Understanding the evo-
lutionary dynamics of cells within these cancerized fields
is key to identifying and then controlling those tissues
that have aged ungracefully towards cancer.

1. Braakhuis, B. J., Tabor, M. P., Kummer, J. A.,
Leemans, C. R. & Brakenhoff, R. H. A genetic
explanation of Slaughter’s concept of field
cancerization: evidence and clinical implications.
Cancer Res. 63, 1727–1730 (2003).
This study presents the updated concept of
field cancerization, which includes a genetic
perspective.
2. Garcia, S. B., Park, H. S., Novelli, M. & Wright, N. A.
Field cancerization, clonality, and epithelial stem cells:
the spread of mutated clones in epithelial sheets.
J. Pathol. 187, 61–81 (1999).
3. Kaz, A. M., Grady, W. M., Stachler, M. D. & Bass, A. J.
Genetic and epigenetic alterations in Barrett’s
esophagus and esophageal adenocarcinoma.
Gastroenterol. Clin. North Am. 44, 473–489
(2015).
4. Virnig, B. A., Tuttle, T. M., Shamliyan, T. & Kane, R. L.
Ductal carcinoma in situ of the breast: a systematic
review of incidence, treatment, and outcomes.
J. Natl Cancer Inst. 102, 170–178 (2010).
5. Bostwick, D. G. & Cheng, L. Precursors of prostate
cancer. Histopathology 60, 4–27 (2012).
6. Prasad, G. A. et al. Long-term survival following
endoscopic and surgical treatment of high-grade
dysplasia in Barrett’s esophagus. Gastroenterology
132, 1226–1233 (2007).
7. Nordenvall, C., Ekbom, A., Bottai, M., Smedby, K. E.
& Nilsson, P. J. Mortality after total colectomy in
3084 patients with inflammatory bowel disease:
a population-based cohort study. Aliment. Pharmacol.
Ther. 40, 280–287 (2014).
8. Laine, L. et al. SCENIC international consensus
statement on surveillance and management of
dysplasia in inflammatory bowel disease.
Gastrointest. Endosc. 81, 489–501.e26 (2015).
9. Shaheen, N. J., Falk, G. W., Iyer, P. G. & Gerson, L. B.
ACG Clinical Guideline: Diagnosis and management
of Barrett’s esophagus. Am. J. Gastroenterol. 111,
30–50 (2016).
10. Blokzijl, F. et al. Tissue-specific mutation accumulation
in human adult stem cells during life. Nature 538,
260–264 (2016).
This study includes a genome-wide measurement
of the accumulation of age-associated mutations
in the intestine and liver.
11. Blanpain, C. & Simons, B. D. Unravelling stem cell
dynamics by lineage tracing. Nat. Rev. Mol. Cell Biol.
14, 489–502 (2013).
12. Klein, A. M., Brash, D. E., Jones, P. H. & Simons, B. D.
Stochastic fate of p53‑mutant epidermal progenitor
cells is tilted toward proliferation by UV B during
preneoplasia. Proc. Natl Acad. Sci. USA 107,
270–275 (2010).
13. Baker, A. M. et al. Quantification of crypt and stem
cell evolution in the normal and neoplastic human
colon. Cell Rep. 8, 940–947 (2014).
14. Kim, K. M., Calabrese, P., Tavare, S. & Shibata, D.
Enhanced stem cell survival in familial adenomatous
polyposis. Am. J. Pathol. 164, 1369–1377(2004).
15. Vortmeyer, A. O. & Alomari, A. K. Pathology of the
nervous system in Von Hippel-Lindau disease.
J. Kidney Cancer VHL 2, 114–129 (2015).
16. Slaughter, D. P., Southwick, H. W. & Smejkal, W.
Field cancerization in oral stratified squamous
epithelium; clinical implications of multicentric origin.
Cancer 6, 963–968 (1953).
This is the original paper proposing the concept of
field cancerization.
17. Franklin, W. A. et al. Widely dispersed p53 mutation
in respiratory epithelium. A novel mechanism for field
carcinogenesis. J. Clin. Invest. 100, 2133–2137
(1997).
This study shows field cancerization in the human
lung by interlobe clonal expansion of a TP53‑mutant
clone.
18. Ushijima, T. & Hattori, N. Molecular pathways:
involvement of Helicobacter pylori-triggered
inflammation in the formation of an epigenetic field
defect, and its usefulness as cancer risk and exposure
markers. Clin. Cancer Res. 18, 923–929 (2012).
This study describes epigenetic fields as a cause
of field cancerization in the stomach.
19. Nosho, K. et al. A prospective cohort study shows
unique epigenetic, genetic, and prognostic features
of synchronous colorectal cancers. Gastroenterology
137, 1609–1620.e3 (2009).
20. Simons, B. D. & Clevers, H. Strategies for homeostatic
stem cell self-renewal in adult tissues. Cell 145,
851–862 (2011).
21. Lochhead, P. et al. Etiologic field effect: reappraisal
of the field effect concept in cancer predisposition
and progression. Mod. Pathol. 28, 14–29 (2015).
22. Hu, B. et al. Multifocal epithelial tumors and field
cancerization from loss of mesenchymal CSL signaling.
Cell 149, 1207–1220 (2012).
This study demonstrates that field cancerization
can be initiated by stromal cell changes rather than
by epithelial cell changes.
