10.1038@s41575 019 0189 8

REvIEWS
Global burden of colorectal cancer:

emerging trends, risk factors and
prevention strategies
NaNa Keum1,2 and Edward Giovannucci1,3,4,5*
Abstract | Globally, colorectal cancer (CRC) is the third most commonly diagnosed malignancy
and the second leading cause of cancer death. Arising through three major pathways, including
adenoma–carcinoma sequence, serrated pathway and inflammatory pathway, CRC represents
an aetiologically heterogeneous disease according to subtyping by tumour anatomical location
or global molecular alterations. Genetic factors such as germline MLH1 and APC mutations have
an aetiologic role, predisposing individuals to CRC. Yet, the majority of CRC is sporadic and
largely attributable to the constellation of modifiable environmental risk factors characterizing
westernization (for example, obesity, physical inactivity, poor diets, alcohol drinking and smoking).
As such, the burden of CRC is shifting towards low-income and middle-income countries as
they become westernized. Furthermore, the rising incidence of CRC at younger ages (before
age 50 years) is an emerging trend. This Review provides a comprehensive summary of CRC
epidemiology, with emphasis on modifiable lifestyle and nutritional factors, chemoprevention
and screening. Overall, the optimal reduction of CRC incidence and mortality will require
concerted efforts to reduce modifiable risk factors, to leverage chemoprevention research
and to promote population-wide and targeted screening.
The majority of colorectal cancers (CRCs) (>90%) are Of note, CRCs with any heritable components are
adenocarcinoma, a malignant neoplasm that develops not fully hereditary, with environmental factors also
from glandular epithelial cells of the colon and rectum; contributing to carcinogenesis.
other rare types include squamous cell carcinoma, There has been a dramatic increase in our understand
1
Department of Nutrition, adenosquamous carcinoma, spindle cell carcinoma and ing of the epidemiology, aetiology, molecular biology
Harvard T.H. Chan School
of Public Health, Boston,
undifferentiated carcinoma1. Approximately 60–65% and clinical aspects of CRC over the past several decades.
MA, USA. of CRC cases arise sporadically (that is, occur in indi Nevertheless, 1.8 million new cases are diagnosed
2
Department of Food Science
viduals without a family history of CRC or inher annually worldwide6. Because CRC is often diagnosed
and Biotechnology, Dongguk ited genetic mutations that increase CRC risk) (Fig. 1) at advanced clinical stages, about 900,000 individuals
University, Goyang, South through acquired somatic genetic and epigenetic aber die from this malignancy6. Furthermore, although CRC
Korea. rations largely attributable to potentially modifiable is thought of as a disease primarily of developed coun
3
Department of Epidemiology, risk factors2. CRC has a hereditary component, with tries, rapid rises in incidence are occurring in countries
Harvard T.H. Chan School of
twin studies having estimated the heritability of CRC undergoing economic development and changes in
Public Health, Boston, MA,
USA.
to be 35–40%3,4. Approximately 25% of CRC cases have diet and lifestyle6. In affluent countries, screening and
4
Channing Division of
a family history of CRC without any obvious genetic improved treatment is reducing incidence and mortality
Network Medicine, Brigham cancer syndrome2. Only 5% are attributed to heredi considerably7, although the rates remain relatively high,
and Women’s Hospital and tary cancer syndromes such as hereditary nonpolyp with age-standardized rates per 100,000 persons per year
Harvard Medical School, osis colorectal cancer (HNPCC, also known as Lynch estimated to be 30.8 for incidence and 10.6 for mortality
Boston, MA, USA.
syndrome) or familial adenomatous polyposis (FAP), (worldwide rates are 19.7 and 8.9, respectively)6. In less
5
Department of Medicine, caused by inherited germline mutations in rare but high- affluent countries, existing screening programmes and
Harvard Medical School,
Boston, MA, USA.
penetrance susceptibility genes (for example, MLH1 and available medical treatment are currently insufficient8 to
APC, respectively)2. Identified from genome-wide asso curb the escalating rise in these rates6. Thus, it is criti
*e-mail: egiovann@
hsph.harvard.edu ciation studies (GWAS), common but low-penetrance cally important for governments, health organizations
https://doi.org/10.1038/ genetic variations known to be associated with CRC risk, and individuals to understand the drivers of CRC inci
s41575-019-0189-8 in aggregate, explain less than 1% of CRC heritability5. dence and primary and secondary prevention strategies
Nature Reviews | Gastroenterology & Hepatology

Reviews
Key points the USA, France and New Zealand), relatively stable
(for example, the UK and Australia) or increasing (for
• Certain global genetic and epigenetic aberrations are disproportionally distributed example, Italy, Norway and Spain)12 (Fig. 3a,b). These
across the colorectum, which corresponds to aetiological heterogeneity of colorectal remarkable fluctuations in incidence rates, which are
cancer (CRC), especially hypermutated cancers, by anatomical location more evident in men than in women, are probably attrib
• With increasing incidence of CRC at younger ages, there is an urgent need to better utable to changing distributions of key risk factors13,14
identify high-risk individuals younger than 50 years, the age when screening typically and CRC screening uptake (discussed later)8. Notably,
starts
countries have undergone a varying degree of changes
• The constellation of factors associated with westernization, such as obesity, physical in the prevalence of CRC risk factors13,14 and of CRC
inactivity, poor diets, alcohol drinking and smoking, is likely to drive increasing CRC
screening practice8, regardless of geographical proximity.
incidence in economically transitioning countries
This aspect might help explain differential increases of
• Evidence indicates that aspirin probably confers chemopreventive benefit against
CRC incidence rates in neighbouring countries.
CRC, though recommendation for its widespread prophylactic use is currently
premature
Increasing CRC incidence rates do not necessarily
translate to escalating mortality rates (Fig. 2c). Despite
• Screening colonoscopy and faecal occult blood test, when implemented
appropriately per national financial and medical resources and CRC incidence, could
rising incidence rates, CRC mortality rates are decreas
contribute to secondary prevention of CRC ing in some countries (for example, Italy and Slovenia)
• The optimal reduction of CRC incidence and mortality will require concerted efforts
probably due to increased screening leading to early
to reduce modifiable risk factors, to leverage chemoprevention research and to detection and improved treatment12. Conversely, in some
promote population-wide and targeted screening countries with rising incidence rates but limited medi
cal resources (for example, Brazil and Russia), mortality
rates are increasing12.
to counter the rise. In this Review, we describe the global
trends of CRC, explain the pathogenesis and molecu Age, sex and ethnicity trends. As cancer is a disease
lar subtypes of CRC, summarize current evidence on of ageing, the rates of CRC development and death
genetic and modifiable risk factors of CRC, and high increase rapidly after age 50 years, with an estimated
light the potential of chemoprevention and importance 90% of global cases and deaths occurring after this
of screening in preventing CRC. age6 (Fig. 4). The age-adjusted rates are higher in men
than in women, yielding an approximately 1.4-fold and
Descriptive epidemiology 1.5-fold difference for incidence (23.6 versus 16.3 cases
Temporal and regional trends. According to estimates per 100,000 persons per year) and death (10.8 versus
from the International Agency for Research on Cancer 7.2 deaths per 100,000 persons per year), respectively6;
in 2018, globally, CRC constituted approximately the difference becomes evident after age 50 years (Fig. 4).
1.8 million new cases and 900,000 deaths annually, The higher CRC rates in men might relate to a com
making it the third most commonly diagnosed malig bination of multiple factors. Compared to women, men
nancy and the second leading cause of cancer deaths6.
Generally, industrialization and economic growth lead
to a western dietary pattern, sedentary lifestyle and
increasing obesity, all of which are major risk factors for Family
CRC9. Thus, the westernization of countries tends to be history
followed by a rise in CRC incidence rates10. This pattern (25%) Hereditary cancer
syndromes (5%)
is supported by higher CRC incidence rates observed Sporadic e.g. HNPCC (2–4%),
in economically developed countries as indicated by a (60–65%) FAP (<1%)
higher human development index, a composite score of Known CRC
life expectancy, education and income11 (Fig. 2a). low-penetrance
Age-s tandardized incidence rates vary more genetic variations (<1%)
than 45-fold between the highest (51.2 cases per Unknown inherited
100,000 persons per year in Hungary) to lowest (1.1 cases genomic aberrations (?%)
per 100,000 persons per year in the Gambia)6 (Fig. 2b,c). Fig. 1 | proportion of colorectal cancer cases associated
The large international variation represents a combined with sporadic and hereditary factors. The majority
effect of multiple factors, including lifestyle, genetics, of colorectal cancers (CRCs) arise sporadically through
life expectancy (for example, underdeveloped coun acquired somatic genomic alterations, whereas 35–40%
tries might have lower incidence rates even after age- of cases are associated with inherited CRC susceptibility.
standardization because fewer people reach the age of The heritable components are contributed by family
65–79 years, when most CRC is diagnosed) and data history of CRC, hereditary cancer syndromes, known
quality of cancer registries6. common but low-penetrance genetic variations and other
Temporal patterns of CRC incidence rates also vary inherited aberrations yet to be discovered. Regardless
of whether CRCs arise sporadically or have hereditary
globally. In South America, Eastern Europe and Asia,
components, environmental factors can influence their
economically transitioning countries are experienc carcinogenesis. For example, a substantial proportion of
ing increasing incidence rates (for example, Brazil, CRCs clustered in individuals with a positive family history
Slovakia and China)12. In high-income countries of are not inherited, but occur through acquired genomic
North America, Europe and Oceania, incidence rates aberrations. FAP, familial adenomatous polyposis; HNPCC,
are either decreasing following a peak (for example, hereditary nonpolyposis colorectal cancer.
www.nature.com/nrgastro
Reviews
a 35 b ASR(W)
per 100,000
30 ≥26.8
16.8–26.8
10.7–16.8
25 6.2–10.7
<6.2
Incidence–ASR(W)
Not available
20
No data
15
c Cumulative risk of CRC

10
development (0–74 years)
Hungary 6.1
Republic of Korea 5.3
5 Slovakia 5.3
Norway 5.0
Slovenia 5.0
0 Denmark 4.8
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Portugal 4.7
Human development index Barbados 4.7
Japan 4.5
The Netherlands 4.5
Africa Asia Europe Oceana
0 10 20 30 40 50 60 Incidence
Latin America and the Caribbean North America Mortality
ASR(W) per 100,000
Fig. 2 | Descriptive epidemiology of colorectal cancer. a | Ecological correlation between a country’s human
development index (HDI) and colorectal cancer (CRC) incidence rates worldwide in 2018. HDI is a composite score of life
expectancy, education and income, reflecting the economic development of a country11. Higher incidence rates of CRC
are observed in countries with higher values of HDI. b | Estimated incidence rates of CRC worldwide in 2018. There are
wide geographical variations in incidence rates of CRC. c | CRC incidence rates of the 10 highest CRC incidence countries
worldwide and corresponding mortality rates in 2018. The highest incidence rate of CRC was observed in Hungary in 2018.
ASR(W), age-standardized rate (cases per 100,000 persons per year) with use of age weights from the world standard
population. Data from GLOBOCAN 2018 (REF.6).
seem to be more strongly influenced by environmental racial disparity22. For instance, novel single-nucleotide
factors than by genetic factors in colorectal carcinogen polymorphisms (SNPs) associated with CRC risk have
esis, with heritability of CRC estimated at 28% for men been identified in a few GWAS conducted among black
and 45% for women3. A migration study also suggests individuals23,24, although our current understanding of
a larger contribution of environmental factors to CRC racial difference in the genetic architecture of CRC is
risk in men15. Among men who immigrated to Sweden limited due to predominance of European ancestry in
when aged <30 years, through five decades of residence GWAS25. However, no racial difference was reported
in Sweden, CRC incidence rates decreased for men from for the frequency of somatic mutations in known
higher-risk countries such as Norway but increased CRC driver genes (for example, APC, KRAS, TP53 and
for men from lower-risk countries such as Finland15. BRAF)22 and of microsatellite instability (MSI)-high
However, the shift in CRC risk towards the host country CRC26, a molecular subtype of CRC caused by muta
was less evident in women. In addition to their potentially tions in DNA mismatch repair genes (see natural his
higher vulnerability to environmental risk factors, men tory of CRC section)27. Furthermore, among those with
have a higher exposure to such risk factors (for example, Lynch syndrome (discussed later), the cumulative risk
visceral fat16, alcoholic beverages17, smoking18 and poor of hereditary CRC was similar between those of black
dietary patterns13) but lower uptake of CRC screening and European descents28. Moreover, the highest rates of
based on faecal samples (OR comparing men versus CRC in African Americans are in sharp contrast with
women: 0.84, 95% CI 0.75–0.79)19. Additionally, men the lowest rates observed in Africa relative to other
do not benefit from the protective effect of endogenous continents6. Collectively, evidence indicates that dis
oestrogen as women do (discussed later)20. Knowledge of proportionate exposure to modifiable risk factors and
these factors could inform prevention strategies. medical care (for example, screening and treatment)
Concerning racial or ethnic disparity, for example, in might predominantly drive the racial disparities.
the USA in 2015, rates standardized to the age distribu
tion of the 2000 USA standard population were highest Natural history of colorectal cancer
in black individuals and lowest in those with Asian or Key stages and features. The natural history of CRC can
Pacific Islander backgrounds in terms of both incidence be divided into four major stages: initiation, promotion,
(43.2 versus 28.8 cases per 100,000 persons per year) progression and metastasis29 (Fig. 5). Initiation involves
and deaths (18.6 versus 9.9 deaths per 100,000 persons irreversible genetic damage (such as DNA adducts) that
per year)21. Genetic factors could have a role in this predisposes the affected cells to subsequent neoplastic

Reviews
transformation. In the promotion stage, the initiated elusive and definitions of CIMP vary widely, with no
cells proliferate, inducing abnormal growth (neoplasm). consensus on markers and criteria to classify CIMP sub
In the subsequent progression phase, by undergoing fur types37. MSI is characterized by alterations in the length
ther genetic and epigenetic alterations that could confer a of microsatellite (that is, short nucleotide tandem repeats
selective growth advantage to cells, benign tumour cells in DNA sequences)27. It is driven by functional loss of
transform into malignant cancer cells and acquire aggres DNA mismatch-repair genes (for example, MLH1), with
sive characteristics and metastatic potential. Metastasis promoter hypermethylation suggested to be the major
is marked by the spread of cancer cells from the primary cause of gene silencing27. Although these molecular
organ to other organs or tissues through the bloodstream phenotypes have distinct characteristics, they are not
or the lymphatic system. The duration of each phase is mutually exclusive38. Most notably, CIMP and MSI sta
difficult to accurately estimate and has wide ranges; tus are correlated since CpG island hypermethylation is a
these processes generally take a long time and decades mechanism that inactivates DNA mismatch-repair genes
are required for all stages to be completed in CRC30. and leads to high levels of MSI36,39. Approximately 70% of
For hereditary CRCs, progression through some of the MSI-high CRC are also CIMP-high38. During colorectal
stages can be more rapid (discussed later). Notably, there tumourigenesis, these molecular aberrations accumulate
has been an increasing consideration of the presence of in a non-random fashion, with CIN, CIMP-positive and
cancer stem cells, which could have a role in colorectal MSI-high status observed in approximately 85%, 20%
carcinogenesis by dividing rapidly and continuously into and 15% of sporadic CRC, respectively30.
cancer cells or identical daughter cells and thus forming A key feature of most CRC carcinogenesis is the pres
a proliferative cancer cell population31. ence of a benign precursor lesion, defined as a polyp
Colorectal carcinogenesis encompasses three major (that is, abnormal growths on the lining of the large
global (as opposed to single) genetic and epigenetic intestine that protrude into the lumen)40. The grossly
aberrations: chromosomal instability (CIN), CpG visible intermediate lesions are readily removable dur
island methylator phenotype (CIMP) and MSI32,33. CIN ing a screening endoscopy and their progression to
is characterized by abnormalities in chromosomal copy CRC generally takes at least 10 years30,41. Thus, there is a
number (for example, aneuploidy and polyploidy) and unique window of opportunity for secondary prevention
structure, which are proposed to be caused by errors of CRC.
during mitosis, including defects in mitotic checkpoint Adenomatous polyps (adenomas) and serrated polyps
proteins and centrosome number34. CIMP, as a form of are two major subtypes that serve as direct precursors to
epigenetic modification, refers to hypermethylation at the majority of CRCs40. Adenomas represent the classic
repetitive CG dinucleotides (so called CpG islands) in precursor of CRC, because approximately 85–90% of
the promoter regions of tumour suppressor genes (such sporadic CRCs evolve from adenomas40. Yet, less than
as MLH1, MINT1, MINT2 and MINT3) that silence gene 10% of adenomas progress to CRC40. Advanced adeno
expression35,36. The underlying cause of CIMP remains mas (≥1 cm in diameter, villous histology or high-grade
a 70 b 70
60 60
50 50
ASR(W) per 100,000
ASR(W) per 100,000
40 40
30 30
20 20
10 10
0 0
1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005
Year Year
Brazil, Goiania USA, SEER (9 registries) China (3 registries) France (8 registries) Italy (8 registries) Norway
Slovakia Spain (7 registries) United Kingdom (9 registries) Australia (6 registries) New Zealand
Fig. 3 | time trends of colorectal cancer incidence rates. a | Incidence rates of colorectal cancer (CRC) in men across
selecteda countries. b | Incidence rates of CRC in women across selecteda countries. Countries undergoing economic
growth and westernization are experiencing increasing incidence rates (for example, Brazil, Slovakia and China).
In high-income countries, incidence rates are decreasing after peaking (for example, the USA, France and New Zealand),
relatively stable (for example, the UK and Australia) or increasing (for example, Italy , Norway and Spain). ASR(W),
age-standardized rate (incident cases per 100,000 persons per year) with use of age weights from the world standard
population. SEER , Surveillance, Epidemiology , and End Results Program. aCountries were selected based on group
(economically developing countries versus developed countries) and availability of time trend data on GLOBOCAN 2018.
Data from GLOBOCAN 2018 (REF.6).
Reviews
180 ensuing inactivation of TP53 tumour suppressor gene

