Processed Meat and Colorectal Cancer Risk: A Pooled Analysis of Three Italian Case-Control

Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=hnuc20
Download by: [University of Warwick] Date: 21 April 2017, At: 05:28
Nutrition and Cancer
ISSN: 0163-5581 (Print) 1532-7914 (Online) Journal homepage: http://www.tandfonline.com/loi/hnuc20
Processed Meat and Colorectal Cancer Risk: A
Pooled Analysis of Three Italian Case-Control
Studies
Valentina Rosato, Alessandra Tavani, Eva Negri, Diego Serraino, Maurizio
Montella, Adriano Decarli, Carlo La Vecchia & Monica Ferraroni
To cite this article: Valentina Rosato, Alessandra Tavani, Eva Negri, Diego Serraino, Maurizio
Montella, Adriano Decarli, Carlo La Vecchia & Monica Ferraroni (2017): Processed Meat and
Colorectal Cancer Risk: A Pooled Analysis of Three Italian Case-Control Studies, Nutrition and
Cancer, DOI: 10.1080/01635581.2017.1310259
To link to this article: http://dx.doi.org/10.1080/01635581.2017.1310259
Published online: 20 Apr 2017.
Submit your article to this journal
View related articles
View Crossmark data

Processed Meat and Colorectal Cancer Risk: A Pooled Analysis of Three Italian
Case-Control Studies
Valentina Rosatoa , Alessandra Tavanib , Eva Negri c
, Diego Serraino d , Maurizio Montellae
, Adriano Decarlia ,f ,
Carlo La Vecchia a
, and Monica Ferraronia
a Branch of Medical Statistics, Biometry and Epidemiology “G.A. Maccacaro”, Department of Clinical Sciences and Community Health, University
of Milan, Milan, Italy; b Milan, Italy; c Department of Biomedical and Clinical Sciences, University of Milan, Milan, Italy; d Cancer Epidemiology Unit,
CRO Aviano National Cancer Institute IRCCS, Aviano, Italy; e Unit of Epidemiology, National Cancer Institute, G. Pascale Foundation, Naples, Italy;
f Unit of Medical Statistics, Biometry and Bioinformatics, National Cancer Institute, IRCCS Foundation, Milan, Italy
ARTICLE HISTORY
Received 7 June 2016
Accepted 9 March 2017
ABSTRACT
To add evidence to the limited data available from southern Europe, we assessed the association
between processed meat consumption and colorectal cancer risk. We analyzed data from three
case-control studies conducted between 1985 and 2010 in various Italian areas, including a total of
3745 incident cases and 6804 hospital-based controls. We calculated odds ratios (ORs) and 95%
confidence intervals (CIs) by unconditional multiple logistic regression models. The median
consumption of processed meat was around 20 g/day both in cases and controls. The OR of
colorectal cancer was 1.02 (95% CI 0.99–1.04) for an increase of 10 g/day of processed meat. The
association was statistically significant for colon cancer (OR 1.03, 95% CI 1.00–1.06), particularly for
proximal colon cancer (OR 1.09, 95% CI 1.04–1.14), while there was no relation with rectal cancer
(OR 0.99, 95% CI 0.95–1.03). The OR of proximal colon cancer was 1.38 (95% CI 1.08–1.75) for the
highest sex-specific tertile of consumption (>25 g/day for men, >21.5 for women) compared with
the lowest (<15 g/day), whereas no significant ORs were found for other anatomical subsites. Our
findings indicate that there is no association with colorectal cancer overall, in the presence,
however, of a positive association with proximal colon cancer.
Introduction
Processed meat denotes meat transformed by salting,
curing, fermentation, smoking, or other processes aimed
to enhancing flavor or improving preservation. Proc-
essed meat includes, among others, ham, bacon, salami,
sausages, and hot dogs (1). Processed meat, compared to
unprocessed meat, generally contains more fat and more
additives due to long-time storage.
The International Agency for Research on Cancer
(IARC) classified processed meat as carcinogenic to
humans based on sufficient evidence that its con-
sumption causes colorectal cancer (1). The IARC
report quoted a meta-analysis published in 2011,
which showed a 18% excess risk of colorectal cancer
for an increase of 50 g/day of processed meat con-
sumption (2). Of the 9 cohort studies included in
that meta-analysis, 4 were conducted in the USA, 1
in Australia, and the remaining 4 in Europe. How-
ever, of the European studies, only the European Pro-
spective Investigation into Cancer and Nutrition
(EPIC) cohort included a minority of centers from
southern Europe (2,3), where type and quantity of
processed meat consumption and, more in general,
diet differ from those of USA/northern Europe.
Meat is a source of proteins, minerals, and vita-
mins, but it is also rich in fats, salt, and some chemi-
cal substances, including n-nitroso compounds,
polycyclic-aromatic hydrocarbons, and heterocyclic
aromatic-amines developed through meat processing,
preservation, and cooking, which have been hypothe-
sized to be associated with increased risk of cancer
and other diseases (4,5). However, the quality of
processed meat varies across countries and, in partic-
ular, processed meat produced and consumed in Italy
is characterized by lower contents of fats, salt, and
nitrites/nitrates (6).
To add evidence to the limited data available from
southern Europe, we assessed the association between
processed meat consumption and colorectal cancer risk,
using pooled data from three Italian case-control studies.
CONTACT Monica Ferraroni [e-mail] Laboratorio di Statistica Medica, Biometria ed Epidemiologia “G.A. Maccacaro”, Dipartimento di
Scienze Cliniche e di Comunita, Universita degli Studi di Milano, Campus Cascina Rosa – Via A. Vanzetti 5, 20133 Milano, Italy.
© 2017 Taylor & Francis Group, LLC
NUTRITION AND CANCER
2017, VOL. 0, NO. 0, 1–7
http://dx.doi.org/10.1080/01635581.2017.1310259