23. Calabrese, P., Tavare, S. & Shibata, D. Pretumor
progression: clonal evolution of human stem cell
populations. Am. J. Pathol. 164, 1337–1346 (2004).
24. Greaves, M. Evolutionary determinants of cancer.
Cancer Discov. 5, 806–820 (2015).
25. Stratton, M. R., Campbell, P. J. & Futreal, P. A.
The cancer genome. Nature 458, 719–724 (2009).
26. Martincorena, I. et al. Tumor evolution. High burden
and pervasive positive selection of somatic mutations
in normal human skin. Science 348, 880–886 (2015).
This study identifies an extensive amount of
cancer-associated mutations in morphologically
normal human skin.
27. Alcolea, M. P. et al. Differentiation imbalance in
single oesophageal progenitor cells causes clonal
immortalization and field change. Nat. Cell Biol. 16,
615–622 (2014).
28. Jonason, A. S. et al. Frequent clones of p53‑mutated
keratinocytes in normal human skin. Proc. Natl Acad.
Sci. USA 93, 14025–14029 (1996).
29. Sanchez-Danes, A. et al. Defining the clonal dynamics
leading to mouse skin tumour initiation. Nature 536,
298–303 (2016).
30. Bonilla, X. et al. Genomic analysis identifies new
drivers and progression pathways in skin basal cell
carcinoma. Nat. Genet. 48, 398–406 (2016).
31. Vermeulen, L. et al. Defining stem cell dynamics in
models of intestinal tumor initiation. Science 342,
995–998 (2013).
This study quantifies the selective advantage of
pretumour mutations in the intestine.
32. Galandiuk, S. et al. Field cancerization in the intestinal
epithelium of patients with Crohn’s ileocolitis.
Gastroenterology 142, 855–864.e8 (2012).
This study identifies examples of field cancerization
by mutant clones in human Crohn’s disease.
33. Leedham, S. J. et al. Clonality, founder mutations,
and field cancerization in human ulcerative colitis-
associated neoplasia. Gastroenterology 136,
542–550.e6 (2009).
This study identifies examples of field cancerization
at individual crypt resolution in human colitis.
34. Lawrence, M. S. et al. Discovery and saturation
analysis of cancer genes across 21 tumour types.
Nature 505, 495–501 (2014).
35. Kato, S., Lippman, S. M., Flaherty, K. T. & Kurzrock, R.
The conundrum of genetic “drivers” in benign
conditions. J. Natl Cancer Inst. 108, djw036
(2016).
36. Bieging, K. T., Mello, S. S. & Attardi, L. D. Unravelling
mechanisms of p53‑mediated tumour suppression.
Nat. Rev. Cancer 14, 359–370 (2014).
37. Weaver, J. M. et al. Ordering of mutations in
preinvasive disease stages of esophageal
carcinogenesis. Nat. Genet. 46, 837–843 (2014).
This study shows the general lack of
stage-specificity of genetic mutations in BE.
38. Li, X. et al. Temporal and spatial evolution of somatic
chromosomal alterations: a case-cohort study of
Barrett’s esophagus. Cancer Prev. Res. 7, 114–127
(2014).
39. Stachler, M. D. et al. Paired exome analysis of
Barrett’s esophagus and adenocarcinoma. Nat. Genet.
47, 1047–1055 (2015).
This study illustrates the multiple genetic
pathways, including a punctuated genome-doubling
pathway, that lead from BE to oesophageal cancer.
40. Curvers, W. L. et al. Low-grade dysplasia in Barrett’s
esophagus: overdiagnosed and underestimated.
Am. J. Gastroenterol. 105, 1523–1530 (2010).
41. Kerkhof, M. et al. Grading of dysplasia in Barrett’s
oesophagus: substantial interobserver variation
between general and gastrointestinal pathologists.
Histopathology 50, 920–927 (2007).
42. Tomasetti, C., Li, L. & Vogelstein, B. Stem cell
divisions, somatic mutations, cancer etiology, and
cancer prevention. Science 355, 1330–1334
(2017).
This study suggests a critical role for pretumour
mutation accumulation in cancer development.
43. Alexandrov, L. B. et al. Clock-like mutational processes
in human somatic cells. Nat. Genet. 47, 1402–1407
(2015).
This study identifies mutational processes that
underlie age-associated mutation accumulation.
44. Driessens, G., Beck, B., Caauwe, A., Simons, B. D. &
Blanpain, C. Defining the mode of tumour growth by
clonal analysis. Nature 488, 527–530 (2012).
45. de Bruin, E. C. et al. Spatial and temporal diversity in
genomic instability processes defines lung cancer
evolution. Science 346, 251–256 (2014).
46. Steiling, K., Ryan, J., Brody, J. S. & Spira, A. The field
of tissue injury in the lung and airway. Cancer Prev.
Res. 1, 396–403 (2008).
47. Burd, E. M. Human papillomavirus and cervical
cancer. Clin. Microbiol. Rev. 16, 1–17 (2003).

NATURE REVIEWS | CANCER ADVANCE ONLINE PUBLICATION | 11

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

48. Yakirevich, E. & Resnick, M. B. Pathology of gastric
cancer and its precursor lesions. Gastroenterol. Clin.
North Am. 42, 261–284 (2013).
49. Wang, K. et al. Whole-genome sequencing and
comprehensive molecular profiling identify new driver
mutations in gastric cancer. Nat. Genet. 46, 573–582
(2014).