160 contributes to the progression to CRC47. The adenoma–
Incidence (M)
Incidence (F)
carcinoma pathway is predominantly associated with the
140
Mortality (M) development of the CIN-positive CRC subtype, although
Rates per 100,000
120 Mortality (F) it remains unclear whether CIN underlies the accumula
100 tion of mutations in the critical tumour suppressor genes
and oncogenes, or vice versa48. Nevertheless, in a mouse
80
model with established colorectal tumours induced by
60 Apc knockout, restoration of APC function led to tumour
40 cell differentiation and sustained regression regardless of
Kras and Tp53 mutation status in lesions46.
20
An alternative carcinogenic pathway is the serrated
0 pathway, in which CRC develops through serrated ade
20–34 35–49 50–64 65–79
nomas (mainly sessile serrated adenoma)49,50. This model
Age (years)
is highlighted by the progression from normal cells to
Fig. 4 | Incidence and mortality of colorectal cancer by age and sex worldwide. hyperplastic polyp, to sessile serrated adenoma and,
Incidence rates and mortality rates of colorectal cancer increase with age. The rates are finally, to CRC49,50 (Fig. 5b). There are two molecular
higher in men than in women, with the difference becoming evident after age 50 years. events characteristic of the serrated pathway. Mutation
Rates are per 100,000 persons per year. F, female sex; M, male sex. Data from GLOBOCAN of the oncogene BRAF is regarded as a critical early
2018 (REF.6).
event which, by inducing uncontrolled cell proliferation
through the constitutive activation of the MAPK path
dysplasia) with or without multiplicity (>3 adenomas) way, contributes to the formation of hyperplastic polyp.
possess a particularly higher risk of progressing into can CIMP fosters subsequent progressions to sessile serrated
cer (30–50%) compared with non-advanced adenomas adenoma and CRC, and thereby CIMP-high adenomas
(1%)40, with estimated transition rates increasing with and CRC frequently arise from the serrated pathway49,50.
age at detection of advanced adenomas42. CIMP positivity was present in approximately 75% of
Serrated polyps represent a group of heterogene sessile serrated adenomas51–53 and 90% of serrated
ous lesions (hyperplastic polyp, traditional serrated adenocarcinomas in studies51.
adenoma, sessile serrated adenoma and mixed polyp) Another distinct carcinogenic pathway involving
combining the saw-toothed morphological appearance chronic inflammation has been suggested. In patients
of hyperplastic polyps and dysplastic features of adeno with inflammatory bowel disease (IBD), particularly
mas40. They are precursors to approximately 10–15% of ulcerative colitis, an estimated 2.4-fold higher risk of
sporadic CRC40. Among serrated polyps, hyperplastic CRC (95% CI 2.1–2.7) compared with the general popu
polyp is the most prevalent type (80–90%) and was previ lation was reported in a meta-analysis of eight cohort
ously considered to be lacking pre-malignant potential43. studies with an average follow-up of 14 years after initial
However, a subset of hyperplastic polyps (especially large diagnosis of ulcerative colitis54. In these patients, car
and/or in the proximal colon) can progress into CRC as cinogenesis progresses from no dysplasia to indefinite
part of the serrated pathway through traditional serrated dysplasia, low-grade dysplasia, high-grade dysplasia
adenoma or sessile serrated adenoma43,44. In a Danish and, finally, to CRC55 (Fig. 5c). Unlike discrete dysplastic
nationwide population-based case–control study, the OR precursor lesions (traditional adenomas or serrated ade
of CRC was 1.79 (95% CI 0.49–6.59, 3 cases and 10 con nomas) in other pathways, dysplasia arising on the back
trols) for traditional serrated adenoma, 3.40 (95% CI ground of chronic inflammation is frequently present in
2.35–4.91, 49 cases and 81 controls) for sessile serrated flat mucosa with multifocality, obscuring detection of
adenoma, and 2.50 (95% CI 2.24–2.80, 727 cases and the lesions. Thus, for dysplasia surveillance in patients
1637 controls) for conventional adenomas, compared with IBD, chromoendoscopy (dye spraying during
with individuals without a history of polyps45. endoscopy) or high-definition endoscopy are preferred
over traditional white-light colonoscopy56. The timing
Carcinogenic pathways. CRCs arise through distinct and frequency of molecular events are also distinct.
carcinogenic pathways: adenoma–carcinoma sequence, Contrary to the adenoma–carcinoma sequence, in which
serrated pathway and inflammatory pathway. The genetic mutations in APC and TP53 respectively occur
adenoma–carcinoma sequence, wherein adenoma early and late in colorectal carcinogenesis, in the inflam
serves as the precursor to CRC, is the classic pathway matory pathway, TP53 mutations represent an early
that explains the majority of CRCs. In this model, event, with APC mutations occurring infrequently and
gradual stepwise accumulation of genetic and epigenetic late in carcinogenesis55. In one study of 31 patients with
alterations drive the transformation of normal cells to CRC and IBD, APC mutation was observed in 13% of
small adenoma, to large adenoma and, finally, to cancer participants, which is much lower than 81% observed in
(Fig. 5a). Inactivating mutations in APC, a tumour sup patients with sporadic CRC without IBD57. By contrast,
pressor gene regarded as the gatekeeper against colorectal TP53 was mutated in approximately 60% of patients with
neoplasms, result in overactivation of the Wnt/β-catenin CRC regardless of IBD status57. Due to the low incidence
signalling pathway, triggering dysregulated cell prolifera of IBD and treatment with prophylactic colectomy,
tion and adenoma development46. Subsequent mutations the inflammation-associated carcinogenic pathway is
of the oncogene KRAS promote adenoma growth and estimated to explain less than 2% of all CRCs58.

Reviews
Initiation Promotion Progression Metastasis
30–60 years 10–20 years 0–5 years

a
Adenoma–carcinoma
sequence
Lung
• 85–90% of
sporadic CRC
• CIN-high Normal Small adenoma Large adenoma Cancer
b
Serrated pathway
• 10–15% of
sporadic CRC Normal Hyperplastic polyp Sessile serrated adenoma Cancer Liver
• CIMP-high
c
Bone
Inﬂammatory
pathway
• <2% of all CRC
Normal Indeﬁnite dysplasia Low-grade dysplasia High-grade dysplasia Cancer
Fig. 5 | pathways of colorectal carcinogenesis. Colorectal carcinogenesis occurs through four stages: initiation,
promotion, progression and metastasis (the liver is the most common metastatic site, followed by the lung and bone).
Although the time span of each stage is difficult to estimate, decades are required for all stages to be completed.
Colorectal cancer (CRC) develops through three distinctive pathways. a | Adenoma–carcinoma sequence. This classic
pathway explains the majority of sporadic CRC. Normal cells progress to small adenoma, to large adenoma and, finally,
to cancer. This pathway is predominantly associated with the development of the chromosomal instability (CIN)-positive
subtype. b | Serrated pathway. Accounting for 10–15% of sporadic CRC, this model is characterized by the progression
from normal cells, to hyperplastic polyp, to sessile serrated adenomas and, finally, to cancer. The CpG island methylator
phenotype (CIMP)-high subtype frequently arises from this pathway. c | Inflammatory pathway. Driven by chronic
inflammation, normal cells progress to indefinite dysplasia, to low-grade dysplasia, to high-grade dysplasia and,
finally, to cancer. Due to the low incidence of inflammatory bowel diseases and treatment with prophylactic colectomy,
this pathway explains less than 2% of all CRC. For all pathways, albeit more notable in the adenoma-carcinoma and
serrated pathways, there are benign precursor lesions that are accessible and removable; because they take years to
progress into cancer, there is a window of opportunity for secondary prevention of CRC.
Subtypes of colorectal cancer to 60% for ≥70 years in the USA)63. Although proxi
Although CRC develops in a single organ, namely the mal colon cancer is the most common subtype among
large intestine, it is a highly heterogeneous disease con white and black individuals (in the USA: for proximal,
sisting of subtypes with variant aetiology and clinical 44% and 49%, respectively; for distal, 27% and 26%,
outcomes. Traditionally, subtypes of CRC have been respectively; and for rectal, 29% and 25%, respectively)64,
defined by tumour anatomical site in three segments rectal cancer accounts for the highest proportion among
of the colorectum: proximal colon (caecum, ascending Asians (in Korea: for proximal, 22%; for distal, 26%; and
colon, hepatic flexure and transverse colon), distal colon for rectal, 52%)65. Yet, among Asian-Pacific Islanders
(splenic flexure, descending colon and sigmoid colon) in the USA, CRC cases were evenly distributed (about
and rectum59 (Fig. 6). Studies have shown that CRCs at 32–34% for each subsite) across the segments of the
different anatomical subsites have distinct risk factors colorectum64. This finding suggests stronger influences
(for example, smoking was associated with increased of environmental factors than genetic factors in the
risk of proximal colon cancer and rectal cancer but not pathogenesis of CRC.
with distal colon cancer)60,61. Aetiological heterogeneity Notably, molecular subtypes of CRC are also dispro
of CRC across the tumour locations might, in part, relate portionally distributed across the colorectum66 (Fig. 6).
to variations in microbial and host characteristics of the MSI-high and CIMP-high subtypes are enriched in the
large intestine. Along the colorectal axis from proxi proximal colon, whereas the CIN-positive subtype pref
mal colon to rectum, there is a progressive increase in erentially occurs in the distal colon66. With genetic and
pH, microbial loads and short-chain fatty acid (namely, epigenetic alterations reflecting the evolutionary his
microbial metabolites) abundance, which could have tory of a tumour, associations between global molecular
divergent implications for colorectal carcinogenesis62. aberrations and segments of the colorectum corrobo
As described previously, exposure to major CRC risk rate the aetiological heterogeneity of CRC across tumour
factors is heterogeneous by demographic factors. Thus, anatomical sites. Further refinement in the concept
aetiological heterogeneity by anatomical subsites of was made by the discovery from the Cancer Genome
CRC inevitably leads to varying distribution of subsite- Atlas project, which comprehensively profiled genomic
specific CRC cases by demographic factors (Fig. 6). changes in 20 cancer types67, that enrichment of hyper
Proximal colon cancer is more prevalent in women than methylation, MSI-high, CIMP-high and BRAF mutation
in men (34% versus 25%, respectively, in the European in the proximal colon was confined to hypermutated
Prospective Investigation into Cancer cohort)60 and cancer (mutation rates, >12 per 106 bases), which
its proportion increases with age (35% for <60 years comprised 16% of the CRC samples67. Distinct from
Reviews
hypermutated CRC, non-hypermutated CRC (mutation CIMP-high subtype32. Discrepant results on the adverse
rates, <8.24 per 106 bases) was characterized by fre prognostic effect of CIMP positivity37 might be, in part,
quent mutations in the well-known genes (for example, explained by adjustment for confounding by MSI sta
APC and TP53) associated with the CIN-p ositive tus74. To better account for confounding by molecular
subtype as well as in other genes (for example, SOX9, correlates, several classifications have been suggested
a gene associated with intestinal stem cell differentia that integrate multiple molecular markers to define a
tion)67. Interestingly, among non-hypermutated CRCs, subtype of CRC. In 2007, Jass proposed five CRC sub
colon and rectal cancers were indistinguishable at the types based on combinations of MSI and CIMP status
genomic level, although the analysis based on <200 cases and presence of BRAF and KRAS mutations75. In 2015,
had limited statistical power67. Overall, the findings the Colorectal Cancer Subtyping Consortium presented
suggest that the concept of aetiological heterogeneity four consensus molecular subtypes based on gene-
by anatomical subsite of CRC may be restricted to a expression profiles76. Nevertheless, a systematic review
subgroup of CRCs (hypermutated cases). published in 2018 concluded that prognostic values and
Molecular classifications of CRC provide insight into applicability of the proposed subtype classifications are
prognosis and treatment response. Considerable evi not optimal enough to be routinely implemented in
dence suggests that MSI-high CRC, despite its tendency clinical practice77.
towards unfavourable histology (such as poorly differ
entiated and mucinous), has a better prognosis than Genetic risk factors
non-MSI-high CRC (RR of all-cause mortality: 0.60, The cumulative risk of developing CRC before age
95% CI 0.53–0.69)68,69. A postulated mechanism relates 75 years is estimated to be 5% in the general population
to neoantigens — peptides produced by tumours as a from a high-incidence country (such as Korea, Norway,
result of MSI-induced frameshift mutations70. As novel Slovakia or Slovenia)6 (Fig. 2c). The lifetime risk of CRC
peptides, neoantigens are highly immunogenic, leading is considerably increased when individuals have a family
to the infiltration of tumours with neoantigen-specific history of CRC or hereditary cancer syndromes2,41.
cytotoxic T cells and, therefore, inducing an antitumour For individuals with a family history of CRC, a meta-
immune response71,72. Although MSI-high CRC might analysis of observational studies found the risk of CRC
not respond to adjuvant 5-fluorouracil-based chemo increased by 2.24-fold (95% CI 2.06–2.43) for those with
therapy68, it is considered as a good candidate for immu at least one affected first-degree relative (parents, sib
notherapy targeting immune checkpoint molecules lings, or children) and by 3.97-fold (95% CI 2.60–6.06)
(such as PD-1 or CTLA-4) that suppress T cell activity73. for those with at least two affected first-degree relatives78.
Associations between CRC subtype defined by a sin The associations became stronger when the relatives
gle molecular event and clinical outcomes are probably were diagnosed with CRC before age 50 years78. Except
inconsistent if correlation across molecular markers for rare hereditary cancer syndromes, most of the known
(that is, molecular correlates) are not adjusted for. For inherited mutations, albeit genetically predisposing to
instance, CIMP-high tumours largely overlap with MSI- CRC, are of low penetrance2. Thus, a substantial pro
high tumours, which comprise approximately 50–70% of portion of CRCs clustered in families are not inherited,
but occur through acquired genomic aberrations, which
points to the importance of environmental risk factors
Transverse
Proximal Distal colon in modulating CRC risk.
colon
colon • CIN-positive
Ascending
Although low-penetrance genetic variants contrib
• MSI-high (e.g. mutations
• CIMP-high colon in APC, KRAS, ute to family history of CRC2, these two factors might
• BRAF Descending TP53) represent distinct concepts. According to a genetic risk
colon
mutation • Men score incorporating 44 SNPs discovered to be related to
• Women Caecum • Younger age CRC risk from GWAS, individuals in the top decile had a
• Older age • FAP
• White, black
threefold increased risk of CRC compared with those in
Sigmoid
• HNPCC colon the lowest decile79. Interestingly, additional adjustment
Rectum
• Early-onset CRC (<50 years of age) for family history of CRC did not materially change the
• Asian result, suggesting that common low-penetrance variants,
in combinations, could confer an elevated risk of CRC,
Fig. 6 | anatomical subtypes of colorectal cancer and their associations with tumour independent of family history79. Indeed, in the joint ana
molecular features and other factors. The colorectum is anatomically partitioned into lysis, compared with individuals with no family history
three segments: proximal colon, starting from caecum through ascending colon to and very low genetic risk score, the OR was 1.68 for those
transverse colon; distal colon, including descending colon and sigmoid colon; and rectum. with affected first-degree relatives but very low genetic
Colorectal cancer (CRC) is aetiologically heterogeneous across anatomical location of risk, and further increased to 6.14 for those with affected
tumours. By demographic factors, proximal colon cancer is more prevalent in women, first-degree relatives and very high genetic risk79.
older individuals, and white and black individuals; distal colon cancer in men and At any given age, individuals affected with heredi
younger individuals; and rectal cancer in early-onset (diagnosis before age 50 years)
tary cancer syndromes are at higher risk of CRC than
and Asian individuals. By tumour molecular markers, proximal colon cancer is enriched
with subtypes characterized by microsatellite instability (MSI)-high, CpG island methylator the general population because they have germline
phenotype (CIMP)-high, or BRAF mutation, and distal colon cancer by the chromosomal mutations in high-p enetrance genes mostly follow
instability (CIN)-positive subtype. By hereditary cancer syndrome, hereditary nonpolyposis ing the autosomal dominant pattern80. Notably, indi
colorectal cancer (HNPCC)-associated CRC predominantly occurs in the proximal colon, viduals could develop hereditary cancer syndrome
while familial adenomatous polyposis (FAP)-associated CRC occurs in the distal colon. without a family history of the syndromes or CRC81.

