Professional Documents
Culture Documents
Patrones Dietarios y Cancer PDF
Patrones Dietarios y Cancer PDF
To cite this article: Francesca Bella, Justyna Godos, Antonella Ippolito, Alessia Di Prima &
Salvatore Sciacca (2016): Differences in the association between empirically derived dietary
patterns and cancer: a meta-analysis, International Journal of Food Sciences and Nutrition,
DOI: 10.1080/09637486.2016.1261087
RESEARCH ARTICLE
CONTACT Justyna Godos justyna.godos@student.uj.edu.pl Integrated Cancer Registry of Catania-Messina-Siracusa-Enna, Azienda Universitario
Ospedaliera Policlinico "Vittorio Emanuale", Catania, Italy
ß 2016 Informa UK Limited, trading as Taylor & Francis Group
2 F. BELLA ET AL.
empirically derived dietary patterns in relation with I2 statistic. The level of significance for the Q test was
cancer risk. defined as p < .10. The I2 statistic represented the
amount of total variation that could be attributed to
heterogeneity. I2 values >50 indicated high heterogen-
Methods
eity. Sensitivity analyzes by excluding one study at the
Study search time were conducted. When data were sufficient to
A systematic search on PubMed (http://www.ncbi.nlm. perform subgroup analyses, we considered whether
nih.gov/pubmed/) and EMBASE (http://www.embase. differences by gender and cancer localization existed.
com/) databases of English language studies published Publication bias was evaluated by a visual investigation
up to March 2016 was performed. Search terms of funnel plots for potential asymmetry. All analyzes
included a combination of keywords, such as “dietary were performed with Review Manager (RevMan) soft-
pattern” and “cancer” and their variants. Inclusion cri- ware version 5.2 (The Nordic Cochrane Center, The
teria were the following: (i) having an observational Cochrane Collaboration).
(case-control or prospective) study design; (ii) evaluat-
ing the association with risk of cancer between a pos- Results
teriori derived dietary patterns based on principal
Study characteristics
component analysis (PCA) or cluster analysis; (iii)
assessed and reported a measure of association for the Out of the 219 unique citations, 71 were excluded
outcome considered as odds ratios (ORs) or hazard after abstract examination and 141 were excluded after
ratios (HRs) and the corresponding 95% confidence full-text reading for the following reasons: 35 studies
intervals (CIs). Reference lists of included studies were explored a priori dietary patterns; four studies explored
scrutinized for any article not previously identified. cancer survival; and 102 studies compared the level of
The article selection process was independently per- adherence within dietary patterns (Figure 1). This selec-
formed by two authors. tion process led to the consideration of seven studies to
be included in this meta-analysis (Chen et al. 2002;
Rouillier et al. 2005; Austin et al. 2007; Edefonti et al.
Data extraction and study quality assessment
2009a; Wirfalt et al. 2009; Bastos et al. 2010; Magalhaes
Data were extracted from selected studies by using a et al. 2011). The main characteristics of the studies
standardized extraction form. The following informa- included are presented in Table 1: case–control studies
tion was collected: (i) author name; (ii) year of publi- (Chen et al. 2002; Rouillier et al. 2005; Austin et al.
cation; (iii) cohort name, length of follow-up, country; 2007; Edefonti et al. 2009a; Bastos et al. 2010;
(iv) number, gender, and age of participants and cases; Magalhaes et al. 2011) accounted for a total of 5042
(v) reference category and OR/HR and 95% CI for the cases and 5516 controls; the prospective study (Wirfalt
categories of exposure; and (vi) covariates used in et al. 2009) included 492,306 individuals and 3135 cases.
adjustments. Most of the studies were conducted on colorectal cancer
The Newcastle–Ottawa Quality Assessment Scale (Rouillier et al. 2005; Austin et al. 2007; Wirfalt et al.
was used to assess the quality of each study (Wells 2009; Magalhaes et al. 2011), two on gastric and esopha-
1999). The instrument consists of 3 domains indicat- geal cancers (Chen et al. 2002; Bastos et al. 2010), and
ing the study quality as follows: selection (4 points), one on breast and ovarian cancers (Edefonti et al.
comparability (2 points), and outcome (3 points) for a 2009a). All studies adjusted for variables potentially
total score of 9 points (9 representing the highest related with cancer incidence, including gender, body
quality). Studies scoring 7–9 points, 3–6 points and mass index, smoking status, physical activity, energy
0–3 points were identified as high, medium and low intake and hormonal and parity status in women. Study
quality, respectively. quality was high for most of the studies included in the
analyses.