50. Rubenstein, J. H. & Taylor, J. B. Meta-analysis: the
association of oesophageal adenocarcinoma with
symptoms of gastro-oesophageal reflux.
Aliment. Pharmacol. Ther. 32, 1222–1227 (2010).
51. Dulak, A. M. et al. Exome and whole-genome
sequencing of esophageal adenocarcinoma identifies
recurrent driver events and mutational complexity.
Nat. Genet. 45, 478–486 (2013).
52. Tomasetti, C., Vogelstein, B. & Parmigiani, G. Half or
more of the somatic mutations in cancers of self-
renewing tissues originate prior to tumor initiation.
Proc. Natl Acad. Sci. USA 110, 1999–2004 (2013).
This study suggests that most somatic mutations
in cancer accrue prior to the initiation of cancer
growth.
53. Robles, A. I. et al. Whole-exome sequencing analyses
of inflammatory bowel disease-associated colorectal
cancers. Gastroenterology 150, 931–943 (2016).
54. Issa, J. P., Ahuja, N., Toyota, M., Bronner, M. P. &
Brentnall, T. A. Accelerated age-related CpG island
methylation in ulcerative colitis. Cancer Res. 61,
3573–3577 (2001).
55. Risques, R. A. et al. Ulcerative colitis is a disease of
accelerated colon aging: evidence from telomere
attrition and DNA damage. Gastroenterology 135,
410–418 (2008).
56. Curtius, K. et al. A molecular clock infers heterogeneous
tissue age among patients with Barrett’s esophagus.
PLoS Comput. Biol. 12, e1004919 (2016).
57. Campisi, J. Aging, cellular senescence, and cancer.
Annu. Rev. Physiol. 75, 685–705 (2013).
58. Thirlwell, C. et al. Clonality assessment and clonal
ordering of individual neoplastic crypts shows
polyclonality of colorectal adenomas. Gastroenterology
138, 1441–1454 (2010).
59. Lin, J. et al. Polyclonality of BRAF mutations in acquired
melanocytic nevi. J. Natl Cancer Inst. 101, 1423–1427
(2009).
60. Halberg, R. B. & Dove, W. F. Polyclonal tumors in
the mammalian intestine: are interactions among
multiple initiated clones necessary for tumor initiation,
growth, and progression? Cell Cycle 6, 44–51 (2007).
61. Bjerknes, M. & Cheng, H. Colossal crypts bordering
colon adenomas in ApcMin mice express full-length Apc.
Am. J. Pathol. 154, 1831–1834 (1999).
62. Thliveris, A. T. et al. Polyclonality of familial murine
adenomas: analyses of mouse chimeras with low tumor
multiplicity suggest short-range interactions. Proc. Natl
Acad. Sci. USA 102, 6960–6965 (2005).
63. Fernandez-Sanchez, M. E. et al. Mechanical induction
of the tumorigenic beta-catenin pathway by tumour
growth pressure. Nature 523, 92–95 (2015).
64. Martens, E. A., Kostadinov, R., Maley, C. C. &
Hallatschek, O. Spatial structure increases the waiting
time for cancer. New J. Phys. 13, 115014 (2011).
65. Greaves, L. C. et al. Mitochondrial DNA mutations are
established in human colonic stem cells, and mutated
clones expand by crypt fission. Proc. Natl Acad. Sci.
USA 103, 714–719 (2006).
66. McDonald, S. A. et al. Mechanisms of field
cancerization in the human stomach: the expansion and
spread of mutated gastric stem cells. Gastroenterology
134, 500–510 (2008).
This study provides a genetic demonstration of field
cancerization in the human stomach.
67. Nicholson, A. M. et al. Barrett’s metaplasia glands are
clonal, contain multiple stem cells and share a common
squamous progenitor. Gut 61, 1380–1389 (2012).
68. Clayton, E. et al. A single type of progenitor cell
maintains normal epidermis. Nature 446, 185–189
(2007).
69. Doupe, D. P. et al. A single progenitor population
switches behavior to maintain and repair esophageal
epithelium. Science 337, 1091–1093 (2012).
70. Teixeira, V. H. et al. Stochastic homeostasis in
human airway epithelium is achieved by neutral
competition of basal cell progenitors. eLife 2, e00966
(2013).
71. Nystul, T. & Spradling, A. An epithelial niche in the
Drosophila ovary undergoes long-range stem cell
replacement. Cell Stem Cell 1, 277–285 (2007).
72. Yui, S. et al. Functional engraftment of colon epithelium
expanded in vitro from a single adult Lgr5+ stem cell.
Nat. Med. 18, 618–623 (2012).
73. Frieda, K. L. et al. Synthetic recording and in situ
readout of lineage information in single cells. Nature
541, 107–111 (2017).
74. McKenna, A. et al. Whole-organism lineage tracing by
combinatorial and cumulative genome editing. Science
353, aaf7907 (2016).
75. Snippert, H. J., Schepers, A. G., van Es, J. H.,
Simons, B. D. & Clevers, H. Biased competition
between Lgr5 intestinal stem cells driven by oncogenic
mutation induces clonal expansion. EMBO Rep. 15,
62–69 (2014).
76. Choi, C. R., Bakir, I. A., Hart, A. L. & Graham, T. A.
Clonal evolution of colorectal cancer in IBD. Nat. Rev.