Reviews
In this situation, possibly, parents have the relevant muta for stage I tumours; 94% versus 95%, P = 0.77 for stage II
tions only in germ cells (egg or sperm) but not in somatic tumours; 78% versus 83%, P = 0.09 for stage III tumours;
cells, or the mutations arose shortly after fertilization81. 39% versus 50%, P = 0.14 for stage IV tumours)93.
In HNPCC, PMS2 mutations have a reduced pene Although the aetiology of sporadic early-onset CRC
trance for CRC compared with mutations in the other remains elusive, considering the long timespan required
causative genes82. Thus, even if an individual inherited to complete carcinogenesis, exposures during early
PMS2 mutations and developed CRC, a family history life (childhood and adolescence) are likely to be crit
of CRC could be absent if the parent(s) carrying PMS2 ical determinants of the risk of early-onset CRC. For
mutations did not develop CRC. instance, men immigrating to Sweden before 30 years
HNPCC is the most common hereditary CRC syn of age reached the CRC risk level of native Swedes at
drome, with an estimated prevalence of 1 in 300 individ around 20 years stay, whereas those who immigrated
uals in western populations83. The hallmark of HNPCC after age 30 years showed no apparent shifts toward the
is the development of MSI-high CRC predominantly CRC risk level of the host country15. Because Australia,
located in the proximal colon84 (Fig. 6). The heredi Canada, New Zealand, the UK and the USA are highly
tary MSI-high CRC in HNPCC is primarily caused by westernized countries, newborn babies during the past
germline genetic mutations in any of the DNA mis decades in these countries might have been exposed to
match repair genes including MLH1, MSH2, MSH6 westernized diets and lifestyles that serve as predominant
and PMS2 (as opposed to epigenetic silencing of MLH1 risk factors for CRC (discussed later) from early life. This
for sporadic MSI-high CRC)36. Unlike ‘nonpolyposis’ aspect might have contributed to the rise in early-onset
in the name, affected individuals develop one or a few CRCs in these countries94–96, although major environ
polyps85, which progress to cancer within 2–3 years (as mental risk factors for late-onset CRC cannot be assumed
opposed to 8–10 years in the general population)86. CRC to serve as shared risk factors for early-onset CRC.
associated with HNPCC is often poorly differentiated, Available evidence, albeit scant, points to the rising
mucinous and abundant in infiltrating lymphocytes86. prevalence of obesity among children and adolescents
HNPCC increases the lifetime risk of CRC up to 60%41, as an important risk factor for early-onset CRC94. Time-
but accounts for 2–4% of all CRC cases2 (Fig. 1). trend data over the past decades have shown a close
The second-most common form of hereditary CRC parallel between the obesity epidemic in children and
syndrome is FAP, which affects approximately 1 in adolescents (aged 5–19 years)95 and the rising rates of
11,300–37,600 individuals in the European Union87. early-onset CRCs in Australia, Canada, New Zealand,
Caused by inherited germline APC mutations (as the UK and the USA (Fig. 7a,b). Furthermore, in a cohort
opposed to acquired somatic APC mutations), FAP is consisting of mainly white women in the USA, women
characterized by the development of hundreds to thou with a BMI of ≥23 at 18 years of age had an approx
sands of adenomas, mainly in the distal colon (Fig. 6), imately 60% increased risk of early-onset CRCs (95%
beginning in adolescence88. The presence of so many CI 1.01–2.61) compared with women with a BMI of
adenomas in FAP ensures that the risk of CRC virtually 18.5–20.9 at the equivalent age96. Although the rising
reaches 100% by age 40 years if the colon, and sometimes trend of early-onset cancer is more pronounced for rec
the rectum, is not prophylactically removed41. Yet, FAP tal cancer than for colon cancer, obesity is a stronger risk
explains <1% of all CRC cases88 (Fig. 1). factor for colon cancer than for rectal cancer in later-
onset cases97,98. This discrepancy indicates a multifac
Early-onset colorectal cancer torial nature of early-onset CRCs and more studies are
In countries such as Canada and the USA, with decreas warranted to identify the major risk factors.
ing or stabilizing trends of CRC incidence rates in the
past few decades12, an increasing trend of early-onset Lifestyle and nutritional factors
CRC (that is, CRC diagnosed before age 50 years) has Genetics contribute to individual risk2, but CRC inci
been witnessed, with the trend more evident for rec dence in a population are largely affected by modifiable
tal cancer than for colon cancer6,89 (Fig. 7a,b). Although diet and lifestyle factors because rates can change drama
hereditary cancer syndromes, family history of CRC and tically over short courses of time and migrants from
IBD predispose individuals to early-onset CRC2,55, the countries with low CRC rates rapidly take on the high
majority of early-onset CRCs arise sporadically from rates of their adoptive country15,99. The 2017 extensive
individuals at average risk90. Among early-onset CRCs summary report by the World Cancer Research Fund
diagnosed mostly from white individuals, only 13% had (WCRF) and American Institute of Cancer Research
germline mutations in genes underlying hereditary CRC (AICR), based on a systematic review of studies availa
syndromes (for example, MLH1, MSH2, MSH6, PMS2 ble globally, concluded that obesity, low physical activity,
and APC)90. Compared with late-onset CRCs, sporadic poor diets (such as high red and processed meat, low
early-onset CRCs have a higher proportion of rectal fibre, low whole grain and low calcium) and alcohol
cancer (Fig. 6) and develop histologically unfavourable increase CRC risk9 (Table 1). By population-attributable
subtypes such as poorly differentiated, mucinous adeno fraction, a measure of public health impact of exposure
carcinomas91,92. Early-onset CRCs are often detected at to a risk factor on a disease outcome in a population100,
a more advanced stage91,92, but a large prospective study 47% of CRC cases in the USA and 45% in the UK were
observed no statistically significant difference in 5-year estimated to be attributable to the aforementioned mod
cancer-specific survival between early-onset and late- ifiable risk factors9. In addition, smoking increases CRC
onset CRCs within each stage (99% versus 98%, P = 0.19 risk101. This constellation of factors is probably the driver
Reviews
a
Australia Canada
12 140 12 140
Rates of early-onset
Rates of late-onset
Rates of late-onset
cases per 100,000
cases per 100,000
cases per 100,000
cases per 100,000

11 130 11 130
10 120 10 120
9 110 9 110
8 100 8 100
7 90 7 90
6 80 6 80
5 70 5 70
4 60 4 60
3 50 3 50
2 40 2 40
19 5
19 6
19 7
19 8
20 9
20 0
20 1
20 1
20 3
20 4
20 5
20 6
20 7
20 8
20 9
10
19 5
19 7
19 9
19 1
19 3
19 5
19 7
20 9
20 1
20 3
20 5
20 7
09
9
9
9
9
9
0
0
0
0
0
0
0
0
0
0
8
8
8
9
9
9
9
9
0
0
0
0
19
19
Year New Zealand Year
12 140
cases per 100,000
cases per 100,000

ates of late-onset
11 130
10 120
9 110
8 100
7 90
6 80
5 70
4 60
3 50
2 40
19 5
19 7
19 9
19 1
93
19 5
19 7
20 9
01
20 3
20 5
20 7
09
8
8
8
9
9
9
9
0
0
0
19
19
20
Year
Scotland USA, SEER (9 registries)
12 140 12 140
Rates of late-onset
Rates of late-onset
cases per 100,000
cases per 100,000
cases per 100,000
cases per 100,000

11 130 11 130
10 120 10 120
9 110 9 110
8 100 8 100
7 90 7 90
6 80 6 80
5 70 5 70
4 60 4 60
3 50 3 50
2 40 2 40
19 0
82
19 4
19 6
19 8
19 0
19 2
19 4
19 6
20 8
20 0
20 4
20 6
20 8
10
19 0
19 2
19 4
19 6
19 8
19 0
19 2
19 4
19 6
20 8
20 0
20 4
20 6
20 8
10
8
8
8
8
9
9
9
9
9
0
0
0
0
8
8
8
8
8
9
9
9
9
9
0
0
0
0
19
19
19
Year Year
b
Australia Canada
12 90 12 90
Rates of late-onset
Rates of late-onset
cases per 100,000
cases per 100,000
cases per 100,000
cases per 100,000

11 11
10 80 10 80
9 9
8 70 8 70
7 7
6 60 6 60
5 5
4 50 4 50
3 3
2 20 2 20
19 5
19 6
19 7
19 8
20 9
20 0
20 1
20 1
20 3
20 4
20 5
20 6
20 7
20 8
20 9
10
19 5
19 7
19 9
19 1
93
19 5
19 7
20 9
01
20 3
20 5
20 7
09
9
9
9
9
9
0
0
0
0
0
0
0
0
0
0
8
8
8
9
9
9
9
0
0
0
19
19
19
20
Year New Zealand Year
12 90
Rates of late-onset
cases per 100,000
cases per 100,000
11
10 80
9
8 70
7
6 60
5
4 50
3
2 20
19 5
87
19 9
19 1
19 3
19 5
19 7
20 9
20 1
20 3
20 5
20 7
09
8
8
9
9
9
9
9
0
0
0
0
19
19
Year
Scotland USA, SEER (9 registries)
12 90 12 90
Rates of late-onset
Rates of late-onset
cases per 100,000
cases per 100,000
cases per 100,000
11 11 cases per 100,000

10 80 10 80
9 9
8 70 8 70
7 7
6 60 6 60
5 5
4 50 4 50
3 3
2 20 2 20
19 0
82
19 4
19 6
19 8
19 0
19 2
19 4
19 6
20 8
20 0
20 4
20 6
20 8
10
19 0
19 2
19 4
19 6
19 8
19 0
19 2
19 4
19 6
20 8
20 0
20 4
20 6
20 8
10
8
8
8
8
9
9
9
9
9
0
0
0
0
8
8
8
8
8
9
9
9
9
9
0
0
0
0
19
19
19
Year Year
Male (25–49 years) Female (25–49 years)
Fig. 7 | time trends of early-onset and late-onset colorectal cancer incidence rates. a | Incidence rates of
early-onset and late-onset colon cancer by sex across selecteda countries. In contrast to decreasing incidence
rates of late-onset colon cancer, early-onset colon cancer shows a modestly increasing trend of incidence rates over
time, particularly in Canada and the USA. b | Incidence rates of early-onset and late-onset rectal cancer by sex across
selecteda countries. In contrast to decreasing or stabilizing incidence rates of late-onset rectal cancer, early-onset
rectal cancer shows an increasing trend of incidence rates over time. Early-onset, diagnosis between age 25–49 years;
late-onset, diagnosis between age 50–74 years. Rates are per 100,000 persons per year. F, female sex; M, male sex;
SEER , Surveillance, Epidemiology , and End Results Program. aCountries were selected based on trend of late-onset
CRC incidence rates (decreasing or stabilizing) and availability of time trend data on GLOBOCAN 2018. Data from
GLOBOCAN 2018 (REF.6).

Reviews
Table 1 | summary of the associations between risk or protective factors and colorectal cancer risk by anatomical subsites
aetiological factors level of Unit increase Colorectal cancer Colon cancer rr rectal cancer refs
evidencea rr (95% CI) (95% CI) rr (95% CI)
Obesity ↑↑ 5 kg/m2 in BMI 1.05 (1.03–1.07) 1.07 (1.05–1.09) 1.01 (1.01–1.04) 116
↑↑ 10 cm in WC 1.02 (1.01–1.03) 1.04 (1.02–1.06) 1.02 (1.00–1.03)