Statistical analysis
Dietary patterns
Random-effects models were used to calculate risk
ratios (RRs) with 95% CIs for various categories of All studies used food frequency questionnaires to
exposure. The risk estimate from the most fully explore the dietary consumption of participants. In all
adjusted models in the analysis of the pooled RR was studies, dietary patterns were identified with PCA and
used. Heterogeneity was assessed using the Q test and cluster analysis: four studies compared dietary patterns
INTERNATIONAL JOURNAL OF FOOD SCIENCES AND NUTRITION 3
characterized by high meat intake with those high in heterogeneity was due to one study, after exclusion of
fruits and vegetables (Chen et al. 2002; Austin et al. which the summary risk estimates remained significant
2007; Bastos et al. 2010; Magalhaes et al. 2011); three with no evidence of heterogeneity (RR ¼1.88, 95% CI:
studies considered as category of reference dietary pat- 1.38, 2.58; I2 ¼ 0%, Pheterogeneity ¼ 0.59). A subgroup ana-
terns with no specific characterization or extreme con- lyzes based on cancer localization showed that the study
sumption of particular foods, which compared with conducted on upper digestive tract reported non-signifi-
patterns high in plant-foods and animal-foods cant summary risk estimates (RR ¼1.77, 95% CI: 0.59,
(Rouillier et al. 2005; Edefonti et al. 2009a; Wirfalt 5.33; I2 ¼ 76%, Pheterogeneity ¼ 0.02) while those conducted
et al. 2009) (Table 1). on colorectal cancer showed a significant risk related to
adoption of high-meat compared to plant-based dietary
patterns (RR ¼1.74, 95% CI: 1.25, 2.44; I2 ¼ 0%,
High-meat versus plant-based dietary patterns Pheterogeneity ¼ 0.55).
and cancer risk
The association of high-meat compared to plant-based
Plant-based versus mixed food patterns and
dietary patterns and cancer risk was explored in four
cancer risk
studies (Chen et al. 2002; Austin et al. 2007; Bastos
et al. 2010; Magalhaes et al. 2011) including 1794 indi- Seven datasets from three studies (Rouillier et al. 2005;
viduals and 1271 cases (Figure 2a): the analysis based Edefonti et al. 2009a; Wirfalt et al. 2009) were ana-
on six datasets showed an increased risk of cancer for lyzed to test the association plant-based compared to
adoption of the former compared with the latter pattern dietary patterns with various foods and cancer risk
(RR ¼1.64, 95% CI: 1.02, 2.63) with significant evidence (Figure 2b). The studies comprised the cohort study
of heterogeneity (I2 ¼ 69%, Pheterogeneity ¼ 0.007) and and 3771 total cases and 3722 controls from case–con-
asymmetry of funnel plot (Figure 3a). A sensitivity ana- trol studies on colon, rectum, breast and ovarian can-
lysis by exclusion of one study at the time revealed that cers. The analysis showed a significant lower risk of
quality
pattern over a mixed one (RR ¼0.88, 95% CI: 0.82,
Study 0.95) with no evidence of heterogeneity (I2 ¼ 6%,
High
Covariates adjustment Pheterogeneity ¼ 0.38) or asymmetry of funnel plot
energy. In women
at the time did not change summary results (data not
mone therapy.