Gastroenterol. Hepatol. 14, 218–229 (2017).
77. Reid, B. J., Li, X., Galipeau, P. C. & Vaughan, T. L.
Barrett’s oesophagus and oesophageal
adenocarcinoma: time for a new synthesis. Nat. Rev.
Cancer 10, 87–101 (2010).
78. Hazelton, W. D. et al. The role of gastroesophageal
reflux and other factors during progression to
esophageal adenocarcinoma. Cancer Epidemiol.
Biomarkers Prev. 24, 1012–1023 (2015).
79. Korolev, K. S., Xavier, J. B. & Gore, J. Turning ecology
and evolution against cancer. Nat. Rev. Cancer 14,
371–380 (2014).
80. Martincorena, I., Jones, P. H. & Campbell, P. J.
Constrained positive selection on cancer mutations
in normal skin. Proc. Natl Acad. Sci. USA 113,
E1128–E1129 (2016).
81. Mantovani, A., Allavena, P., Sica, A. & Balkwill, F.
Cancer-related inflammation. Nature 454, 436–444
(2008).
82. Davis, H. et al. Aberrant epithelial GREM1
expression initiates colonic tumorigenesis from cells
outside the stem cell niche. Nat. Med. 21, 62–70
(2015).
83. Rutter, M. et al. Severity of inflammation is a risk factor
for colorectal neoplasia in ulcerative colitis.
Gastroenterology 126, 451–459 (2004).
84. Saadi, A. et al. Stromal genes discriminate preinvasive
from invasive disease, predict outcome, and highlight
inflammatory pathways in digestive cancers.
Proc. Natl Acad. Sci. USA 107, 2177–2182 (2010).
This study shows consistent changes in the gene
expression profile of cells within the stromal
compartment across pre-malignant gastrointestinal
diseases.
85. Bronisz, A. et al. Reprogramming of the tumour
microenvironment by stromal PTEN-regulated
miR‑320. Nat. Cell Biol. 14, 159–167 (2011).
86. Quail, D. F. & Joyce, J. A. Microenvironmental
regulation of tumor progression and metastasis.
Nat. Med. 19, 1423–1437 (2013).
87. Tomlinson, I. P. et al. Multiple common susceptibility
variants near BMP pathway loci GREM1, BMP4,
and BMP2 explain part of the missing heritability
of colorectal cancer. PLoS Genet. 7, e1002105 (2011).
88. Elinav, E. et al. Inflammation-induced cancer: crosstalk
between tumours, immune cells and microorganisms.
Nat. Rev. Cancer 13, 759–771 (2013).
89. Gatenby, R. A. & Gillies, R. J. A microenvironmental
model of carcinogenesis. Nat. Rev. Cancer 8, 56–61
(2008).
This perspective proposes critical
microenvironmental barriers that must be overcome
during tumorigenesis.
90. Perez-Mancera, P. A., Young, A. R. & Narita, M.
Inside and out: the activities of senescence in cancer.
Nat. Rev. Cancer 14, 547–558 (2014).
91. Welch, H. G. & Black, W. C. Overdiagnosis in cancer.
J. Natl Cancer Inst. 102, 605–613 (2010).
92. Mori, H. et al. Chromosome translocations and covert
leukemic clones are generated during normal fetal
development. Proc. Natl Acad. Sci. USA 99,
8242–8247 (2002).
93. Greaves, M. F., Maia, A. T., Wiemels, J. L. & Ford, A. M.
Leukemia in twins: lessons in natural history. Blood
102, 2321–2333 (2003).
94. Genovese, G. et al. Clonal hematopoiesis and blood-
cancer risk inferred from blood DNA sequence. N. Engl.
J. Med. 371, 2477–2487 (2014).
This study shows how the mutation burden in
ostensibly normal white blood cells confers
leukaemia risk.
95. Hvid-Jensen, F., Pedersen, L., Drewes, A. M.,
Sorensen, H. T. & Funch-Jensen, P. Incidence of
adenocarcinoma among patients with Barrett’s
esophagus. N. Engl. J. Med. 365, 1375–1383 (2011).
96. Spechler, S. J. Barrett esophagus and risk of
esophageal cancer: a clinical review. JAMA 310,
627–636 (2013).
97. Greaves, M. Does everyone develop covert cancer?
Nat. Rev. Cancer 14, 209–210 (2014).
This commentary elegantly discusses the inevitably
of cancerization.
98. Runge, T. M., Abrams, J. A. & Shaheen, N. J.
Epidemiology of Barrett’s esophagus and esophageal
adenocarcinoma. Gastroenterol. Clin. North Am. 44,
203–231 (2015).
99. Odze, R. D. Diagnosis and grading of dysplasia in
Barrett’s oesophagus. J. Clin. Pathol. 59, 1029–1038
(2006).
100. Choi, C. H. et al. Forty-year analysis of colonoscopic
surveillance for neoplasia in ulcerative colitis: an
updated overview. Am. J. Gastroenterol. 110,
1022–1034 (2015).
101. Bird-Lieberman, E. L. et al. Population-based study
reveals new risk-stratification biomarker panel for
Barrett’s esophagus. Gastroenterology 143,
927–935.e3 (2012).
102. Reid, B. J. et al. Predictors of progression in Barrett’s
esophagus II: baseline 17p (p53) loss of heterozygosity
identifies a patient subset at increased risk for
neoplastic progression. Am. J. Gastroenterol. 96,
2839–2848 (2001).