Total physical activity ↓↓ 5 MET-hours per week 0.97 (0.94–0.99) 0.92 (0.86–0.99) 1.02 (0.95–1.10) 116
Western dietary pattern ↑↑ Highest versus lowest 1.12 (1.01–1.24) 1.30 (1.04–1.63) 1.09 (0.91–1.29) 133
Prudent dietary pattern ↓↓ Highest versus lowest 0.89 (0.84–0.95) 0.89 (0.80–0.99) 0.96 (0.83–1.10) 133
Processed meat ↑↑ 50 g per day 1.16 (1.08–1.26) 1.23 (1.11–1.35) 1.08 (1.00–1.18) 116
Red meat ↑ 100 g per day 1.12 (1.00–1.25) 1.22 (1.06–1.39) 1.13 (0.96–1.34) 116
Total fibre ↓ 10 g per day 0.93 (0.87–1.00) 0.91 (0.84–1.00) 0.93 (0.85–1.01) 116
Whole grain ↓ 90 g per day 0.83 (0.79–0.89) 0.82 (0.73–0.92) 0.82 (0.57–1.16) 116
Alcohol (as ethanol) ↑↑ 10 g per day 1.07 (1.05–1.09) 1.07 (1.05–1.09) 1.08 (1.07–1.10) 116
Smoking b
↑ Current versus never smokers 1.15 (1.00–1.32) 1.10 (0.89–1.36) 1.19 (0.94–1.54) 101
Aspirinb ↑↑ 75–1200 mg per day versus control 0.76 (0.63–0.94) 0.76 (0.60–0.96) 0.90 (0.63–1.30) 185
Total calcium b
↓ 300 mg per day 0.92 (0.89–0.95) 0.91 (0.87–0.96) 0.95 (0.83–1.08) 194
↑↑, convincing risk factor ; ↑, probable risk factor ; ↓↓, convincing protective factor ; ↓, probable protective factor ; BMI, body mass index; CI, confidence interval;
CRC, colorectal cancer ; MET, metabolic equivalent of task; RR , relative risk; WC, waist circumference. aLevel of evidence as indicated by WCRF−AICR summary
report for CRC9, except for smoking and aspirin (based on evidence from observational studies and randomized controlled trials).bLong latency was required to
observe an effect on CRC.
of increasing CRC incidence in populations undergoing significantly associated with increased colorectal ade
economic transition. noma risk after adjustment for SAT (RR comparing
extreme categories: 2.51, 95% CI 1.56–4.02), but SAT
Obesity. Excess adiposity is an established risk factor was not after adjustment for VAT (RR comparing
for CRC (stronger for colon cancer than rectal cancer), extreme categories: 1.12, 95% CI 0.72–1.73)109, which
which is consistently supported by epidemiological suggests an underlying role of VAT in linking excess
studies using diverse anthropometric measures97,98. The adiposity to colorectal carcinogenesis.
two most commonly used measures are BMI, which For any given BMI, visceral obesity is more prevalent
represents overall body fatness, and waist circumference in Asians than white populations110 and in men than in
(WC), which largely reflects abdominal fatness. Some women16, possibly owing to genetic variations111 and sex
evidence suggests that WC is a stronger risk factor for hormones112. This disparate susceptibility towards vis
CRC than BMI102,103. For instance, when BMI and WC ceral obesity corresponds to a sharply elevated risk of
were simultaneously included in the statistical model for CRC in Asian individuals within the normal BMI range
colon cancer risk, the relative risk for BMI diminished (RR comparing 23.0–24.9 versus <23.0 kg/m2: 1.21, 95%
but not the relative risk for WC103. Approximately, a CI 1.05–1.41)113, and a stronger association between BMI
10 cm increase in WC was associated with a 4% increased and CRC risk observed in men (RR per 5 kg/m2: 1.24,
risk of colon cancer (Table 1). 95% CI 1.18–1.31) than in women (RR per 5 kg/m2:
Abdominal fat is further categorized into two distinct 1.08, 95% CI 1.02–1.34)97. Besides VAT, other factors
compartments: visceral adipose tissue (VAT) and sub could also explain a weaker association between obesity
cutaneous adipose tissue (SAT). Compared with SAT, and CRC risk in women114. Most of the available stud
VAT secretes more proinflammatory adipokines (such as ies on obesity and CRC risk have focused on adulthood
TNF) and less adiponectin (an insulin-sensitizing hor obesity, but obesity in early life could also be aetiolog
mone)104, and is more heavily infiltrated with immune ically relevant to CRC risk in women. For instance,
cells (such as macrophages)105. All of these traits contri in a Mendelian randomization study in which genetically
bute to the development of chronic low-grade systemic determined BMI preferentially reflects early-life adipos
inflammation and insulin resistance106. As exemplified ity, the Mendelian BMI score was positively associated
by colitis-induced CRC, inflammatory conditions in with CRC risk in women (RR per 5 kg/m2 as estimated
the tumour microenvironment promote tumour growth by genetic scores: 1.82, 95% CI 1.26–2.61)115, which
and progression107. However, it is not known if obesity- is much more pronounced than the adult BMI–CRC
related inflammation increases CRC risk directly or sec association in women (RR per 5 kg/m2: 1.05, 95% CI
ondarily through processes such as insulin resistance. 1.02–1.08)116. Endogenous oestrogens confer protection
Insulin resistance and subsequent hyperinsulinaemia against CRC only in women and adipocytes become the
lead to an increase in free insulin-like growth factor 1 main site for oestrogen production after menopause20,117.
(IGF1), and the insulin−IGF1 signalling pathway has Thus, concerning excess adiposity of women in later life,
been suggested to promote colorectal carcinogenesis by its cancer-promoting effect through insulin and IGF1
increasing cell proliferation and decreasing apoptosis108. could be counterbalanced by the anticancer effect of
In a meta-analysis of observational studies, VAT was oestrogens.
Reviews
Physical activity and sedentary lifestyle. CRC, espe Dietary patterns can be derived a posteriori from
cially colon cancer, is one of few cancers for which lack a dataset to identify the uncorrelated habitual dietary
of physical activity is acknowledged as a risk factor118. patterns of the population. Across studies conducted in
There is no definitive answer to the optimal intensity different populations worldwide, two dietary patterns
and dose of physical activity for CRC prevention, but have been consistently observed: the ‘healthy’ pattern
the American Cancer Society recommends adults (characterized by high intakes of fruits and vegetables
engage in at least 150 min of moderate-intensity activity as well as at least one of whole grains, nuts or legumes,
(3–5.9 metabolic equivalents of task (MET)), 75 min of fish or other seafood, and low-fat milk or dairy products;
vigorous-intensity activity (≥6 MET) or an equivalent for example, prudent dietary pattern) and the ‘unhealthy’
combination of the two (of note, 2 min of moderate pattern (characterized by high intakes of red and pro
intensity activity = 1 min of vigorous intensity activity) cessed meat, sugar-sweetened beverages, refined grains,
throughout the week119,120. Flexibility in the choice of desserts and potatoes; for example, western dietary pat
exercise intensity implies the importance of accumulat tern)132. A meta-analysis of cohort studies found that
ing a certain level of energy expenditure through phys the RR of CRC, comparing the highest versus lowest
ical activity rather than specific regimes of exercise. An adherence to a dietary pattern, was 1.12 (95% CI 1.01–
increase in the amount of exercise by 5 MET-hours per 1.24) for the western dietary pattern and 0.89 (95% CI
week was associated with an approximately 8% reduced 0.84–0.95) for the prudent pattern, with the associations
risk of colon cancer116 (Table 1). stronger for colon cancer than for rectal cancer133.
Physical activity might reduce CRC risk through Biological mechanisms linking healthy and unhealthy
its beneficial effects on gut motility, the immune dietary patterns and CRC are probably multifaceted,
system, inflammation and metabolic hormones 121. reflecting a complex interplay of various dietary com
These responses might be, in part, the direct con ponents. For instance, in the unhealthy dietary pattern,
sequences of physical activity, but could be largely red and processed meat could directly contribute to
mediated through VAT loss 122. Among adults who colorectal carcinogenesis. According to a meta-analysis
were overweight or obese, 3–4 months of intervention of prospective studies by WCRF−AICR, each 100 g
engaging in daily walking or jogging reduced VAT per day increase in red and processed meat intake was
(P <0.01), even in the absence of overall weight associated with 12% increased CRC risk (95% CI 1.04–
loss123,124. In a meta-analysis of randomized controlled 1.21)134, with the association stronger with processed
trials (RCTs), significant VAT loss was induced by aer meat than with red meat (Table 1). The influence of red
obic exercise (standardized mean difference: –0.33 unit and processed meat on CRC can occur through carcino
of standard deviation, P <0.001) but not by resistance genic compounds such as haem iron from red meat,
exercise125. Consistently, in a cohort study that cross- exogenous N-nitroso compounds from processed meat,
classified participants by level of aerobic exercise and ionized fatty acids and secondary bile acids attributable
engagement in weight lifting, the greatest reduction to fat in meats, and heterocyclic amines and polycyclic
in the risk of digestive system cancers (of which CRC aromatic hydrocarbons formed when meats are cooked
accounted for 56%) occurred in the most active group at high temperatures135–138.
exclusively engaging in aerobic exercise, with the opti A novel approach that offers insight into biological
mal level suggested to be 30 MET-hours per week pathways from diet to CRC development is the deriva
(RR 0.68, 95% CI 0.56–0.83)126. tion of dietary patterns associated with biomarkers of
Regardless of the level of physical activity, prolonged pathways hypothesized to influence CRC risk (such as
sitting, as captured by long hours spent watching TV, is insulin and inflammation)108. In large cohorts of men
emerging as a risk factor for CRC127, with each 2 h per day and women, food groups most predictive of circulat
increase in TV viewing hours associated with a 7% ele ing C-peptide levels (a marker of insulin secretion)
vated CRC risk (95% CI 1.05–1.10)128. Prolonged sitting and inflammatory markers (such as C-reactive pro
impairs skeletal muscle function, which facilitates tein (CRP), IL-6 and TNF receptor 2) were selected to
insulin resistance and promotes colorectal carcino define C-peptide and proinflammatory dietary patterns,
genesis129. Notably, a cross-sectional study found that, respectively139,140. These dietary patterns were shown to
independent of total sitting time, increased breaks in be partially characterized by a low intake of food groups
sedentary time (for example, frequent transition from of a healthy diet but a high intake of food groups of an
sitting to standing) was inversely associated with met unhealthy diet. For instance, both C-peptide and pro
abolic markers implicated in colorectal carcinogenesis inflammatory dietary patterns are low in whole grain
such as BMI, WC and 2-hour plasma glucose levels intake, but high in meat and sweetened beverage intakes.
(P = 0.03 for all)130. Positive associations of C-peptide and proinflammatory
dietary patterns with CRC risk139,140 suggest that hyper
Dietary patterns. To reflect actual eating behaviours insulinaemia and inflammation could mediate, in part,
and account for the potential additive and interactive the effects of overall diet on CRC risk.
effects of foods consumed together, nutrition research
to identify the optimal diet for disease prevention has Fibre and whole grains. In the early 1970s, Denis Burkitt
shifted its focus from individual nutrients or specific attributed the rarity of CRC among rural Africans to
foods to overall dietary patterns131. A dietary pattern, by higher fibre intakes upon observing their swiftly passed
the combined effect of its components, could increase or bulky stool141. Dietary fibre, particularly insoluble fibre,
decrease CRC risk. reduces the exposure of the colorectal epithelium to

Reviews
carcinogens in the lumen by decreasing the transit time colorectal carcinogenesis by causing DNA damage and
and increasing the bulk of faeces142. Another proposed destroying intracellular folate, required for proper DNA
mechanism involves interaction with intestinal micro synthesis and methylation157,158.
biota. Soluble fibre reaching the colon undigested is read ALDH2 enzyme is abundant in the liver and its well-
ily fermented by the anaerobic intestinal microbiota into known genetic polymorphism ALDH2*2 leads to ele
short-chain fatty acids (mainly acetate, propionate and vated circulating levels of acetaldehyde, which can reach
butyrate)143. In vitro studies of human intestinal cell lines colonocytes159. The ALDH2*2 variant is highly prevalent
showed that butyrate promoted the survival of normal among East Asians (as known by ‘Asian Flush’) but vir
colonocytes and apoptosis of neoplastic colonocytes144, tually absent in other parts of the world160. According
and downregulated proinflammatory pathways 145. to pooled studies, the RR associated with ≥45 g per day
Additionally, Fusobacterium nucleatum, found enriched intakes of alcohol (about three alcoholic drinks) was
in CRC tissue relative to adjacent normal tissue, has 2.09 (95% CI 1.65, 2.64) in Japan161 but 1.41 (95% CI
been proposed to promote colorectal carcinogenesis by 1.16–1.72) in North America and Europe155. Yet, direct
contributing to an immunosuppressive tumour micro evidence for the role of ALDH2*2 in colorectal car
environment146. Diets rich in fibre and whole grains were cinogenesis remains inconclusive159. A meta-analysis
associated with a reduced risk of F. nucleatum-positive even found an approximately 20% reduced CRC risk
CRC but not with a F. nucleatum-negative subtype147. for carriers of one ALDH2*2 allele, possibly because
Despite the strong mechanistic plausibility link of the unpleasant discomforts (such as facial flush
ing fibre intake and CRC risk, epidemiological studies ing, nausea or headache) that carriers experience after
have shown divergent results by food sources of fibre. drinking preventing them from frequent alcohol con
In a meta-analysis of prospective observational stud sumption162. However, the lack of subgroup analysis
ies, cereal fibre was inversely associated with the risk among non-drinkers and drinkers limits our understand
of CRC (RR for 10 g per day increment: 0.90, 95% CI ing of the interaction between alcohol consumption and
0.83–0.97) but fruit, vegetable and legume fibres were ALDH2*2 (REF.162).
not148. The same pattern was observed with adenoma
outcome149. Likewise, among patients diagnosed with Smoking. Cigarette smoke contains a mixture of com
stage I to III CRC, a reduced CRC mortality associated pounds that can readily reach the colorectal mucosa
with a higher post-diagnostic fibre intake was confined through the circulatory system or direct ingestion and
to cereal fibre (RR for 5 g per day increment: 0.67, 95% induce genetic and epigenetic aberrations163. Meta-
CI 0.50–0.90)150. Whole grains, a major source of cereal analyses of observational studies found CRC risk
fibre, are inversely associated with CRC incidence increasing with pack-years (RR for 5 pack-years: 1.06,
and mortality148,150. However, in a RCT of wheat-bran 95% CI 1.03–1.08; RR for 30 pack-years: 1.26, 95% CI
fibre supplements (13.5 versus 2 g per day), 3 years 1.17–1.36)164, and decreasing with age of smoking ini
of high fibre supplementation did not confer benefit tiation (approximately 4% reduction per 10-year delay,
against colorectal adenoma recurrence151. Reflecting P <0.0001)101.
the uncertainty, the CRC report by WCRF−AICR Smoking might affect CRC risk differentially by anato
demoted the level of evidence supporting a protective mical subsites. In a cohort study based on 10 European
role of dietary fibre from ‘convincing’ in 2011 (ref.152) tocountries published in 2018, current smoking was asso
‘probable’ in 2017, and added whole grain as a probable ciated with an elevated risk of proximal colon cancer
protective factor9. (RR 1.19, 95% CI 1.05–1.34) and rectal cancer (RR 1.27,
95% CI 1.14–1.42) but not with distal colon cancer
Alcohol. Ethanol in alcoholic drinks of any type is an (HR 1.08, 95% CI 0.94–1.23)60. Consistent findings
established risk factor for CRC, with its first metabolite, were reported from molecular epidemiological studies.
acetaldehyde, evaluated as carcinogenic to humans by Proximal colon cancer is enriched with MSI-high, CIMP-
the International Agency for Research153. In the meta- high, BRAF-mutant subtypes66. Current smoking was
analysis of 14 cohort studies from North America, associated with an approximately twofold increased risk
Europe and Asia published in 2018, even light drink of these CRC subtypes, but not associated with their coun
ing (≤1 alcoholic drink per day) was associated with a terparts such as CIMP-negative and BRAF-wild type165 or
slightly but significantly increased CRC risk (RR 1.04, only marginally associated with MSI-low or microsatellite
95% CI 1.01–1.06) compared with no or occasional stable166. BRAF mutations and CIMP are the key molec
drinking of alcohol154. Generally, the association is ular alterations driving the serrated pathway and smok
stronger in men than in women, probably due to higher ing was more strongly associated with serrated polyps
alcohol intake in men155. (RR 2.33, 95% CI 1.76–3.07) than with classical adeno
Ingested alcohol reaches colonocytes through the sys mas (RR 1.31, 95% CI 1.08–1.58)167. Taken together, epi
tematic circulation and probably diffuses into the lumen. genetic alterations seem to be an important contributor
The luminal ethanol is metabolized by microbial alco to smoking-induced colorectal neoplasms and, indeed,
hol dehydrogenase into acetaldehyde156, which causes a genome-wide methylation study observed extensive
mucosal injury and regenerative cellular prolifera changes in DNA methylation patterns in smokers com
tion157. The toxic acetaldehyde also enters the intestinal pared with never smokers168. Notably, the study observed
epithelial cells and accumulates due to the low activ similar methylation levels between former and never
ity of colonic mucosal acetaldehyde dehydrogenase smokers, suggesting smoking cessation could revert
(ALDH)156. Intracellular acetaldehyde might promote aberrant methylation to the normal168. Compared with
Reviews
current smoking, at least 10 years of smoking cessation colorectal adenoma, benign polyps and ‘clean controls’
was associated with approximately 50% reduced risk of without colorectal neoplasms, CRC-associated micro
CIMP-high CRC (Ptrend = 0.001) but not with CIMP-low organisms (such as F. nucleatum) and plasma inflam
or CIMP-negative subtypes169. matory markers (such as CRP levels) tended to change
Smoking could affect rectal cancer through a diffe along the adenoma−carcinoma sequence, with CRP level
rent mechanism than CIMP-high cancer163, which is an positively correlated with microorganisms increased
infrequent subtype among rectal cancer170. Unlike the in patients with CRC (for example, F. nucleatum) but
short-term, reversible effect of smoking on the CIMP- negatively correlated with microorganisms decreased in
high subtype169, for rectal cancer, smoking seems to have patients with CRC (for example, Eubacterium eligens)178.
an irreversible effect that requires a long time to emerge. With diet179, exercise180 and weight change181,182 capable
For example, in a cohort study of insured US veterans of inducing gut microbiota composition changes even
mostly from World War I, the RR of death for current within several weeks, future studies are warranted to bet
smokers was 1.1 for both colon and rectal cancer after ter understand the interactions between environmental
16 years of follow-up from 1954 through 1969 (REF.171). factors and the gut microbiota and their influences on
When an effect emerged after 26 years of follow-up in colorectal carcinogenesis. Additionally, there is growing
1980, an increased risk of death was more prominent for interest in the contribution of non-bacterial components
rectal cancer (RR 1.4, 95% CI 1.2–1.7) than colon cancer of the gut microbiota to colorectal carcinogenesis, with
(RR 1.2, 95% CI 1.1–1.4)172. The lagged effect suggests studies having identified CRC-associated faecal mark
that smoking might act early on the pathogenesis of rec ers of fungi (such as Malasseziomycetes)183 and viruses
tal cancer. Tobacco carcinogens (such as polycyclic aro (such as Orthobunyavirus)184.
matic hydrocarbons) form DNA adducts, which cause
irreversible damage to DNA163. This harm of smoking is Chemopreventive potential
unlikely to be mitigated by smoking cessation. Aspirin. Aspirin, a non-steroidal anti-inflammatory
drug, has been extensively examined for its chemopre
Other potential aetiological factors ventive potential against CRC. In an individual partici
In addition to established risk factors for CRC, there pant data meta-analysis of four RCTs, daily aspirin use
are additional promising but unproven hypotheses that (75–300 mg) was inversely associated with long-term
could offer insights to unexplained features of the epi risk of CRC development (HR 0.75, 95% CI 0.56–0.97)
demiology of CRC. The higher risk of CRC and height and death (HR 0.61, 95% CI 0.43–0.87) 185. In the
ened susceptibility to visceral obesity in men relative Women’s Health Study, alternate-day use of low-dose
to women remain unexplained. An obvious explana (100 mg) aspirin had no effect on CRC incidence dur
tory candidate is sex hormones. Evidence suggests that ing 10 years of active intervention period186, but reduced
high levels of endogenous oestrogens confer protection the risk (RR 0.58, 95% CI 0.42–0.80) after an additional
against CRC in women and high levels of endogenous 8 years of post-trial observational follow-up187. A latency
testosterone might lower risk in men173. Interestingly, of 10 years was suggested in a prospective observational
increasing adiposity with ageing has differential effects study188. The delayed effect implies that aspirin might
on these hormones; adipocytes become the main site act on early stages of colorectal carcinogenesis. Indeed,
for oestrogen production after menopause in women174, aspirin was effective in preventing adenoma recurrence
whereas testosterone levels decline with adiposity in after 3–4 years of intervention189. The optimal dose and
men175. Thus, the positive relationship between obesity duration of aspirin use for the prevention of CRC remain
and CRC risk might be attenuated in postmenopausal inconclusive. A cohort study with 32 years of follow-up
women with increased oestrogens from excess adiposity, estimated that the minimum intake of 0.5 to 1.5 standard
but amplified in men with decreasing testosterone. (325 mg) aspirin tablets per week for at least 6 years of
Another unresolved question concerns the dispropor regular use might be required to obtain benefit against
tionate distribution of molecular subtypes of CRC across gastrointestinal cancer including CRC190.
anatomical segments of the large intestine66. A plausi Aspirin could exert its anticancer effect in part by
ble contributor is the gut microbiota, which also varies inhibiting cyclooxygenase 2 (COX2), an enzyme that
in its abundance and composition along the colorectal fuels tumour-promoting inflammation and suppresses
axis62,176,177. Advances in characterizing the microbiome T cell-mediated antitumour immunity191. In a cohort
have offered insights for the aetiology of CRC. study that integrated tumour tissue biomarkers, the
In a study of biopsy samples of CRC and non- chemopreventive benefit of regular aspirin use was
cancerous tissues, the mucosal microbiota profile dif confined to CRC subtype dependent on COX2 for
fered not only between proximal cancers (for example, tumour growth, as indicated by COX2 overexpres
higher abundance of some Clostridium species) and sion192, and to CRC subtype with suppressed antitumour
distal cancers (for example, higher abundance of some immune response as reflected by low-level lymphocytic
Alistipes species), but also across non-cancerous tissues infiltrates193.
from patients with CRC (for example, higher abundance
of Bacteroides or F. nucleatum), individuals with polyps Calcium. An inverse association between calcium intake
and healthy individuals as controls177. The gut microbi and CRC risk has been found in long-term observational
ota could be implicated in colorectal carcinogenesis by studies (RR for 300 mg per day increment of total cal
promoting inflammation. In a study that collected stool cium 0.92, 95% CI 0.89–0.95)194, but not in RCTs of
and blood samples from patients with CRC, advanced calcium supplements alone195 or calcium and vitamin D