shown). Subgroup analysis to explore potential con-
founding factors showed no variations of previous
findings: the results remained significant when consid-
ering women (RR ¼0.91, 95% CI: 0.83, 0.99; I2 ¼ 0%,
Pheterogeneity ¼ 0.54) and gastrointestinal cancer (RR
sweets, potatoes, wine and alcoholic
beverages, and low intake of whole-
but not female cancers (RR ¼0.89, 95% CI: 0.76, 1,05;
I2 ¼ 51%, Pheterogeneity ¼ 0.15).
analysis
analysis
Discussion
BMI: body mass index; FFQ: food frequency questionnaire; PCA: principal component analysis.
rectal cancer
colon cancer
colon cancer
No. of cases
women, 258
cases; in
and 707
men and
(293,576
women)
492,306
198,730
5-year follow-
follow-up
(country)
1995–2000,
and Health
Figure 2. Forest plots of studies evaluating summary risk ratios for cancer between (a) high-meat versus plant-based, (b) plant-
based versus mixed foods, and (c) meat-based dietary versus mixed foods dietary patterns.
olive oil and fish as main source of fats and low intake In the present study, unhealthy dietary patterns
of meat, and has been associated with decreased risk of characterized by red and, especially, processed meat,
cancer, including colorectal cancer (Grosso et al. 2014a; have been associated with increased risk of cancer.
Schwingshackl & Hoffmann 2015; D'Alessandro et al. Constituents of meat themselves, such as vitamin B,
2016). Plant-based dietary patterns are rich of several vitamin A, zinc and iron, as well as proteins provide
compounds that could play a role in the prevention of important nutritional benefits. In contrast, meat has a
cancer, for instance dietary fiber and antioxidants not negligible content in fat, including saturated fat
(Turati et al. 2015). Among antioxidants, vitamins, pol- and cholesterol, which may play a role in increasing
yphenols and omega-3 PUFA are the best candidates to the risk of certain cancers (Abid et al. 2014). Other
be responsible of anti-inflammatory effects, protective unhealthy factors have been hypothesized to be
action toward DNA damage (through prevention of responsible for the association between meat and can-
oxidation) and improving cell signaling and cell cycle cer, including (i) heterocyclic amines (HCAs) and
regulation, which are biological pathways related to polycyclic aromatic hydrocarbons (PAH) occurring
cancer initiation (Grosso et al. 2013a; Fantini et al. after cooking meat at high temperatures, and (ii) add-
2015). As most of the existing evidence on the role of ition of chemical preservatives such as sodium nitrite
nutrition and cancer risk is related to gastrointestinal (salt enriched with nitrates and nitrites) to increase its
cancers, dietary fiber may exert a key role through a durability, which leads to the formation of N-nitroso
direct action into the digestive tract by reducing transit compounds (NOC). From a mechanistic point of
time and contact of carcinogens with the colonic view, meat may exerts its effects (i) locally, with con-
mucosa, increasing binding of carcinogens and produc- tact with the lumen in the case of gastrointestinal can-
tion of short-chain fatty acids (Grosso et al. 2013b). cers; (ii) in target organs, through the hormone-like
INTERNATIONAL JOURNAL OF FOOD SCIENCES AND NUTRITION 7
Figure 3. Funnel plots of the analyses comparing high-meat versus plant-based, (b) plant-based versus mixed foods, and (c) meat-
based dietary versus mixed foods dietary patterns and risk of cancer.
properties of certain HCAs, which may differentially limited to, better nutrition behaviors have been asso-
affect organ normal functioning and homeostasis; and ciated with higher diet quality and lower metabolic
(iii) at cellular level, activating/deactivating enzymes issues (De Vriendt et al. 2009; McLeod et al. 2011;
across organs and tissues, which may either promote Grosso et al. 2013c; Jones & Zidenberg-Cherr 2015).
or suppress carcinogenesis (Abid et al. 2014). Indeed, healthy and unhealthy dietary habits are pri-
However, despite low meat intake can be associated marily associated with body weight, which is an
with lower cancer risk, there is no evidence on vege- independent risk factor for cancer risk (Grosso et al.
tarian diets that total depletion of animal sources of 2014b; Godos et al. 2016b; Rodriguez-Monforte et al.