103. Timmer, M. R. et al. Derivation of genetic biomarkers
for cancer risk stratification in Barrett’s oesophagus:
a prospective cohort study. Gut 65, 1602–1610
(2016).
104. Silvestri, G. A. et al. A bronchial genomic classifier
for the diagnostic evaluation of lung cancer.
N. Engl. J. Med. 373, 243–251 (2015).
105. Spira, A. et al. Airway epithelial gene expression in the
diagnostic evaluation of smokers with suspect lung
cancer. Nat. Med. 13, 361–366 (2007).
106. Dhawan, A., Graham, T. A. & Fletcher, A. G. A.
Computational modeling approach for deriving
biomarkers to predict cancer risk in premalignant
disease. Cancer Prev. Res. 9, 283–295 (2016).
107. Greaves, M. & Maley, C. C. Clonal evolution in cancer.
Nature 481, 306–313 (2012).
108. Maley, C. C. et al. Genetic clonal diversity predicts
progression to esophageal adenocarcinoma.
Nat. Genet. 38, 468–473 (2006).
This study shows how measurement of the
cancerized field — in this case, the evolvability as
measured by genetic diversity — predicts future
cancer risk.
109. Martinez, P. et al. Dynamic clonal equilibrium and
predetermined cancer risk in Barrett’s oesophagus.
Nat. Commun. 7, 12158 (2016).
110. Andor, N. et al. Pan-cancer analysis of the extent and
consequences of intratumor heterogeneity. Nat. Med.
22, 105–113 (2016).
111. Maley, C. C. et al. The combination of genetic
instability and clonal expansion predicts progression
to esophageal adenocarcinoma. Cancer Res. 64,
7629–7633 (2004).
112. Willenbucher, R. F. et al. Genomic instability is an early
event during the progression pathway of ulcerative-
colitis-related neoplasia. Am. J. Pathol. 154,
1825–1830 (1999).
113. Salk, J. J. et al. Clonal expansions in ulcerative colitis
identify patients with neoplasia. Proc. Natl Acad. Sci.
USA 106, 20871–20876 (2009).
This study demonstrates that evidence of recent
clonal expansion in patients with colitis indicates
increased cancer risk.
114. Curtius, K., Hazelton, W. D., Jeon, J. &
Luebeck, E. G. A. Multiscale model evaluates screening
for neoplasia in Barrett’s esophagus. PLoS Comput.
Biol. 11, e1004272 (2015).
This study demonstrates how computational
modelling of field cancerization can help optimize
screening protocols.
115. de Koning, H. J. et al. Benefits and harms of computed
tomography lung cancer screening strategies:
a comparative modeling study for the U.S. Preventive
Services Task Force. Ann. Intern. Med. 160, 311–320
(2014).
116. Hanin, L. & Pavlova, L. Optimal screening schedules
for prevention of metastatic cancer. Stat. Med. 32,
206–219 (2013).
117. Jeon, J., Meza, R., Moolgavkar, S. H. & Luebeck, E. G.
Evaluation of screening strategies for pre-malignant
lesions using a biomathematical approach.
Math. Biosci. 213, 56–70 (2008).
118. Kong, C. Y. et al. Exploring the recent trend in
esophageal adenocarcinoma incidence and
mortality using comparative simulation modeling.
Cancer Epidemiol. Biomarkers Prev. 23, 997–1006
(2014).

12 | ADVANCE ONLINE PUBLICATION www.nature.com/nrc

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

119. Ryser, M. D., Lee, W. T., Ready, N. E., Leder, K. Z. &
Foo, J. Quantifying the dynamics of field cancerization
in tobacco-related head and neck cancer: a multiscale
modeling approach. Cancer Res. 76, 7078–7088
(2016).
120. Galipeau, P. C. et al. NSAIDs modulate CDKN2A, TP53,
and DNA content risk for progression to esophageal
adenocarcinoma. PLoS Med. 4, e67 (2007).
121. Kostadinov, R. L. et al. NSAIDs modulate clonal
evolution in Barrett’s esophagus. PLoS Genet. 9,
e1003553 (2013).
This study suggests that NSAIDs reduce cancer risk
by inhibiting clonal evolution within a cancerized
field in BE.
122. Zhao, L. N. et al. 5‑Aminosalicylates reduce the risk of
colorectal neoplasia in patients with ulcerative colitis:
an updated meta-analysis. PLoS ONE 9, e94208
(2014).
123. Rothwell, P. M. et al. Effect of daily aspirin on long-term
risk of death due to cancer: analysis of individual
patient data from randomised trials. Lancet 377,
31–41 (2011).
124. Hvid-Jensen, F., Pedersen, L., Funch-Jensen, P. &
Drewes, A. M. Proton pump inhibitor use may not
prevent high-grade dysplasia and oesophageal
adenocarcinoma in Barrett’s oesophagus: a nationwide
study of 9883 patients. Aliment. Pharmacol. Ther. 39,
984–991 (2014).
125. Gee, J., Sabichi, A. L. & Grossman, H. B.
Chemoprevention of superficial bladder cancer.
Crit. Rev. Oncol. Hematol. 43, 277–286 (2002).