Reviews
supplements combined196,197. In observational studies, which tests for both bleeding (through immunoassay
an inverse association was found from both food and for haemoglobin) and molecular aberrations associated
supplemental sources, and over a wide range of intake with colorectal neoplasms (through quantitative molec
(250–1900 mg per day), suggesting that a calcium ular assays for biomarkers such as KRAS mutations and
intake higher than the recommended dietary allowance aberrant NDRG4 and BMP3 methylation)206. In a study
for adults (1000–1200 mg per day) might be beneficial of asymptomatic individuals at average risk of CRC,
in preventing CRC194. The failure to confirm the benefit in
compared with FIT, FIT-DNA had a higher sensitivity
RCTs could indicate the absence of a causal effect, but afor detecting advanced precancerous lesions (42.4%
versus 23.8%, P <0.001) and CRC (92.3% versus 73.8%,
latent effect of calcium on CRC risk is plausible. A large
prospective study of 47,740 men and 88,509 women P = 0.002), but lower specificity (86.6% versus 94.9%
found a required lag of approximately 10 years to among individuals with non-advanced or negative find
observe an effect of calcium intake on CRC risk198. ings, P <0.001)206. Yet, in another population representing
In observational studies, higher calcium intakes were an average risk of CRC, FIT had a slightly higher sen
associated with reduced risk of colorectal adenoma199, sitivity than the reported FIT-DNA test207 at the same
the likely precursor to most sporadic CRC. Considered specificity level of 86.6%206. The FIT-DNA test, owing to
together, calcium might inhibit the progression from its lower specificity that likely requires more diagnostic
pre-adenoma to adenoma. Because adenoma takes at colonoscopies and its substantially higher cost and need
least 10 years to progress to CRC30,41, for RCTs with a for collection of an entire stool sample, has a limited
CRC end point, <5 years of follow-up are insufficient tovalue as a screening tool208.
capture the effect of calcium acting early in colorectal Colonoscopy, the gold-standard screening method, is
carcinogenesis. On the contrary, for RCTs with an ade a highly sensitive test but involves the highest cost and
noma end point, a recent meta-analysis including three resources204. RCTs for screening colonoscopy are not yet
high-quality RCTs that ensured allocation concealment, available. In a meta-analysis of observational studies,
adequate randomization and blinding found that daily compared with no endoscopy, screening colonoscopy
supplementation with calcium (1200–2000 mg) reduced was associated with reduced CRC risk (HR 0.31, 95% CI
colorectal adenoma recurrence by 12% (95% CI 0.79–0.99) 0.12–0.77) and mortality (HR 0.32, 95% CI 0.23–0.43)7.
over a period of 3–5 years200. Complications, including colonic perforations and intes
tinal bleeding, are relatively rare204. Over-diagnosis and
A role of calcium against colorectal neoplasms is sup
ported by several biological mechanisms. According to consequent over-treatment are not as common for CRC
experimental studies, calcium precipitates secondary as for breast and prostate cancer209.
bile acids, ionized fatty acids and haem iron in the Given several screening options with varying perfor
colorectal lumen, diminishing their carcinogenic effects mance (regarding sensitivity and specificity), cost and
on the colorectal mucosa138,201. When calcium binds complications, cost-effective analyses might inform
calcium-sensing receptors on intestinal epithelial cells,
the optimal screening strategy. Despite heterogeneity
diverse intracellular signalling pathways are activated, in underlying modelling assumptions and study popu
inhibiting proliferation and inducing differentiation andlations, studies have consistently found that screening,
apoptosis201,202. In a RCT conducted among individuals irrespective of the screening methods, is cost-effective
with a history of adenoma, calcium supplementation compared with no screening, with the costs per life-year
of 2000 mg per day for 6 months induced favourable gained less than US$60,000 across different countries210.
changes in the relative expression of APC and β-catenin However, studies disagree on the most cost-effective
in macroscopically normal rectal mucosa201,203, which screening strategy, which suggests that factors such
provides mechanistic evidence for an early action of as population acceptability, screening adherence and
calcium in the adenoma–carcinoma pathway. availability of resources for colonoscopy might be of
critical consideration in determining the optimal screen
Screening: secondary prevention ing modality to implement210. For instance, relative to
Screening can reduce CRC incidence and death by ena endoscopic screening tests, stool-based tests such as
bling removal of precancerous lesions before malignant guaiac-based FOBT or FIT are more acceptable to the
transformation or early detection and treatment of public for their lower cost and non-invasiveness, despite
CRC204. Several screening modalities are available, with requiring more frequent tests211. Therefore, stool-based
varying cost and infrastructure requirements. Widely tests might offer more attractive and practical screen
adopted globally, the faecal occult blood test (FOBT) ing options in many parts of the world. In the USA,
checks stool samples for small amounts of blood, which screening recommendations by professional organiza
can indicate the presence of precancerous lesions or tions do not emphasize a specific screening approach
cancer. Among FOBT, immunochemical FOBT (com but rather offer diverse screening strategies208,212, which
monly known as faecal immunochemical test or FIT) is help increase screening uptake213.
often preferred compared with the older guaiac-based CRC screening guidelines vary by countries and
FOBT. FIT does not require dietary restriction before professional organizations. For average-r isk adults,
sample collection and has a higher sensitivity and a many countries, such as Denmark, France and
lower incidence of interval CRC after a negative screen Norway, recommend FIT to those aged 50–74 years214.
ing test result (20 versus 34 cases per 100,000 person- Colonoscopy is not used as a screening tool except for in
years, P value not reported)205. An emerging screening a few countries (for example, Austria, Germany, Poland
method is multitarget stool DNA testing (FIT-DNA), and the USA)8,214. For older adults aged ≥75 years,
Reviews
a Italy Latvia Poland Spain
M M
50 50 50 50
30 F 30 M 30 30 F
ASR(W) per 100,000
20 20 F M 20 M M 20 M
15 M 15 15 F 15
F F
10 F 10 10 10 F
5 5 5 5
3 3 3 3
2 2 2 2
1960 1985 2010 1960 1985 2010 1960 1985 2010 1960 1985 2010
Year Year Year Year
b Austria Czech Republic Israel
M
50 50 50 M
30 M 30 30
F F
ASR(W) per 100,000
20 F 20 M 20
15 M 15 15
M
F
10 10 10 F
F
5 5 5
3 3 3
2 2 2
1960 1985 2010 1960 1985 2010 1960 1985 2010
Year Year Year
Japan USA: Black USA: White
50 M 50 M 50
30 30 F 30 M
Incidence of CRC
ASR(W) per 100,000
F
20 20 20 F
15 M 15 M 15 Mortality of CRC
10 10 F 10 M
F FIT
F
5 5 5
gFOBT
3 3 3
Colonoscopy
2 2 2
1960 1985 2010 1960 1985 2010 1960 1985 2010
Year Year Year
Fig. 8 | Introduction of colorectal cancer screening programmes and time trend of colorectal cancer incidence rates
and mortality rates. a | Stabilizing mortality rates after introduction of colorectal cancer (CRC) screening programmes.
In Italy, Latvia, Poland and Spain, for example, mortality rates became plateaued after the start of a CRC screening
programme, although CRC incidence rates continued to increase. b | Stabilizing or decreasing incidence rates and
decreasing mortality rates after introduction of CRC screening programmes. In Austria, the Czech Republic, Israel, Japan
and the USA, for example, in addition to decreasing mortality, CRC incidence rates stabilized or decreased with varying
time lags after screen programme implementation. Information on timing of screening programme was obtained from
Schreuders et al.8 ASR(W), age-standardized rate (cases per 100,000 persons per year) with use of age weights from the
world standard population; F, female sex; FIT, faecal immunochemical test for haemoglobin; gFOBT, guaiac faecal occult
blood test; M, male sex. Adapted by permission from BMJ Publishing Group Limited. Arnold, M. et al. Global patterns and
trends in colorectal cancer incidence and mortality. Gut 66, 683–691 (2017)12.
routine screening is recommended in Austria but recommend screening via FIT beginning as early as age
not in Canada; screening is decided on an individual 40 years214.
basis with consultation of their health-care providers Despite heterogeneous screening programmes world
and accounting for the overall health status and prior wide, introduction of a screening programme seems to
screening history in the UK and USA214. Some countries, be followed by a reduction in CRC mortality (Fig. 8a,b),
such as New Zealand and Finland, do not recommend with a caveat that causality cannot be inferred from eco
screening until age 59 years, whereas Austria and Japan logical observations. For Italy, Latvia, Poland and Spain,

Reviews
mortality rates plateaued after the start of a screening estimate that at least 40% of new CRC could be prevented
programme, although CRC incidence rates continued through healthy changes in these principal lifestyle fac
to increase (Fig. 8a). Of note, an increased detection of tors9. Smoking has decreased in some populations with
asymptomatic CRC by screening generally leads to an initially high levels of smoking, such as the Netherlands,
initial rise in CRC incidence and, therefore, time lag is Norway and Spain, but is increasing in many popula
expected between screening introduction and reduction tions18. While the motivated individual can follow life
in CRC incidence rates215. This aspect is particularly evi style guidelines for prevention, some of these changes
dent in Austria and Israel (Fig. 8b), in which a reduction (for example, increased opportunities for physical
in CRC incidence rates occurred later than a reduction in activity) can be initiated at the societal level (government,
mortality rates following implementation of a screen community, work environment and schools).
ing programme. On the contrary, in the Czech Republic Chemopreventive drugs could attenuate the increases
and Japan, there was a relatively immediate reduction in in CRC wrought by suboptimal diets and lifestyles. To
incidence rates after a screening programme started. date, the most promising efficacious agent is aspirin221.
In the USA, compared with black individuals, a decreas Yet, recommendation for its widespread prophylactic
ing trend in CRC incidence and deaths was observed in use is premature. Even low-dose aspirin increases the
white individuals before the initiation of mandated risk of gastrointestinal bleeding222 and at least 10 years
insurance coverage for CRC cancer screening in 1998 of use might be required to receive the chemopreven
(REF. 216) . The distinct pattern in white individuals tive benefit223. Currently, the US Preventive Services
might, in part, reflect a long-term effect of screen Task Force is the only expert body that recommends
ing, considering that white individuals, who generally routine low-dose aspirin use for the primary preven
have higher household income, better insurance status tion of CRC and cardiovascular diseases to a subgroup
and higher educational level, probably started opportun of adults aged 50–59 years with elevated cardiovascular
istic screenings earlier than 1998 when the mandated risk profiles but not at an increased risk of bleeding223.
insurance coverage began217. The ASPIRED trial, an ongoing trial of individuals
In the USA, the American Cancer Society reduced previously diagnosed with colorectal adenomas, aims
the starting age of regular CRC screening for average- to determine risk-stratification biomarkers to identify
risk adults from 50 to 45 years in 2018 in response to an individuals with favourable risk–benefit ratio for aspirin
increase in early-onset CRC in the past few decades212. chemoprevention against CRC224.
By contrast, the US Preventive Services Task Force Nutrient-based chemoprevention could also ame
maintains the starting age of 50 years, noting discordant liorate CRC risk safely. While no factor is considered
findings across microsimulation analyses for starting to have definitive evidence, the evidence for calcium
screening at age 45 years208. It remains to be examined supplementation has been deemed as ‘probable’ by the
whether the new guideline by the American Cancer WCRF–AICR9. The potential benefit might be pro
Society will substantially contribute to reducing mor nounced in populations that have developed a western
tality from early-onset CRC. Although relative rates have diet and lifestyle but whose intake of dairy products
increased, the absolute risk of early-onset CRC is still is relatively low, partly due to lactose intolerance. For
low (<1%) in the USA6. Empirical data from long-term instance, in Asian populations, inverse associations have
observational studies and RCTs on outcomes of screen been observed with CRC risk even within low ranges
ing before age 50 years are needed to better understand of calcium intake225–227. Besides calcium, there is some
the risk–benefit ratio for early screening and to iden evidence for chemopreventive effects of other micro
tify the optimal starting age among individuals at aver nutrients, including vitamin D and folate221. RCTs with
age risk. For individuals with predisposing conditions, long durations (greater than 10 years) might be required
in the USA, screening colonoscopy is recommended to show definitive evidence for efficacy as shown with
starting before age 50 years and more frequently (every aspirin188.
1–2 years for those with hereditary CRC syndromes218 Incidence and mortality for CRC can be greatly
and every 5 years for those with a family history of CRC reduced by screening, especially with colonoscopy7. The
or advanced adenomas in a first-degree relative aged main hurdles to CRC screening include cost, underuti
<60 years219) than every 10 years (recommendation for lization and optimizing surveillance schedules when a
average-risk individuals)208,212. serrated polyp is identified213,228. As of 2015, 32 countries,
primarily those with high income and high incidence
Future challenges and opportunities of CRC, had organized screening programmes8. CRC
Increased incidence and mortality of CRC inevita screening programmes might not be cost-efficient in
bly follow economic development and its sequelae of many low-income or medium-income countries with
westernization of diet and lifestyle. Reducing adiposity, low CRC incidence, in which other common diseases
increasing physical activity while decreasing sedentary might be of a public health priority. In these countries,
time, and improving diet in addition to reducing exces interventions to promote healthy lifestyle probably
sive alcohol consumption and smoking rates, are likely to confer greater benefit for overall health beyond the
have a substantial effect on reducing CRC incidence and reduction of CRC229. However, many low-income or
mortality166, with added benefits of reducing diabetes, medium-income countries undergoing rapid western
cardiovascular diseases and some other cancers220. Yet, ization are now experiencing escalating CRC incidence
thus far, obesity, low physical activity and poor diet have rates, and they might merit a CRC screening programme
been difficult to reverse in most populations, despite an based on FOBT, which is less expensive and non-invasive
Reviews
compared with colonoscopy210. Current recommen Conclusions