Material and Methods
Study Population
Between 1985 and 2010, we conducted three case-control
studies on risk factors for colorectal cancer in various Ital-
ian areas, using similar study protocols. The first study was
conducted between 1985 and 1991 in the urban area of
Milan (northern Italy) and included 1,326 colorectal cancer
cases (median age 62, range 20–74 yr) and 2,081 controls
(median age 55, range 19–74 yr) (7); the second was con-
ducted between 1992 and 1996 in the urban areas of Milan
and Genoa, the provinces of Pordenone, Gorizia, Forlı
(northern Italy), Latina (central Italy), and the urban area
of Naples (southern Italy), and included 1,953 colorectal
cancer cases (median age 62, range 19–74 yr) and 4,154
controls (median age 58, range 19–74 yr) (8); the third one
was conducted between 2008 and 2010 in the greater Milan
and the provinces of Pordenone and Udine (northern
Italy), and included 466 colorectal cancer cases (median
age 67, range 35–80 yr) and 569 controls (median age 66,
range 31–80 yr) (9). Thus, overall, 3,745 colorectal cancer
cases and 6,804 controls were enrolled. Cases were incident,
histologically confirmed colorectal cancer patients, admit-
ted to major teaching and general hospitals of the study
areas. Overall, 488 cases had a diagnosis of proximal colon
cancer (i.e., appendix, caecum, ascending colon, hepatic
flexure, and transverse colon, ICD-10 C18.0–C18.4); 1,078
of distal colon cancer (i.e., splenic flexure, descending, and
sigmoid colon, ICD-10 C18.5–C18.7); 788 of overlapping
or not otherwise specified colon cancer (ICD-10 C18.8–
C18.9); 1,383 had a diagnosis of rectal cancer (i.e., rectum
and rectosigmoid junction, ICD-10 C19.9–C20.9); and the
remaining 8 subjects had a not otherwise specified colorec-
tal cancer. In the three studies, controls were subjects
admitted to the same hospitals as cases for a wide spectrum
of acute, nonneoplastic conditions unrelated to factors
likely related to colorectal cancer. Overall, 28.0% of the con-
trol subjects were admitted for traumatic conditions, 24.1%
for other orthopedic disorders, 23.2% for acute surgical
conditions, 12.4% for eye diseases, and 12.3% for miscella-
neous other illnesses, including nose, ear, skin, or dental
disorder.
Ad-hoc-trained personnel interviewed by face to face all
subjects during their hospital stay by using structured ques-
tionnaires that included sociodemographic factors, lifestyle
habits (e.g., tobacco smoking, and alcohol drinking),
anthropometric measures, physical activity, a problem-ori-
ented medical history, and family history of cancer. Infor-
mation on patients’ usual diet before diagnosis/interview
was assessed through similar food frequency question-
naires, collecting information on weekly consumption of 29
selected food items in the first study (10), 78 foods, food
groups, or recipes in the second study, and 56 food items in
the most recent one (11). Occasional intake (lower than
once a week, but at least once a month) was coded as 0.5/
wk. Total processed meat was calculated as the sum of three
food items (cured ham, boiled ham, and salami/sausages)
for the first study, three food items (cured ham/bresaola/
speck, boiled ham, salami/mortadella/w€urstel/bacon/sau-
sages) for the second study, and one food item (all proc-
essed meats) for the third study. Processed meat
consumption was expressed in grams per day (g/day), by
dividing weekly frequency of consumption—an average
serving in the Italian diet, i.e., 50 g—by 7. The question-
naires of the three studies were tested for reproducibility
(10–12), and the second study was also tested for validity
(13), giving satisfactory results. Intake of nutrients and total
energy was computed using ad-hoc-developed food compo-
sition databases (14,15).
The study protocols were approved by the ethical review
boards of the participating centers, according to the regula-
tions at the time of each study conduction, and for the
third, most recent study, all subjects signed an informed
consent before recruitment. On average, less than 5% of
the cases and controls approached refused to participate.
Statistical Analysis
Odds ratios (ORs) for an increase of 10 g/day of proc-
essed meat consumption, and the corresponding 95%
confidence intervals (CIs), were estimated by uncondi-
tional multiple logistic regression models (16). Processed
meat consumption was also categorized in approximate
sex-specific tertiles of consumption among all controls
(corresponding to <15 g/day for both men and women,
in the first tertile; 15–25 g/day for men and 15–21.5 for
women, in the second tertile; and >25 g/day for men
and >21.5 g/day for women, in the third tertile). The
models included terms for age (quinquennia), sex, cen-
ter, study period, year of interview, education (<7, 7–11,
12 yr), tobacco smoking (never, former since at least
1 yr, current <20 cigarettes/day, current 20 cigarettes/
day), alcohol drinking (<1, 1–<4, 4 drinks/day), body
mass index (<25, 25–<30, 30 kg/m 2 ), occupational
physical activity (low, medium, high activity level, except
for Study I), vegetables (study-specific tertiles among
controls), fruit (study-specific tertiles among controls),
total energy intake (study-specific quintiles among con-
trols), and family history of intestinal cancer in first-
degree relatives. Missing values for adjustment covariates
were imputed as median value and included in the mod-
els. Test for linear trend was based on the likelihood-
ratio test between the models with and without a linear
term identifying exposure category. Stratified analyses
were carried out according to selected variables, and het-
erogeneity across strata was tested by computing the
2 V. ROSATO ET AL.