126. Herr, H. W. et al. Intravesical bacillus Calmette-Guerin
therapy prevents tumor progression and death from
superficial bladder cancer: ten-year follow‑up of a
prospective randomized trial. J. Clin. Oncol. 13,
1404–1408 (1995).
127. Lamm, D. L. et al. Maintenance Bacillus Calmette-
Guerin immunotherapy for recurrent TA, T1 and
carcinoma in situ transitional cell carcinoma of the
bladder: a randomized Southwest Oncology Group
Study. J. Urol. 163, 1124–1129 (2000).
128. Redelman-Sidi, G., Glickman, M. S. & Bochner, B. H.
The mechanism of action of BCG therapy for bladder
cancer — a current perspective. Nat. Rev. Urol. 11,
153–162 (2014).
129. Foo, J., Leder, K. & Ryser, M. D. Multifocality and
recurrence risk: a quantitative model of field
cancerization. J. Theor. Biol. 355, 170–184
(2014).
130. Gupta, M. et al. Recurrence of esophageal intestinal
metaplasia after endoscopic mucosal resection and
radiofrequency ablation of Barrett’s esophagus: results
from a US Multicenter Consortium. Gastroenterology
145, 79–86.e1 (2013).
131. Rutgeerts, P. et al. Predictability of the postoperative
course of Crohn’s disease. Gastroenterology 99,
956–963 (1990).
132. Choi, C. H. et al. Low-grade dysplasia in ulcerative
colitis: risk factors for developing high-grade dysplasia
or colorectal cancer. Am. J. Gastroenterol. 110,
1461–1471 (2015).
133. Morrow, M. et al. Society of Surgical Oncology–
American Society for Radiation Oncology–American
Society of Clinical Oncology consensus guideline on
margins for breast-conserving surgery with whole-
breast irradiation in ductal carcinoma in situ. Ann. Surg.
Oncol. 23, 3801–3810 (2016).
134. Okosun, J. et al. Integrated genomic analysis identifies
recurrent mutations and evolution patterns driving the
initiation and progression of follicular lymphoma.
Nat. Genet. 46, 176–181 (2014).
135. Esserman, L. J. et al. Addressing overdiagnosis and
overtreatment in cancer: a prescription for change.
Lancet Oncol. 15, e234–e242 (2014).
This article highlights the current scale of the
overdiagnosis and overtreatment problem in
pre-malignant disease.
136. Campbell, J. D. et al. The case for a Pre-Cancer Genome
Atlas (PCGA). Cancer Prev. Res. 9, 119–124 (2016).
137. Koizumi, K. et al. Array-based identification of common
DNA methylation alterations in ulcerative colitis.
Int. J. Oncol. 40, 983–994 (2012).
138. van Dekken, H. et al. Genomic analysis of a case of
multifocal adenocarcinoma in ulcerative colitis.
Virchows Arch. 449, 716–721 (2006).
139. Alonso, S. et al. Methylation of MGMT and ADAMTS14
in normal colon mucosa: biomarkers of a field defect for
cancerization preferentially targeting elder African-
Americans. Oncotarget 6, 3420–3431 (2015).
140. Asada, K. et al. FHL1 on chromosome X is a single-hit
gastrointestinal tumor-suppressor gene and contributes
to the formation of an epigenetic field defect. Oncogene
32, 2140–2149 (2013).
141. Damania, D. et al. Nanocytology of rectal colonocytes
to assess risk of colon cancer based on field
cancerization. Cancer Res. 72, 2720–2727 (2012).
142. Hawthorn, L., Lan, L. & Mojica, W. Evidence for field
effect cancerization in colorectal cancer. Genomics 103,
211–221 (2014).
143. Kamiyama, H. et al. DNA demethylation in normal
colon tissue predicts predisposition to multiple cancers.
Oncogene 31, 5029–5037 (2012).
144. Kaz, A. M. et al. Patterns of DNA methylation in the
normal colon vary by anatomical location, gender, and
age. Epigenetics 9, 492–502 (2014).
145. Milicic, A. et al. Ectopic expression of P‑cadherin
correlates with promoter hypomethylation early in
colorectal carcinogenesis and enhanced intestinal crypt
fission in vivo. Cancer Res. 68, 7760–7768 (2008).
146. Shen, L. et al. MGMT promoter methylation and field
defect in sporadic colorectal cancer. J. Natl Cancer Inst.
97, 1330–1338 (2005).
147. Gutierrez-Gonzalez, L. et al. The clonal origins of
dysplasia from intestinal metaplasia in the human
stomach. Gastroenterology 140, 1251–1260.e6
(2011).
148. Kang, G. H. et al. Genetic evidence for the multicentric
origin of synchronous multiple gastric carcinoma.
Lab. Invest. 76, 407–417 (1997).
149. Takeshima, H. et al. Frequent involvement of chromatin
remodeler alterations in gastric field cancerization.
Cancer Lett. 357, 328–338 (2015).
150. Yamanoi, K. et al. Epigenetic clustering of gastric
carcinomas based on DNA methylation profiles at the
precancerous stage: its correlation with tumor
aggressiveness and patient outcome. Carcinogenesis
36, 509–520 (2015).
151. Zaky, A. H. et al. Clinicopathologic implications of
genetic instability in intestinal-type gastric cancer and
intestinal metaplasia as a precancerous lesion: proof of
field cancerization in the stomach. Am. J. Clin. Pathol.
129, 613–621 (2008).