protein from the diet lead to further decreased risk 2016). Moreover, healthy and unhealthy lifestyle
(Godos et al. 2016 (ahead of print)). choices tend to cluster, and thus, dietary choices
Results of the present meta-analysis should be con- may be associated with smoking habit and physical
sidered in light of some limitations. The main issue activity (Alvarez-Cubero et al. 2015). Overall, it is
difficult to isolate the effect of one factor from
relied on the observational nature of the studies
another.
included, as most of them had a case-control design,
In conclusion, plant-based dietary patterns can be
which may be affected by recall bias and confounding.
considered a healthy choice over meat-based dietary
Moreover, heterogeneity in dietary patterns could lead
patterns. Future studies will provide additional evi-
to study bias. Regarding the latter, a limitation of
dence and better investigate the potential causative
exploring dietary habits and quality is the potential roles.
confounding effect of the overall lifestyle and back-
ground of a person, including higher socioeconomic
status and education, which may lead to better access Disclosure statement
to cure and health-related information (Rametta et al. The authors report no conflicts of interest. The authors
2013). Among the latter, higher nutrition knowledge alone are responsible for the content and writing of this
and health conscious choices, including, but not article.
8 F. BELLA ET AL.
McLeod ER, Campbell KJ, Hesketh KD. 2011. Nutrition adenomacarcinoma sequence of colorectal cancer. Eur J
knowledge: a mediator between socioeconomic position Nutr. 44:311–318.
and diet quality in Australian first-time mothers. J Am Schwingshackl L, Hoffmann G. 2015. Adherence to
Diet Assoc. 111:696–704. Mediterranean diet and risk of cancer: an updated sys-
Mourouti N, Kontogianni MD, Papavagelis C, Panagiotakos tematic review and meta-analysis of observational studies.
DB. 2015. Diet and breast cancer: a systematic review. Int Cancer Med. 4:1933–1947.
J Food Sci Nutr. 66:1–42. Turati F, Rossi M, Pelucchi C, Levi F, La Vecchia C. 2015.
Ogino S, Lochhead P, Chan AT, Nishihara R, Cho E, Fruit and vegetables and cancer risk: a review of southern
Wolpin BM, Meyerhardt JA, Meissner A, Schernhammer European studies. Br J Nutr. 2:S102–S110.
ES, Fuchs CS, Giovannucci E. 2013. Molecular patho- Wells GASB, O’connell D, Peterson J, Welch V, Losos M,
logical epidemiology of epigenetics: emerging integrative Tugwell P. The Newcastle-Ottawa Scale (NOS) for assess-
science to analyze environment, host, and disease. Mod ing the quality of nonrandomised studies in meta-analy-
Pathol. 26:465–484. ses. Proceedings of the Ottawa (Canada): Ottawa Health
Rametta S, Grosso G, Galvano F, Mistretta A, Marventano S, Research Institute; 1999.
Nolfo F, Buscemi S, Gangi S, Basile F, Biondi A. 2013. Wirfalt E, Midthune D, Reedy J, Mitrou P, Flood A, Subar
Social disparities, health risk behaviors, and cancer. BMC AF, Leitzmann M, Mouw T, Hollenbeck AR, Schatzkin A,
Surg. 13:S17. Kipnis V. 2009. Associations between food patterns
Rodriguez-Monforte M, Sanchez E, Barrio F, Costa B, defined by cluster analysis and colorectal cancer incidence
Flores-Mateo G. 2016. Metabolic syndrome and dietary in the NIH-AARP diet and health study. Eur J Clin Nutr.
patterns: a systematic review and meta-analysis of obser- 63:707–717.
vational studies. Eur J Nutr. [Epub ahead of print]. doi: Wiseman M. 2008. The second World Cancer Research Fund/
10.1007/s00394-016-1305-y. American Institute for Cancer Research expert report.
Rouillier P, Senesse P, Cottet V, Valleau A, Faivre J, Food, nutrition, physical activity, and the prevention of
Boutron-Ruault MC. 2005. Dietary patterns and the cancer: a global perspective. Proc Nutr Soc. 67:253–256.