dations to start screening at age 50 years might miss a The optimal reduction of CRC incidence and mortal
proportion of early-onset cases. Targeted screening for ity will require concerted efforts to reduce modifiable
individuals at higher risk of CRC approaches should be risk factors, to leverage chemoprevention research and
considered. Currently, family history is primarily consid to promote population-wide and targeted screening.
ered for risk stratification208,212,219, but a study of two large The effort devoted to each approach needs to be bal
international consortia found that a model integrating anced with the overall health priorities of the specific
environmental and genetic (CRC-associated SNPs) population, taking economic resources and health-care
factors along with family history predicted CRC risk infrastructure into account.
and screening starting age with greater accuracy than a
model with family history alone230. Published online xx xx xxxx
1. Fleming, M., Ravula, S., Tatishchev, S. F. & Wang, H. L. 19. Clarke, N., Sharp, L., Osborne, A. & Kearney, P. M. molecular pathology. J. Mol. Diagn. 14, 91–103
Colorectal carcinoma: pathologic aspects. Comparison of uptake of colorectal cancer screening (2012).
J. Gastrointest. Oncol. 3, 153–173 (2012). based on fecal immunochemical testing (FIT) in males 39. Weisenberger, D. J. et al. CpG island methylator
2. Jasperson, K. W., Tuohy, T. M., Neklason, D. W. and females: a systematic review and meta-analysis. phenotype underlies sporadic microsatellite instability
& Burt, R. W. Hereditary and familial colon cancer. Cancer Epidemiol. Biomarkers Prev. 24, 39–47 and is tightly associated with BRAF mutation in
Gastroenterology 138, 2044–2058 (2010). (2015). colorectal cancer. Nat. Genet. 38, 787–793 (2006).
3. Graff, R. E. et al. Familial risk and heritability of 20. Murphy, N. et al. A prospective evaluation of 40. Conteduca, V., Sansonno, D., Russi, S. & Dammacco, F.
colorectal cancer in the nordic twin study of cancer. endogenous sex hormone levels and colorectal cancer Precancerous colorectal lesions (Review). Int. J. Oncol.
Clin. Gastroenterol. Hepatol. 15, 1256–1264 (2017). risk in postmenopausal women. J. Natl Cancer. Inst. 43, 973–984 (2013).
4. Lichtenstein, P. et al. Environmental and heritable 107, djv210 (2015). 41. Winawer, S. J. et al. Colorectal cancer screening:
factors in the causation of cancer—analyses of cohorts 21. U.S. Cancer Statistics Working Group. U.S. Cancer clinical guidelines and rationale. Gastroenterology
of twins from Sweden, Denmark, and Finland. N. Engl. Statistics Data Visualizations Tool, based on 112, 594–642 (1997).
J Med. 343, 78–85 (2000). November 2017 submission data (1999-2015): U.S. 42. Brenner, H. et al. Risk of progression of advanced
5. Jiao, S. et al. Estimating the heritability of colorectal Department of Health and Human Services, Centers adenomas to colorectal cancer by age and sex:
cancer. Hum. Mol. Genet. 23, 3898–3905 (2014). for Disease Control and Prevention and National estimates based on 840,149 screening colonoscopies.
6. International Agency for Research on Cancer. Cancer Institute. Centers for Disease Control and Gut 56, 1585–1589 (2007).
Globocan 2018: Cancer Fact Sheets — Colorectal Prevention www.cdc.gov/cancer/dataviz (2016). 43. Strum, W. B. Colorectal adenomas. N. Engl J. Med.
Cancer. IARC http://gco.iarc.fr/today/data/factsheets/ 22. Augustus, G. J. & Ellis, N. A. Colorectal cancer 374, 1065–1075 (2016).
cancers/10_8_9-Colorectum-fact-sheet.pdf (2018). disparity in african americans: risk factors and 44. East, J. E. et al. British Society of Gastroenterology
7. Brenner, H., Stock, C. & Hoffmeister, M. Effect of carcinogenic mechanisms. Am. J. Pathol. 188, position statement on serrated polyps in the colon
screening sigmoidoscopy and screening colonoscopy 291–303 (2018). and rectum. Gut 66, 1181–1196 (2017).
on colorectal cancer incidence and mortality: 23. Ashktorab, H. et al. Identification of novel mutations 45. Erichsen, R. et al. Increased risk of colorectal cancer
systematic review and meta-analysis of randomised by exome sequencing in African American colorectal development among patients with serrated polyps.
controlled trials and observational studies. BMJ 348, cancer patients. Cancer 121, 34–42 (2015). Gastroenterology 150, 895–902e5 (2016).
g2467 (2014). 24. Wang, H. et al. Novel colon cancer susceptibility 46. Dow, L. E. et al. Apc restoration promotes cellular
8. Schreuders, E. H. et al. Colorectal cancer screening: variants identified from a genome-wide association differentiation and reestablishes crypt homeostasis
a global overview of existing programmes. Gut 64, study in African Americans. Int. J. Cancer 140, in colorectal cancer. Cell 161, 1539–1552 (2015).
1637–1649 (2015). 2728–2733 (2017). 47. Armaghany, T., Wilson, J. D., Chu, Q. & Mills, G.
9. World Cancer Research Fund International/American 25. Huyghe, J. R. et al. Discovery of common and rare Genetic alterations in colorectal cancer. Gastrointest.
Institute for Cancer Research. Continuous update genetic risk variants for colorectal cancer. Nat. Genet. Cancer Res. 5, 19–27 (2012).
project report: Diet, nutrition, physical activity, 51, 76–87 (2019). 48. Pino, M. S. & Chung, D. C. The chromosomal
and colorectal cancer. American Institute for Cancer 26. Ashktorab, H. et al. A meta-analysis of MSI frequency instability pathway in colon cancer. Gastroenterology
Research https://www.aicr.org/continuous-update- and race in colorectal cancer. Oncotarget 7, 138, 2059–2072 (2010).
project/reports/colorectal-cancer-2017-report.pdf 34546–34557 (2016). 49. Leggett, B. & Whitehall, V. Role of the serrated
(2018). 27. Pawlik, T. M., Raut, C. P. & Rodriguez-Bigas, M. A. pathway in colorectal cancer pathogenesis.
10. Center, M. M., Jemal, A. & Ward, E. International Colorectal carcinogenesis: MSI-H versus MSI-L. Gastroenterology 138, 2088–2100 (2010).
trends in colorectal cancer incidence rates. Cancer Dis. Markers 20, 199–206 (2004). 50. Kedrin, D. & Gala, M. K. Genetics of the serrated
Epidemiol. Biomarkers Prev. 18, 1688–1694 (2009). 28. Guindalini, R. S. et al. Mutation spectrum and risk pathway to colorectal cancer. Clin. Transl
11. United Nations Development Programme. 2018 of colorectal cancer in African American families with Gastroenterol. 6, e84 (2015).
Statistical Update and Human Development Index. Lynch syndrome. Gastroenterology 149, 1446–1453 51. O’Brien, M. J. et al. Comparison of microsatellite
United Nations Development Programme http:// (2015). instability, CpG island methylation phenotype, BRAF
hdr.undp.org/ (2018). 29. Pitot, H. C. The molecular biology of carcinogenesis. and KRAS status in serrated polyps and traditional
12. Arnold, M. et al. Global patterns and trends in Cancer 72, 962–970 (1993). adenomas indicates separate pathways to distinct
colorectal cancer incidence and mortality. Gut 66, 30. Carethers, J. M. & Jung, B. H. Genetics and genetic colorectal carcinoma end points. Am. J. Surg. Pathol.
683–691 (2017). biomarkers in sporadic colorectal cancer. 30, 1491–1501 (2006).
13. Imamura, F. et al. Dietary quality among men and Gastroenterology 149, 1177–1190.e3 (2015). 52. Kim, K. M. et al. Molecular features of colorectal
women in 187 countries in 1990 and 2010: 31. Munro, M. J., Wickremesekera, S. K., Peng, L., Tan, S. T. hyperplastic polyps and sessile serrated adenoma/
a systematic assessment. Lancet Glob. Health 3, & Itinteang, T. Cancer stem cells in colorectal cancer: polyps from Korea. Am. J. Surg. Pathol. 35,
e132–e142 (2015). a review. J. Clin. Pathol. 71, 110–116 (2018). 1274–1286 (2011).
14. World Health Organization. Mean Body Mass Index 32. Ogino, S. & Goel, A. Molecular classification and 53. Yang, S., Farraye, F. A., Mack, C., Posnik, O. &
Trends Among Adults, Age-Standardized Estimates by correlates in colorectal cancer. J. Mol. Diagn. 10, O’Brien, M. J. BRAF and KRAS mutations in
Country. World Health Organization http://apps.who. 13–27 (2008). hyperplastic polyps and serrated adenomas of the
int/gho/data/node.main.A904?lang=en (2017). 33. Grady, W. M. & Carethers, J. M. Genomic and colorectum: relationship to histology and CpG island
15. Mousavi, S. M., Fallah, M., Sundquist, K. & epigenetic instability in colorectal cancer methylation status. Am. J. Surg. Pathol. 28,
Hemminki, K. Age- and time-dependent changes pathogenesis. Gastroenterology 135, 1079–1099 1452–1459 (2004).
in cancer incidence among immigrants to Sweden: (2008). 54. Jess, T., Rungoe, C. & Peyrin-Biroulet, L. Risk of
colorectal, lung, breast and prostate cancers. Int. J. 34. Bakhoum, S. F. et al. The mitotic origin of colorectal cancer in patients with ulcerative colitis:
Cancer 131, E122–E128 (2012). chromosomal instability. Curr. Biol. 24, R148–R149 a meta-analysis of population-based cohort studies.
16. Karastergiou, K., Smith, S. R., Greenberg, A. S. & (2014). Clin. Gastroenterol. Hepatol. 10, 639–645 (2012).
Fried, S. K. Sex differences in human adipose tissues 35. Nazemalhosseini Mojarad, E., Kuppen, P. J., 55. Itzkowitz, S. H. & Yio, X. Inflammation and cancer IV.
– the biology of pear shape. Biol. Sex Differ. 3, 13 Aghdaei, H. A. & Zali, M. R. The CpG island methylator Colorectal cancer in inflammatory bowel disease: the
(2012). phenotype (CIMP) in colorectal cancer. Gastroenterol. role of inflammation. Am. J. Physiol. Gastrointest.
17. Wilsnack, R. W., Wilsnack, S. C., Kristjanson, A. F., Hepatol. Bed Bench 6, 120–128 (2013). Liver Physiol. 287, G7–G17 (2004).
Vogeltanz-Holm, N. D. & Gmel, G. Gender and alcohol 36. Markowitz, S. D. & Bertagnolli, M. M. Molecular 56. Buchner, A. M. & Lichtenstein, G. R. Evaluation and
consumption: patterns from the multinational origins of cancer: molecular basis of colorectal cancer. detection of dysplasia in IBD: the role of
GENACIS project. Addiction 104, 1487–1500 N. Engl J. Med. 361, 2449–2460 (2009). chromoendoscopy and enhanced imaging techniques.
(2009). 37. Jia, M., Gao, X., Zhang, Y., Hoffmeister, M. & Brenner, H. Curr. Treat. Options Gastroenterol. 14, 73–82
18. World Health Organization. WHO Global Report on Different definitions of CpG island methylator (2016).
Trends in Prevalence of Tobacco Smoking. 2015. phenotype and outcomes of colorectal cancer: 57. Robles, A. I. et al. Whole-exome sequencing analyses
World Health Organization http://apps.who.int/iris/ a systematic review. Clin. Epigenetics 8, 25 (2016). of inflammatory bowel disease-associated colorectal
bitstream/handle/10665/156262/9789241564922_ 38. Funkhouser, W. K. Jr. et al. Relevance, pathogenesis, cancers. Gastroenterology 150, 931–943 (2016).
eng.pdf;jsessionid=3E603A72D553D128773D0836 and testing algorithm for mismatch repair-defective 58. Triantafillidis, J. K., Nasioulas, G. & Kosmidis, P. A.
AECE325E?sequence=1 (2015). colorectal carcinomas: a report of the association for Colorectal cancer and inflammatory bowel disease:

Reviews
epidemiology, risk factors, mechanisms of 87. Half, E., Bercovich, D. & Rozen, P. Familial 111. Wang, T. et al. Effects of obesity related genetic
carcinogenesis and prevention strategies. Anticancer adenomatous polyposis. Orphanet J. Rare Dis. 4, 22 variations on visceral and subcutaneous fat
Res. 29, 2727–2737 (2009). (2009). distribution in a chinese population. Sci. Rep. 6,
59. Li, F. Y. & Lai, M. D. Colorectal cancer, one entity or 88. Patel, S. G. & Ahnen, D. J. Familial colon cancer 20691 (2016).
three. J. Zhejiang Univ. Sci. B 10, 219–229 (2009). syndromes: an update of a rapidly evolving field. 112. Leeners, B., Geary, N., Tobler, P. N. & Asarian, L.
60. Murphy, N. et al. Heterogeneity of colorectal cancer Curr. Gastroenterol. Rep. 14, 428–438 (2012). Ovarian hormones and obesity. Hum. Reprod. Update
risk factors by anatomical subsite in 10 European 89. Murphy, C. C. & Singal, A. G. Establishing a research 23, 300–321 (2017).
countries: a multinational cohort study. Clin. agenda for early-onset colorectal cancer. PLOS Med. 113. Ning, Y., Wang, L. & Giovannucci, E. L. A quantitative
Gastroenterol. Hepatol. 17, 1323–1331.e6 (2019). 15, e1002577 (2018). analysis of body mass index and colorectal cancer:
61. Wei, E. K. et al. Comparison of risk factors for colon 90. Pearlman, R. et al. Prevalence and spectrum of findings from 56 observational studies. Obes. Rev. 11,
and rectal cancer. Int. J. Cancer 108, 433–442 germline cancer susceptibility gene mutations among 19–30 (2010).
(2004). patients with early-onset colorectal cancer. JAMA 114. Kim, H. & Giovannucci, E. L. Sex differences in the
62. Drewes, J. L., Housseau, F. & Sears, C. L. Sporadic Oncol. 3, 464–471 (2017). association of obesity and colorectal cancer risk.
colorectal cancer: microbial contributors to disease 91. Crosbie, A. B. et al. Trends in colorectal cancer Cancer Causes Control 28, 1–4 (2017).
prevention, development and therapy. Br. J. Cancer incidence among younger adults-disparities by age, 115. Thrift, A. P. et al. Mendelian randomization study of
115, 273–280 (2016). sex, race, ethnicity, and subsite. Cancer Med. 7, body mass index and colorectal cancer risk. Cancer
63. Yang, L. et al. Proximal shift of colorectal cancer with 4077–4086 (2018). Epidemiol. Biomarkers Prev. 24, 1024–1031 (2015).
increasing age in different ethnicities. Cancer Manag. 92. Ballester, V., Rashtak, S. & Boardman, L. Clinical and 116. World Cancer Research Fund/American Institute for
Res. 10, 2663–2673 (2018). molecular features of young-onset colorectal cancer. Cancer Research. World Cancer Research Fund
64. Murphy, G. et al. Sex disparities in colorectal cancer World J. Gastroenterol. 22, 1736–1744 (2016). Internationl Systematic Literature Review: The
incidence by anatomic subsite, race and age. Int. J. 93. Kim, T. J., Kim, E. R., Hong, S. N., Chang, D. K. Associations between Food, Nutrition and Physical
Cancer 128, 1668–1675 (2011). & Kim, Y. H. Long-term outcome and prognostic Activity and the Risk of Colorectal Cancer, 2017.
65. Shin, A. et al. Increasing trend of colorectal cancer factors of sporadic colorectal cancer in young patients: World Cancer Research Fund https://www.wcrf.org/
incidence in Korea, 1999-2009. Cancer Res. Treat. a large institutional-based retrospective study. sites/default/files/colorectal-cancer-slr.pdf (2017).
44, 219–226 (2012). Medicine 95, e3641 (2016). 117. Hetemaki, N. et al. Estrogen metabolism in abdominal
66. Missiaglia, E. et al. Distal and proximal colon cancers 94. Nimptsch, K. & Wu, K. Is timing important? the role subcutaneous and visceral adipose tissue in
differ in terms of molecular, pathological, and clinical of diet and lifestyle during early life on colorectal postmenopausal women. J. Clin. Endocrinol. Metab.
features. Ann. Oncol. 25, 1995–2001 (2014). neoplasia. Curr. Colorectal Cancer Rep. 14, 1–11 102, 4588–4595 (2017).
67. Cancer Genome Atlas Network. Comprehensive (2018). 118. Rezende, L. F. M. et al. Physical activity and cancer:
molecular characterization of human colon and rectal 95. NCD Risk Factor Collaboration (NCD-RisC). Worldwide an umbrella review of the literature including 22 major
cancer. Nature 487, 330–337 (2012). trends in body-mass index, underweight, overweight, anatomical sites and 770 000 cancer cases.
68. Guastadisegni, C., Colafranceschi, M., Ottini, L. and obesity from 1975 to 2016: a pooled analysis of Br. J. Sports Med. 52, 826–833 (2018).
& Dogliotti, E. Microsatellite instability as a marker 2416 population-based measurement studies in 119. Kushi, L. H. et al. American Cancer Society Guidelines
of prognosis and response to therapy: a meta-analysis of 128.9 million children, adolescents, and adults. on Nutrition and Physical Activity for Cancer
colorectal cancer survival data. Eur. J. Cancer 46, Lancet 390, 2627–2642 (2017). Prevention: reducing the risk of cancer with healthy
2788–2798 (2010). 96. Liu, P. H. et al. Association of obesity with risk of early- food choices and physical activity. CA Cancer J. Clin.
69. Boland, C. R. & Goel, A. Microsatellite instability in onset colorectal cancer among women. JAMA Oncol. 62, 30–67 (2012).
colorectal cancer. Gastroenterology 138, 5, 37–44 (2019). 120. Ainsworth, B. E. et al. 2011 Compendium of Physical
2073–2087.e3 (2010). 97. Renehan, A. G., Tyson, M., Egger, M., Heller, R. F. & Activities: a second update of codes and MET values.
70. Schumacher, T. N. & Schreiber, R. D. Neoantigens in Zwahlen, M. Body-mass index and incidence of cancer: Med. Sci. Sports Exerc. 43, 1575–1581 (2011).
cancer immunotherapy. Science 348, 69–74 (2015). a systematic review and meta-analysis of prospective 121. Ruiz-Casado, A. et al. Exercise and the hallmarks
71. Westdorp, H. et al. Opportunities for immunotherapy observational studies. Lancet 371, 569–578 (2008). of cancer. Trends Cancer 3, 423–441 (2017).
in microsatellite instable colorectal cancer. Cancer 98. Dong, Y. et al. Abdominal obesity and colorectal 122. Giovannucci, E. An integrative approach for
Immunol. Immunother. 65, 1249–1259 (2016). cancer risk: systematic review and meta-analysis of deciphering the causal associations of physical activity
72. Giannakis, M. et al. Genomic correlates of immune-cell prospective studies. Biosci. Rep. 37, BSR20170945 and cancer risk: the role of adiposity. J. Natl Cancer
infiltrates in colorectal carcinoma. Cell Rep. 15, (2017). Inst. 110, 935–941 (2018).
857–865 (2016). 99. Kune, G. A. The Melbourne Colorectal Cancer Study: 123. Ross, R. et al. Reduction in obesity and related
73. Passardi, A., Canale, M., Valgiusti, M. & Ulivi, P. reflections on a 30-year experience. Med. J. Aust. comorbid conditions after diet-induced weight loss or
Immune checkpoints as a target for colorectal cancer 193, 648–652 (2010). exercise-induced weight loss in men. A randomized,
treatment. Int. J. Mol. Sci. 18, E1324 (2017). 100. Mansournia, M. A. & Altman, D. G. Population controlled trial. Ann. Intern. Med. 133, 92–103
74. Jia, M. et al. No association of CpG island methylator attributable fraction. BMJ 360, k757 (2018). (2000).
phenotype and colorectal cancer survival: population- 101. Liang, P. S., Chen, T. Y. & Giovannucci, E. Cigarette 124. Ross, R. et al. Exercise-induced reduction in obesity
based study. Br. J. Cancer 115, 1359–1366 (2016). smoking and colorectal cancer incidence and mortality: and insulin resistance in women: a randomized
75. Jass, J. R. Classification of colorectal cancer based on systematic review and meta-analysis. Int. J. Cancer controlled trial. Obes. Res. 12, 789–798 (2004).
correlation of clinical, morphological and molecular 124, 2406–2415 (2009). 125. Ismail, I., Keating, S. E., Baker, M. K. & Johnson, N. A.
features. Histopathology 50, 113–130 (2007). 102. Song, M. et al. Long-term status and change of body A systematic review and meta-analysis of the effect of
76. Guinney, J. et al. The consensus molecular subtypes fat distribution, and risk of colorectal cancer: aerobic vs. resistance exercise training on visceral fat.
of colorectal cancer. Nat. Med. 21, 1350–1356 (2015). a prospective cohort study. Int. J. Epidemiol. 45, Obes. Rev. 13, 68–91 (2012).
77. Alwers, E. et al. Associations between molecular 871–883 (2016). 126. Keum, N. et al. Association of physical activity by type
classifications of colorectal cancer and patient survival: 103. Moore, L. L. et al. BMI and waist circumference as and intensity with digestive system cancer risk. JAMA
a systematic review. Clin. Gastroenterol. Hepatol. 17, predictors of lifetime colon cancer risk in Framingham Oncol. 2, 1146–1153 (2016).
402–410.e2 (2019). Study adults. Int. J. Obes. Relat. Metab. Disord. 28, 127. Schmid, D. & Leitzmann, M. F. Television viewing and
78. Butterworth, A. S., Higgins, J. P. & Pharoah, P. 559–567 (2004). time spent sedentary in relation to cancer risk: a meta-
Relative and absolute risk of colorectal cancer for 104. Samaras, K., Botelho, N. K., Chisholm, D. J. & Lord, R. V. analysis. J. Natl Cancer Inst. 106, dju098 (2014).
individuals with a family history: a meta-analysis. Subcutaneous and visceral adipose tissue gene 128. Ma, P., Yao, Y., Sun, W., Dai, S. & Zhou, C. Daily
Eur. J. Cancer 42, 216–227 (2006). expression of serum adipokines that predict type 2 sedentary time and its association with risk for
79. Weigl, K. et al. Strongly enhanced colorectal cancer diabetes. Obesity 18, 884–889 (2010). colorectal cancer in adults: a dose-response meta-
risk stratification by combining family history and 105. Bruun, J. M., Lihn, A. S., Pedersen, S. B. & analysis of prospective cohort studies. Medicine 96,
genetic risk score. Clin. Epidemiol. 10, 143–152 Richelsen, B. Monocyte chemoattractant protein-1 e7049 (2017).
(2018). release is higher in visceral than subcutaneous human 129. Lynch, B. M. Sedentary behavior and cancer:
80. Rahner, N. & Steinke, V. Hereditary cancer syndromes. adipose tissue (AT): implication of macrophages a systematic review of the literature and proposed
Dtsch. Arztebl. Int. 105, 706–714 (2008). resident in the AT. J. Clin. Endocrinol. Metab. 90, biological mechanisms. Cancer Epidemiol. Biomarkers
81. Mork, M. E. et al. High prevalence of hereditary 2282–2289 (2005). Prev. 19, 2691–2709 (2010).
cancer syndromes in adolescents and young adults 106. Ouchi, N., Parker, J. L., Lugus, J. J. & Walsh, K. 130. Healy, G. N. et al. Breaks in sedentary time: beneficial
with colorectal cancer. J. Clin. Oncol. 33, 3544–3549 Adipokines in inflammation and metabolic disease. associations with metabolic risk. Diabetes Care 31,
(2015). Nat. Rev. Immunol. 11, 85–97 (2011). 661–666 (2008).
82. Goodenberger, M. L. et al. PMS2 monoallelic mutation 107. Park, J., Morley, T. S., Kim, M., Clegg, D. J. 131. Cespedes, E. M. & Hu, F. B. Dietary patterns: from
carriers: the known unknown. Genet. Med. 18, 13–19 & Scherer, P. E. Obesity and cancer–mechanisms nutritional epidemiologic analysis to national
(2016). underlying tumour progression and recurrence. guidelines. Am. J. Clin. Nutr. 101, 899–900 (2015).
83. Win, A. K. et al. Prevalence and penetrance of Nat. Rev. Endocrinol. 10, 455–465 (2014). 132. Tabung, F. K., Brown, L. S. & Fung, T. T. Dietary
major genes and polygenes for colorectal cancer. 108. Calle, E. E. & Kaaks, R. Overweight, obesity and patterns and colorectal cancer risk: a review of
Cancer Epidemiol. Biomarkers Prev. 26, 404–412 cancer: epidemiological evidence and proposed 17 years of evidence (2000-2016). Curr. Colorectal
(2017). mechanisms. Nat. Rev. Cancer 4, 579–591 (2004). Cancer Rep. 13, 440–454 (2017).
84. Lynch, H. T. & de la Chapelle, A. Hereditary colorectal 109. Keum, N., Lee, D. H., Kim, R., Greenwood, D. C. & 133. Garcia-Larsen, V. et al. Dietary patterns derived from
cancer. N. Engl. J. Med. 348, 919–932 (2003). Giovannucci, E. L. Visceral adiposity and colorectal principal component analysis (PCA) and risk of
85. Lindgren, G., Liljegren, A., Jaramillo, E., Rubio, C. & adenomas: dose-response meta-analysis of colorectal cancer: a systematic review and meta-
Lindblom, A. Adenoma prevalence and cancer risk in observational studies. Ann. Oncol. 26, 1101–1109 analysis. Eur. J. Clin. Nutr. 73, 366–386 (2019).
familial non-polyposis colorectal cancer. Gut 50, (2015). 134. Vieira, A. R. et al. Foods and beverages and colorectal
228–234 (2002). 110. Lim, U. et al. Asian women have greater abdominal cancer risk: a systematic review and meta-analysis of
86. Jass, J. R. et al. Pathology of hereditary non-polyposis and visceral adiposity than Caucasian women with cohort studies, an update of the evidence of the
colorectal cancer. Anticancer Res. 14, 1631–1634 similar body mass index. Nutr. Diabetes 1, e6 WCRF-AICR Continuous Update Project. Ann. Oncol.
(1994). (2011). 28, 1788–1802 (2017).
Reviews
135. Cascella, M. et al. Dissecting the mechanisms and a public health perspective. J. Biomed. Sci. 24, 19 follow-up of five randomised trials. Lancet 376,
molecules underlying the potential carcinogenicity (2017). 1741–1750 (2010).
of red and processed meat in colorectal cancer (CRC): 160. Li, H. et al. Refined geographic distribution of the 186. Cook, N. R. et al. Low-dose aspirin in the primary
an overview on the current state of knowledge. oriental ALDH2*504Lys (nee 487Lys) variant. prevention of cancer the women’s health study:
Infect. Agent Cancer 13, 3 (2018). Ann. Hum. Genet. 73, 335–345 (2009). a randomized controlled trial. JAMA 294, 47–55
136. Helmus, D. S., Thompson, C. L., Zelenskiy, S., 161. Mizoue, T. et al. Alcohol drinking and colorectal cancer (2005).
Tucker, T. C. & Li, L. Red meat-derived heterocyclic in Japanese: a pooled analysis of results from five 187. Cook, N. R., Lee, I. M., Zhang, S. M. M., Moorthy, M. V.
amines increase risk of colon cancer: a population- cohort studies. Am. J. Epidemiol. 167, 1397–1406 & Buring, J. E. Alternate-day, low-dose aspirin and
based case-control study. Nutr. Cancer 65, (2008). cancer risk: long-term observational follow-up of a
1141–1150 (2013). 162. Zhao, H., Liu, K. J., Lei, Z. D., Lei, S. L. & Tian, Y. Q. randomized trial. Ann. Int. Med. 159, 77–85 (2013).
137. Bastide, N. M., Pierre, F. H. & Corpet, D. E. Heme iron Meta-analysis of the aldehyde dehydrogenases-2 188. Giovannucci, E. et al. Aspirin and the risk of
from meat and risk of colorectal cancer: a meta- (ALDH2) Glu487Lys polymorphism and colorectal colorectal-cancer in women. N. Engl. J. Med. 333,
analysis and a review of the mechanisms involved. cancer risk. PLOS ONE 9, e88656 (2014). 609–614 (1995).
Cancer Prev. Res. 4, 177–184 (2011). 163. Giovannucci, E. & Martinez, M. E. Tobacco, colorectal 189. Cole, B. F. et al. Aspirin for the chemoprevention
138. Gamage, S. M. K., Dissabandara, L., Lam, A. K. & cancer, and adenomas: a review of the evidence. of colorectal adenomas: meta-analysis of the
Gopalan, V. The role of heme iron molecules derived J. Natl Cancer Inst. 88, 1717–1730 (1996). randomized trials. J. Natl Cancer Inst. 101,
from red and processed meat in the pathogenesis of 164. Johnson, C. M. et al. Meta-analyses of colorectal 256–266 (2009).
colorectal carcinoma. Crit. Rev. Oncol. Hematol. 126, cancer risk factors. Cancer Causes Control 24, 190. Cao, Y. et al. Population-wide impact of long-term use
121–128 (2018). 1207–1222 (2013). of aspirin and the risk for cancer. JAMA Oncol. 2,
139. Fung, T. T. et al. A dietary pattern that is associated 165. Limsui, D. et al. Cigarette smoking and colorectal 762–769 (2016).
with C-peptide and risk of colorectal cancer in women. cancer risk by molecularly defined subtypes. J. Natl 191. Zelenay, S. et al. Cyclooxygenase-dependent tumor
Cancer Causes Control 23, 959–965 (2012). Cancer Inst. 102, 1012–1022 (2010). growth through evasion of immunity. Cell 162,
140. Tabung, F. K. et al. Association of dietary inflammatory 166. Carr, P. R. et al. Lifestyle factors and risk of sporadic 1257–1270 (2015).
potential with colorectal cancer risk in men and colorectal cancer by microsatellite instability status: 192. Chan, A. T., Ogino, S. & Fuchs, C. S. Aspirin and the
women. JAMA Oncol. 4, 366–373 (2018). a systematic review and meta-analyses. Ann. Oncol. risk of colorectal cancer in relation to the expression
141. Burkitt, D. P. Epidemiology of cancer of the colon and 29, 825–834 (2018). of COX-2. N. Engl. J. Med. 356, 2131–2142 (2007).
rectum. Cancer 28, 3–13 (1971). 167. Figueiredo, J. et al. Smoking-association risks of 193. Cao, Y. et al. Regular aspirin use associates with lower
142. Holscher, H. D. Dietary fiber and prebiotics and the conventional adenomas and serrated polyps in the risk of colorectal cancers with low numbers of tumor-
gastrointestinal microbiota. Gut Microbes 8, colorectum. Cancer Causes Control 26, 377–389 infiltrating lymphocytes. Gastroenterology 151,
172–184 (2017). (2015). 879–892.e4 (2016).
143. Zeng, H., Lazarova, D. L. & Bordonaro, M. 168. Zeilinger, S. et al. Tobacco smoking leads to extensive 194. Keum, N., Aune, D., Greenwood, D. C., Ju, W. &
Mechanisms linking dietary fiber, gut microbiota and genome-wide changes in DNA methylation. PLOS ONE Giovannucci, E. L. Calcium intake and colorectal
colon cancer prevention. World J. Gastrointest. Oncol. 8, e63812 (2013). cancer risk: dose-response meta-analysis of
6, 41–51 (2014). 169. Nishihara, R. et al. A prospective study of duration prospective observational studies. Int. J. Cancer 135,
144. Zeng, H., Taussig, D. P., Cheng, W. H., Johnson, L. K. & of smoking cessation and colorectal cancer risk by 1940–1948 (2014).
Hakkak, R. Butyrate inhibits cancerous HCT116 colon epigenetics-related tumor classification. Am. J. 195. Bristow, S. M. et al. Calcium supplements and cancer
cell proliferation but to a lesser extent in noncancerous Epidemiol. 178, 84–100 (2013). risk: a meta-analysis of randomised controlled trials.
NCM460 colon cells. Nutrients 9, E25 (2017). 170. Bae, J. M., Kim, J. H. & Kang, G. H. Molecular Br. J. Nutr. 110, 1384–1393 (2013).
145. Elce, A. et al. Butyrate modulating effects on pro- subtypes of colorectal cancer and their 196. Wactawski-Wende, J. et al. Calcium plus vitamin D
inflammatory pathways in human intestinal epithelial clinicopathologic features, with an emphasis on the supplementation and the risk of colorectal cancer.
cells. Benef. Microbes 8, 841–847 (2017). serrated neoplasia pathway. Arch. Pathol. Lab. Med. N. Engl. J. Med. 354, 684–696 (2006).
146. Shang, F. M. & Liu, H. L. Fusobacterium nucleatum 140, 406–412 (2016). 197. Lappe, J. et al. Effect of vitamin D and calcium
and colorectal cancer: a review. World J. Gastrointest. 171. Rogot, E. & Murray, J. L. Smoking and causes of death supplementation on cancer incidence in older women:
Oncol. 10, 71–81 (2018). among U.S. veterans: 16 years of observation. Public a randomized clinical trial. JAMA 317, 1234–1243
147. Mehta, R. S. et al. Association of dietary patterns with Health Rep. 95, 213–222 (1980). (2017).
risk of colorectal cancer subtypes classified by 172. Heineman, E. F., Zahm, S. H., McLaughlin, J. K. & 198. Zhang, X. et al. Calcium intake and colorectal cancer
fusobacterium nucleatum in tumor tissue. JAMA Vaught, J. B. Increased risk of colorectal cancer among risk: results from the Nurses’ Health Study and Health
Oncol. 3, 921–927 (2017). smokers: results of a 26-year follow-up of US veterans Professionals Follow-up Study. Int. J. Cancer 139,
148. Aune, D. et al. Dietary fibre, whole grains, and risk and a review. Int. J. Cancer 59, 728–738 (1994). 2232–2242 (2016).
of colorectal cancer: systematic review and dose- 173. Lin, J. H. et al. Association between sex hormones and 199. Keum, N., Lee, D. H., Greenwood, D. C., Zhang, X.
response meta-analysis of prospective studies. BMJ colorectal cancer risk in men and women. Clin. & Giovannucci, E. L. Calcium intake and colorectal
343, d6617 (2011). Gastroenterol. Hepatol. 11, 419–424.e1 (2013). adenoma risk: dose-response meta-analysis of
149. Kunzmann, A. T. et al. Dietary fiber intake and risk of 174. Freeman, E. W., Sammel, M. D., Lin, H. & Gracia, C. R. prospective observational studies. Int. J. Cancer 136,
colorectal cancer and incident and recurrent adenoma Obesity and reproductive hormone levels in the 1680–1687 (2015).
in the prostate, lung, colorectal, and ovarian cancer transition to menopause. Menopause 17, 718–726 200. Veettil, S. K. et al. Effects of calcium on the incidence
screening trial. Am. J. Clin. Nutr. 102, 881–890 (2010). of recurrent colorectal adenomas: a systematic review
(2015). 175. Kelly, D. M. & Jones, T. H. Testosterone and obesity. with meta-analysis and trial sequential analysis of
150. Song, M. et al. Fiber intake and survival after Obes. Rev. 16, 581–606 (2015). randomized controlled trials. Medicine 96, e7661
colorectal cancer diagnosis. JAMA Oncol. 4, 71–79 176. Hillman, E. T., Lu, H., Yao, T. & Nakatsu, C. H. (2017).
(2018). Microbial ecology along the gastrointestinal tract. 201. Lamprecht, S. A. & Lipkin, M. Chemoprevention
151. Alberts, D. S. et al. Lack of effect of a high-fiber cereal Microbes Environ. 32, 300–313 (2017). of colon cancer by calcium, vitamin D and folate:
supplement on the recurrence of colorectal adenomas. 177. Flemer, B. et al. Tumour-associated and non-tumour- molecular mechanisms. Nat. Rev. Cancer 3, 601–614
phoenix colon cancer prevention physicians’ network. associated microbiota in colorectal cancer. Gut 66, (2003).
N. Engl. J. Med. 342, 1156–1162 (2000). 633–643 (2017). 202. Tennakoon, S., Aggarwal, A. & Kallay, E. The
152. World Cancer Research Fund International/American 178. Zhang, Y. et al. Changes in gut microbiota and plasma calcium-sensing receptor and the hallmarks of cancer.
Institute for Cancer Research. Continuous update inflammatory factors across the stages of colorectal Biochim. Biophys. Acta 1863, 1398–1407 (2016).
project report: Food, nutrition, physical activity, and tumorigenesis: a case-control study. BMC Microbiol. 203. Ahearn, T. U., Shaukat, A., Flanders, W. D.,
the prevention of colorectal cancer. World Cancer 18, 92 (2018). Rutherford, R. E. & Bostick, R. M. A randomized
Research Fund https://www.wcrf.org/sites/default/files/ 179. O’Keefe, S. J. et al. Fat, fibre and cancer risk in African clinical trial of the effects of supplemental calcium
Colorectal-Cancer-2011-Report.pdf (2011). Americans and rural Africans. Nat. Commun. 6, 6342 and vitamin D3 on the APC/β-catenin pathway in the
153. International Agency for Research on Cancer. Agents (2015). normal mucosa of colorectal adenoma patients.
Classified by the IARC Monographs, volumes 1-122. 180. Allen, J. M. et al. Exercise alters gut microbiota Cancer Prev. Res. 5, 1247–1256 (2012).
International Agency for Research on Cancer https:// composition and function in lean and obese humans. 204. PDQ Screening and Prevention Editorial Board.
monographs.iarc.fr/agents-classified-by-the-iarc/ Med. Sci. Sports Exerc. 50, 747–757 (2018). Colorectal Cancer Screening (PDQ®): Health
(2018). 181. Louis, S., Tappu, R. M., Damms-Machado, A., Professional Version. PDQ Cancer Information
154. Choi, Y. J., Myung, S. K. & Lee, J. H. Light alcohol Huson, D. H. & Bischoff, S. C. Characterization of the Summaries. NIH.gov https://www.ncbi.nlm.nih.gov/
drinking and risk of cancer: a meta-analysis of cohort gut microbial community of obese patients following books/NBK65825/ (2012).
studies. Cancer Res. Treat. 50, 474–487 (2018). a weight-loss intervention using whole metagenome 205. Wieten, E. et al. Incidence of faecal occult blood test
155. Cho, E. et al. Alcohol intake and colorectal cancer: shotgun sequencing. PLOS ONE 11, e0149564 interval cancers in population-based colorectal cancer
a pooled analysis of 8 cohort studies. Ann. Intern. Med. (2016). screening: a systematic review and meta-analysis. Gut
140, 603–613 (2004). 182. Ley, R. E., Turnbaugh, P. J., Klein, S. & Gordon, J. I. 68, 873-881 (2019).
156. Salaspuro, M. Microbial metabolism of ethanol and Microbial ecology: human gut microbes associated 206. Imperiale, T. F. et al. Multitarget stool DNA testing for
acetaldehyde and clinical consequences. Addict. Biol. with obesity. Nature 444, 1022–1023 (2006). colorectal-cancer screening. N. Engl. J. Med. 370,
2, 35–46 (1997). 183. Coker, O. O. et al. Enteric fungal microbiota dysbiosis 1287–1297 (2014).
157. Seitz, H. K. & Stickel, F. Molecular mechanisms of and ecological alterations in colorectal cancer. Gut 68, 207. Brenner, H. & Chen, H. Fecal occult blood versus DNA
alcohol-mediated carcinogenesis. Nat. Rev. Cancer 7, 654–662 (2019). testing: indirect comparison in a colorectal cancer
599–612 (2007). 184. Nakatsu, G. et al. Alterations in enteric virome are screening population. Clin. Epidemiol. 9, 377–384
158. Giovannucci, E. Alcohol, one-carbon metabolism, and associated with colorectal cancer and survival (2017).
colorectal cancer: recent insights from molecular outcomes. Gastroenterology 155, 529–541.e5 208. US Preventive Services Task Force, Bibbins-Domingo,
studies. J. Nutr. 134, 2475S–2481S (2004). (2018). K. et al. Screening for colorectal cancer: us preventive
159. Chang, J. S., Hsiao, J. R. & Chen, C. H. ALDH2 185. Rothwell, P. M. et al. Long-term effect of aspirin on services task force recommendation statement. JAMA
polymorphism and alcohol-related cancers in Asians: colorectal cancer incidence and mortality: 20-year 315, 2564–2575 (2016).