152. Cense, H. A., van Lanschot, J. J., Fockens, P.,
Obertop, H. & Offerhaus, G. J. A patient with seven
primary tumors of the upper aerodigestive tract: the
process of field cancerization versus distant monoclonal
expansion. Dis. Esophagus 10, 139–142 (1997).
153. Kammori, M. et al. Squamous cell carcinomas of the
esophagus arise from a telomere-shortened epithelial
field. Int. J. Mol. Med. 20, 793–799 (2007).
154. Lee, Y. C. et al. Revisit of field cancerization in
squamous cell carcinoma of upper aerodigestive tract:
better risk assessment with epigenetic markers.
Cancer Prev. Res. 4, 1982–1992 (2011).
155. Matsuda, Y. et al. Hypomethylation of Alu repetitive
elements in esophageal mucosa, and its potential
contribution to the epigenetic field for cancerization.
Cancer Causes Control 23, 865–873 (2012).
156. Oka, D. et al. The presence of aberrant DNA
methylation in noncancerous esophageal mucosae in
association with smoking history: a target for risk
diagnosis and prevention of esophageal cancers.
Cancer 115, 3412–3426 (2009).
157. Roesch-Ely, M. et al. Proteomic analysis of field
cancerization in pharynx and oesophagus: a
prospective pilot study. J. Pathol. 221, 462–470
(2010).
158. Yakoub, D., Keun, H. C., Goldin, R. & Hanna, G. B.
Metabolic profiling detects field effects in nondysplastic
tissue from esophageal cancer patients. Cancer Res.
70, 9129–9136 (2010).
159. Galipeau, P. C., Prevo, L. J., Sanchez, C. A.,
Longton, G. M. & Reid, B. J. Clonal expansion and
loss of heterozygosity at chromosomes 9p and 17p
in premalignant esophageal (Barrett’s) tissue.
J. Natl Cancer Inst. 91, 2087–2095 (1999).
160. Maley, C. C. et al. Selectively advantageous mutations
and hitchhikers in neoplasms: p16 lesions are selected
in Barrett’s esophagus. Cancer Res. 64, 3414–3427
(2004).
161. van Dekken, H., Vissers, C. J., Tilanus, H. W.,
Tanke, H. J. & Rosenberg, C. Clonal analysis of a case of
multifocal oesophageal (Barrett’s) adenocarcinoma by
comparative genomic hybridization. J. Pathol. 188,
263–266 (1999).
162. Chang, Y. L. et al. Clonality and prognostic implications
of p53 and epidermal growth factor receptor somatic
aberrations in multiple primary lung cancers.
Clin. Cancer Res. 13, 52–58 (2007).
163. Kadara, H. & Wistuba, I. I. Field cancerization in
non-small cell lung cancer: implications in disease
pathogenesis. Proc. Am. Thorac Soc. 9, 38–42 (2012).
164. McCaughan, F. et al. Genomic evidence of pre-invasive
clonal expansion, dispersal and progression in
bronchial dysplasia. J. Pathol. 224, 153–159 (2011).
165. Pipinikas, C. P. et al. Cell migration leads to spatially
distinct but clonally related airway cancer precursors.
Thorax 69, 548–557 (2014).
166. Sozzi, G. et al. Genetic evidence for an independent
origin of multiple preneoplastic and neoplastic lung
lesions. Cancer Res. 55, 135–140 (1995).
167. Gomperts, B. N., Walser, T. C., Spira, A. &
Dubinett, S. M. Enriching the molecular definition of
the airway “field of cancerization:” establishing new
paradigms for the patient at risk for lung cancer.
Cancer Prev. Res. 6, 4–7 (2013).
168. Kadara, H. et al. Transcriptomic architecture of the
adjacent airway field cancerization in non-small cell lung
cancer. J. Natl Cancer Inst. 106, dju004 (2014).
169. Lin, M. W., Wu, C. T., Kuo, S. W., Chang, Y. L. &
Yang, P. C. Clinicopathology and genetic profile of
synchronous multiple small adenocarcinomas:
implication for surgical treatment of an uncommon
lung malignancy. Ann. Surg. Oncol. 21, 2555–2562
(2014).
170. Nakachi, I. et al. Application of SNP microarrays to the
genome-wide analysis of chromosomal instability in
premalignant airway lesions. Cancer Prev. Res. 7,
255–265 (2014).
171. Weichert, W. & Warth, A. Early lung cancer with lepidic
pattern: adenocarcinoma in situ, minimally invasive
adenocarcinoma, and lepidic predominant
adenocarcinoma. Curr. Opin. Pulm. Med. 20,
309–316 (2014).
172. Wirtschafter, E., Walts, A. E., Liu, S. T. &
Marchevsky, A. M. Diffuse idiopathic pulmonary
neuroendocrine cell hyperplasia of the lung
(DIPNECH): current best evidence. Lung 193,
659–667 (2015).
173. Angadi, P. V., Savitha, J. K., Rao, S. S. & Sivaranjini, Y.
Oral field cancerization: current evidence and future
perspectives. Oral Maxillofac. Surg. 16, 171–180
(2012).
174. Boscolo-Rizzo, P. et al. Telomere shortening in mucosa
surrounding the tumor: biosensor of field cancerization
and prognostic marker of mucosal failure in head and
neck squamous cell carcinoma. Oral Oncol. 51,
500–507 (2015).
