Lifetime alcohol intake is associated with an increased risk of
KRAS+ and BRAF-/KRAS- but not BRAF+ colorectal cancer
Running title: Alcohol intake and molecular subtypes of colorectal cancer
Harindra Jayasekara
1,2
, Robert J. MacInnis
1,2
, Elizabeth J. Williamson
3,4
, Allison M.
Hodge
1
,
Mark Clendenning
5
, Christophe Rosty
5,6,7
, Rhiannon Walters
8
, Robin
Room
9,10,11
,
Melissa C. Southey
12
, Mark A. Jenkins
2
, Roger L. Milne
1,2
, John L.
Hopper
1,2
, Graham G. Giles
1,2
, Daniel D. Buchanan
2,5
and Dallas R. English
1,2
1
Cancer Epidemiology Centre, Cancer Council Victoria, 615 St Kilda Road, Melbourne,
Victoria 3004, Australia
2
Centre for Epidemiology and Biostatistics, Melbourne School of Population and Global
Health, The University of Melbourne, 207 Bouverie Street, Melbourne, Victoria 3010,
Australia
3
Farr Institute of Health Informatics Research, London, NW1 2DA, United Kingdom
4
Department of Medical Statistics, London School of Hygiene & Tropical Medicine, London,
WC1E 7HT, United Kingdom
5
Colorectal Oncogenomics Group, Genetic Epidemiology Laboratory, Department of
Pathology, The University of Melbourne, Parkville, Victoria, Australia
6
Envoi Specialist Pathologists, Brisbane, Queensland, Australia
7
School of Medicine, The University of Queensland, Brisbane, Queensland, Australia
8
Cancer and Population Studies Group, Queensland Institute of Medical Research, Herston,
Queensland, Australia
9
Centre for Alcohol Policy Research, La Trobe University, 215 Franklin Street, Melbourne,
Victoria 3000, Australia
10
Centre for Health Equity, Melbourne School of Population and Global Health, The
University of Melbourne, 207 Bouverie Street, Carlton, Victoria 3010, Australia
11
Centre for Social Research on Alcohol and Drugs, Stockholm University, SE - 106 91,
Stockholm, Sweden
12
Genetic Epidemiology Laboratory, Department of Pathology, The University of Melbourne,
Melbourne, Australia.
Correspondence to: Harindra Jayasekara, Cancer Epidemiology Centre, Cancer Council
Victoria, 615 St Kilda Road, Melbourne, Victoria 3004, Australia; Phone: +61 4 33469782;
Fax: +61 3 93495815; E-mail: harindra.jayasekara@cancervic.org.au
This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process which may lead to
differences between this version and the Version of Record. Please cite this article as an
‘Accepted Article’, doi: 10.1002/ijc.30568
This article is protected by copyright. All rights reserved.
2
Keywords: Alcohol intake; BRAF; colorectal cancer; KRAS
Abbreviations: CI, confidence interval; CIMP, CpG island methylator phenotype; HR,
hazard ratio; ICD-O-3, International Classification of Diseases for Oncology; MCCS,
Melbourne Collaborative Cohort Study; MSI, microsatellite instability; MSS, microsatellite
stable; VCR, Victorian Cancer Registry
Article category: Cancer Epidemiology
Abstract: 247 words Text: 3,316 words
Figures: 1 Tables: 3 References: 54
Novelty and Impact
Ethanol in alcoholic beverages has a causal association with colorectal cancer. Differences in
associations of alcohol intake with colorectal cancer subtypes defined by the presence of
somatic mutations in oncogenes BRAF and KRAS are not yet established. In the present study,
lifetime alcohol intake was associated with increased risks of KRAS+ and BRAF-/KRAS-
tumors (originating via specific molecular pathways including the traditional adenoma-
carcinoma pathway) but not with BRAF+ tumors, a hallmark of tumor development via the
‘serrated’ pathway.
Page 3 of 27
John Wiley & Sons, Inc.
International Journal of Cancer
This article is protected by copyright. All rights reserved.
3
Abstract
Ethanol in alcoholic beverages is a causative agent for colorectal cancer. Colorectal cancer is
a biologically heterogeneous disease, and molecular subtypes defined by the presence of
somatic mutations in BRAF and KRAS are known to exist. We examined associations
between lifetime alcohol intake and molecular and anatomic subtypes of colorectal cancer.