Reviews
209. Welch, H. G. & Black, W. C. Overdiagnosis in cancer. syndromes. Am. J. Gastroenterol. 110, 223–262; 227. Ishihara, J., Inoue, M., Iwasaki, M., Sasazuki, S. &
J. Natl Cancer Inst. 102, 605–613 (2010). quiz 263 (2015). Tsugane, S. Dietary calcium, vitamin D, and the risk
210. Lansdorp-Vogelaar, I., Knudsen, A. B. & Brenner, H. 219. Rex, D. K. et al. Colorectal cancer screening: of colorectal cancer. Am. J. Clin. Nutr. 88, 1576–1583
Cost-effectiveness of colorectal cancer screening. recommendations for physicians and patients from (2008).
Epidemiol. Rev. 33, 88–100 (2011). the U.S. Multi-Society Task Force on Colorectal Cancer. 228. Jones, R. M., Devers, K. J., Kuzel, A. J. & Woolf, S. H.
211. Tinmouth, J., Lansdorp-Vogelaar, I. & Allison, J. E. Gastroenterology 153, 307–323 (2017). Patient-reported barriers to colorectal cancer screening:
Faecal immunochemical tests versus guaiac faecal 220. Veronese, N. et al. Combined associations of body a mixed-methods analysis. Am. J. Prev. Med. 38,
occult blood tests: what clinicians and colorectal weight and lifestyle factors with all cause and cause 508–516 (2010).
cancer screening programme organisers need to know. specific mortality in men and women: prospective 229. Li, Y. et al. Impact of healthy lifestyle factors on life
Gut 64, 1327–1337 (2015). cohort study. BMJ 355, i5855 (2016). expectancies in the US population. Circulation 138,
212. Wolf, A. M. D. et al. Colorectal cancer screening for 221. Dulai, P. S. et al. Chemoprevention of colorectal cancer 345–355 (2018).
average-risk adults: 2018 guideline update from the in individuals with previous colorectal neoplasia: 230. Jeon, J. et al. Determining risk of colorectal cancer
American Cancer Society. CA Cancer J. Clin. 68, systematic review and network meta-analysis. and starting age of screening based on lifestyle,
250–281 (2018). BMJ 355, i6188 (2016). environmental, and genetic factors. Gastroenterology
213. Inadomi, J. M. et al. Adherence to colorectal cancer 222. Cea Soriano, L., Lanas, A., Soriano-Gabarro, M. 154, 2152–2164.e19 (2018).
screening: a randomized clinical trial of competing & Garcia Rodriguez, L. A. Incidence of upper and lower
strategies. Arch. Intern. Med. 172, 575–582 gastrointestinal bleeding in new users of low-dose Acknowledgements
(2012). aspirin. Clin. Gastroenterol. Hepatol. 17, 887-895.e6 N.K. was supported by grants from the National Research
214. Ebell, M. H., Thai, T. N. & Royalty, K. J. Cancer (2019). Foundation of Korea (NRF-2018R1C1B6008822;
screening recommendations: an international 223. Chubak, J., Kamineni, A., Buist, D. S. M., NRF-2018R1A4A1022589).
comparison of high income countries. Public Health Anderson, M. L. & Whitlock, E. P. Aspirin Use for
Rev. 39, 7 (2018). the Prevention of Colorectal Cancer: An Updated Author contributions
215. Keum, N. & Giovannucci, E. L. Folic acid fortification Systematic Evidence Review for the U.S. Preventive All authors contributed equally to researching data for the
and colorectal cancer risk. Am. J. Prev. Med. 46, Services Task Force (Report No.: 15-05228-EF-1), article, discussion of content, and writing, reviewing and editing
S65–S72 (2014). NIH.gov https://www.ncbi.nlm.nih.gov/books/ of the manuscript before submission.
216. Cokkinides, V., Bandi, P., Shah, M., Virgo, K. & Ward, E. NBK321661/ (2015).
The association between state mandates of colorectal 224. Drew, D. A. et al. ASPirin Intervention for the Competing interests
cancer screening coverage and colorectal cancer REDuction of colorectal cancer risk (ASPIRED): a study The authors declare no competing interests.
screening utilization among US adults aged 50 to protocol for a randomized controlled trial. Trials 18,
64 years with health insurance. BMC Health Serv. Res. 50 (2017). Publisher’s note
11, 19 (2011). 225. Han, C. et al. Dietary calcium intake and the risk of Springer Nature remains neutral with regard to jurisdictional
217. James, T. M., Greiner, K. A., Ellerbeck, E. F., Feng, C. colorectal cancer: a case control study. BMC Cancer claims in published maps and institutional affiliations.
& Ahluwalia, J. S. Disparities in colorectal cancer 15, 966 (2015).
screening: a guideline-based analysis of adherence. 226. Shin, A. et al. Dietary intake of calcium, fiber and Reviewer information
Ethn. Dis. 16, 228–233 (2006). other micronutrients in relation to colorectal cancer Nature Reviews Gastroenterology & Hepatology thanks
218. Syngal, S. et al. ACG clinical guideline: genetic testing risk: results from the Shanghai Women’s Health Study. M. Hoffmeister, and the other, anonymous, reviewer(s) for
and management of hereditary gastrointestinal cancer Int. J. Cancer 119, 2938–2942 (2006). their contribution to the peer review of this work.

10.1038@s41575 019 0189 8

Uploaded by

Copyright:

Available Formats

10.1038@s41575 019 0189 8

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

10.1038@s41575 019 0189 8

Uploaded by

Copyright:

Available Formats

REvIEWS

Global burden of colorectal cancer:

Nature Reviews | Gastroenterology & Hepatology

c Cumulative risk of CRC

Nature Reviews | Gastroenterology & Hepatology

ASR(W) per 100,000

180 ensuing inactivation of TP53 tumour suppressor gene

Nature Reviews | Gastroenterology & Hepatology

Initiation Promotion Progression Metastasis

30–60 years 10–20 years 0–5 years

Nature Reviews | Gastroenterology & Hepatology

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

cases per 100,000

11 11 cases per 100,000

Nature Reviews | Gastroenterology & Hepatology

↑↑ 10 cm in WC 1.02 (1.01–1.03) 1.04 (1.02–1.06) 1.02 (1.00–1.03)

Nature Reviews | Gastroenterology & Hepatology

Nature Reviews | Gastroenterology & Hepatology

a Italy Latvia Poland Spain

b Austria Czech Republic Israel

Nature Reviews | Gastroenterology & Hepatology

compared with colonoscopy210. Current recommen­ Conclusions

Nature Reviews | Gastroenterology & Hepatology

Nature Reviews | Gastroenterology & Hepatology

You might also like

compared with colonoscopy210. Current recommen Conclusions