175. Califano, J. et al. Genetic progression model for head
and neck cancer: implications for field cancerization.
Cancer Res. 56, 2488–2492 (1996).
176. Narayana, A., Vaughan, A. T., Fisher, S. G. &
Reddy, S. P. Second primary tumors in laryngeal cancer:
results of long-term follow‑up. Int. J. Radiat. Oncol.
Biol. Phys. 42, 557–562 (1998).
177. Pentenero, M. et al. Field effect in oral precancer as
assessed by DNA flow cytometry and array-CGH.
J. Oral Pathol. Med. 41, 119–123 (2012).
178. Shaw, R. J. et al. Molecular staging of surgical margins
in oral squamous cell carcinoma using promoter
methylation of p16INK4A, cytoglobin, E‑cadherin, and
TMEFF2. Ann. Surg. Oncol. 20, 2796–2802 (2013).
179. Van der Vorst, S., Dekairelle, A. F., Weynand, B.,
Hamoir, M. & Gala, J. L. Assessment of p53 functional
activity in tumor cells and histologically normal mucosa
from patients with head and neck squamous cell
carcinoma. Head Neck 34, 1542–1550 (2012).
180. Dworkin, A. M., Huang, T. H. & Toland, A. E. Epigenetic
alterations in the breast: Implications for breast cancer
detection, prognosis and treatment. Semin. Cancer
Biol. 19, 165–171 (2009).
181. Ellsworth, D. L. et al. Outer breast quadrants
demonstrate increased levels of genomic instability.
Ann. Surg. Oncol. 11, 861–868 (2004).
182. Foschini, M. P. et al. Genetic clonal mapping of in situ
and invasive ductal carcinoma indicates the field
cancerization phenomenon in the breast. Hum. Pathol.
44, 1310–1319 (2013).
183. Rivenbark, A. G. & Coleman, W. B. Field cancerization
in mammary carcinogenesis - Implications for
prevention and treatment of breast cancer.
Exp. Mol. Pathol. 93, 391–398 (2012).
184. Trujillo, K. A. et al. Breast field cancerization: isolation
and comparison of telomerase-expressing cells in tumor
and tumor adjacent, histologically normal breast tissue.
Mol. Cancer Res. 9, 1209–1221 (2011).
185. Haaland, C. M. et al. Differential gene expression in
tumor adjacent histologically normal prostatic tissue
indicates field cancerization. Int. J. Oncol. 35,
537–546 (2009).
186. Nonn, L., Ananthanarayanan, V. & Gann, P. H. Evidence
for field cancerization of the prostate. Prostate 69,
1470–1479 (2009).

NATURE REVIEWS | CANCER ADVANCE ONLINE PUBLICATION | 13

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

187. Trujillo, K. A., Jones, A. C., Griffith, J. K. & Bisoffi, M.
Markers of field cancerization: proposed clinical
applications in prostate biopsies. Prostate Cancer
2012, 302894 (2012).
188. Hafner, C., Knuechel, R., Stoehr, R. & Hartmann, A.
Clonality of multifocal urothelial carcinomas: 10 years
of molecular genetic studies. Int. J. Cancer 101, 1–6
(2002).
189. Vriesema, J. L., Aben, K. K., Witjes, J. A.,
Kiemeney, L. A. & Schalken, J. A. Superficial and
metachronous invasive bladder carcinomas are
clonally related. Int. J. Cancer 93, 699–702 (2001).
190. Wang, Y., Lang, M. R., Pin, C. L. & Izawa, J. I.
Comparison of the clonality of urothelial carcinoma
developing in the upper urinary tract and those
developing in the bladder. SpringerPlus 2, 412
(2013).
191. Kanjilal, S. et al. p53 mutations in nonmelanoma skin
cancer of the head and neck: molecular evidence for
field cancerization. Cancer Res. 55, 3604–3609
(1995).
192. Stern, R. S., Bolshakov, S., Nataraj, A. J. &
Ananthaswamy, H. N. p53 mutation in nonmelanoma
skin cancers occurring in psoralen ultraviolet a‑treated
patients: evidence for heterogeneity and field
cancerization. J. Invest. Dermatol. 119, 522–526
(2002).
193. Szeimies, R. M. et al. Clinical, histopathological and
immunohistochemical assessment of human skin field
cancerization before and after photodynamic therapy.
Br. J. Dermatol. 167, 150–159 (2012).
194. Vatve, M., Ortonne, J. P., Birch-Machin, M. A. &
Gupta, G. Management of field change in actinic
keratosis. Br. J. Dermatol. 157 (Suppl. 2), 21–24
(2007).
195. Shain, A. H. et al. The genetic evolution of melanoma
from precursor lesions. N. Engl. J. Med. 373,
1926–1936 (2015).
Acknowledgements
The authors thank D. Brash for helpful conversations during
the writing of this manuscript. The authors acknowledge fund-
ing from Cancer Research UK (grants A19771 to T.A.G. and
A21870 to N.A.W.), the Wellcome Trust (202778/Z/16/Z to
T.A.G.) and the Barts Charity (472–2300 to K.C. and T.A.G.).
Author contributions
All authors researched data for the article, made substan­
tial contributions to discussion of the content, wrote the
article and reviewed and edited the manuscript before
submission.
Competing interests statement
The authors declare no competing interests.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

14 | ADVANCE ONLINE PUBLICATION www.nature.com/nrc

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.