We calculated usual alcohol intake for 10-year periods from age 20 using recalled frequency
and quantity of beverage-specific consumption for 38,149 participants aged 40-69 years from
the Melbourne Collaborative Cohort Study. Cox regression was performed to derive hazard
ratios (HRs) and 95% confidence intervals (CIs) for the association between lifetime alcohol
intake and colorectal cancer risk. Heterogeneity in the HRs across subtypes of colorectal
cancer was assessed. A positive dose-dependent association between lifetime alcohol intake
and overall colorectal cancer risk (mean follow-up=14.6 years; n=596 colon and n=326 rectal
cancer) was observed (HR = 1.08, 95% CI: 1.04-1.12 per 10 g/day increment). The risk was
greater for rectal than colon cancer (p
homogeneity
=0.02). Alcohol intake was associated with
increased risks of KRAS+ (HR = 1.07, 95% CI: 1.00-1.15) and BRAF-/KRAS- (HR = 1.05,
95% CI: 1.00-1.11) but not BRAF+ tumors (HR = 0.89, 95% CI: 0.78-1.01; p
homogeneity
=0.01).
Alcohol intake is associated with an increased risk of KRAS+ and BRAF-/KRAS- tumors
originating via specific molecular pathways including the traditional adenoma-carcinoma
pathway but not with BRAF+ tumors originating via the serrated pathway. Therefore,
limiting alcohol intake from a young age might reduce colorectal cancer originating via the
traditional adenoma-carcinoma pathway.
Page 4 of 27
John Wiley & Sons, Inc.
International Journal of Cancer
This article is protected by copyright. All rights reserved.
4
Introduction
Ethanol in alcoholic beverages is a carcinogen
1
that increases the risk of colorectal cancer.
2
Although colorectal cancer is generally referred to as a single, broad disease entity, it is a
heterogeneous group of diseases in terms of molecular pathology and prognosis.
3, 4
A number
of molecularly defined subtypes of colorectal cancer have been described related to the
presence of key somatic events including microsatellite instability (MSI), the CpG island
methylator phenotype (CIMP), chromosomal instability, and somatic mutations in the
oncogenes BRAF, KRAS and PIK3CA.
5
For instance, colorectal cancers with BRAF mutation
are considered a distinct group
3, 6
while a combination of features sets KRAS-mutated
colorectal cancers apart from tumors harboring neither BRAF nor KRAS mutation.
7
Smoking has been consistently shown to have differences in associations with the risk of
specific molecular subtypes of colorectal cancer.
8-10
Findings for alcohol thus far have been
inconsistent: increased risks of MSI-low
8
and MSI-high
11, 12
colorectal cancer as well as an
absence of a difference in association with MSI
13, 14
or BRAF and CIMP
15, 16
subtypes have
been reported; associations for KRAS or combined BRAF/KRAS subtypes are not available.
Similarly, uncertainty remains whether alcohol consumption poses a greater risk for rectal
cancer over colon cancer: mechanistically, this is plausible considering that the rectal mucosa
is exposed to a greater carcinogenic effect of acetaldehyde due to its higher concentration.
17
In the present study, we examined the associations between lifetime alcohol intake and
colorectal cancer risk, overall and by subtypes defined by BRAF V600E and KRAS codons 12
and 13 somatic mutation status, and anatomic location (colon versus rectal), using data from
the Melbourne Collaborative Cohort Study (MCCS).
Page 5 of 27
John Wiley & Sons, Inc.
International Journal of Cancer
This article is protected by copyright. All rights reserved.
5
Materials and Methods
Study population
The MCCS is a prospective cohort study of 41,514 people (99.2% aged 40-69 years; 58.9%
women) recruited during 1990-94 from Melbourne.
18
Participants were recruited through the
electoral rolls (registration to vote is compulsory for adults in Australia), advertisements and
community announcements in local media (such as television, radio, newspapers).
Participants attended clinics where demographic, anthropometric, lifestyle and dietary
information were collected and anthropometric measurements were performed. Participants
aged <40 (n=194) and 70+ years (n=131) at baseline, with a confirmed cancer diagnosis
before baseline (n=1,467), missing alcohol consumption data for any age period (n=22),
reporting implausibly high alcohol intake (n=616) or extreme values of total energy intake
(<1
st
percentile and >99
th
percentile) (n=779), and missing data on any of the covariates
modelled (n=156) were excluded, leaving 38,149 (91.9% of all participants) eligible for this
analysis (Fig. 1). The study protocol was approved by the Cancer Council Victoria’s Human
Research Ethics Committee. Participants gave written informed consent to participate and for
investigators to obtain access to their medical records.
Baseline data collection
A structured interview schedule was used at baseline to obtain information on potential risk
factors including age, sex, country of birth, education, smoking habits, physical activity, and
previous medical conditions. A 121-item food frequency questionnaire was used to collect
dietary information.
19
Waist circumference was measured using a standard protocol. Baseline
residential addresses were used to classify participants into quintiles of an area-based
measure of socioeconomic status.
20
Page 6 of 27
John Wiley & Sons, Inc.
International Journal of Cancer
This article is protected by copyright. All rights reserved.