, Volume 44, Issue 5, p 318
Date: 30 Nov 2004

Dietary patterns and the adenomacarcinoma sequence of colorectal cancer

Rent the article at a discount

Rent now

* Final gross prices may vary according to local VAT.

Get Access

Summary

Background

Food components of a diet are highly related, so that building up dietary patterns may help understand the relationship between chronic diseases and diet, and identify high risk groups that need preventive advice.

Aim

The aim of this study was to determine dietary patterns associated with the colorectal adenoma–carcinoma pathway.

Methods

We performed a two–step analysis using first principal component analysis to select the most appropriate food groups, then a hierarchical agglomerative clustering method, in order to determine dietary patterns in 1372 subjects included in a case–control study. Patients with hyperplastic polyps (n = 103), adenomas < 10mm, (n = 154) or larger adenomas (n = 208) were then compared with polyp–free controls (n = 426), and colorectal cancer cases (n = 171) compared with population controls (n = 309) using unconditional logistic regression adjusted on age and gender.

Results

Cluster analysis determined five food patterns. Cluster 1 identified a low-energy diet; cluster 2 a high–starch, highfat, and low–fruit diet; cluster 3 a high–processed meat, –energy, –alcohol, and –starchy foods diet; cluster 4 a high–fish, –cereals, –honey, –olive oil, –fruit and –vegetables diet; and cluster 5 a high–flour, –sugar, –chocolate, –animal fats, and –eggs diet. Logistic regression identified cluster 2 as significantly associated with risk of small adenomas (OR = 1.7; 95% confidence interval 1.0–2.7), large adenomas (OR = 1.9; 1.2–3.0) and cancers (OR = 1.7; 1.1–2.8) compared with cluster 1. Cluster 4 diet was inversely associated with risk of small adenomas (OR = 0.4; 0.2–1.0). There was no relationship between patterns and risk of hyperplastic polyps. Multiple adjustment decreased the strength of the relationships with cluster 2, which remained significantly associated with adenomas, but not cancer.

Conclusion

A lowenergy diet appeared as protective all along the adenoma–carcinoma sequence, contrary to a high–energy, high–processed meat and –animal fat diet.

In the html abstract, html-text and pdf of the summary of the article
“Rouillier P, et al.(2004) Dietary patterns and the adenomacarcinoma sequence of colorectal cancer. Eur J Nutr (published online 20th August 2004)”The online version of the original article can be found athttp://dx.doi.org/10.1007/s00394-004-0525-8the numbering of the different clusters in the sub-section “Results” had been mixed up. Instead of:
“...Logistic regression identified cluster 1 as significantly associated with risk of small adenomas (OR=1.7; 95% confidence interval 1.0–2.7), large adenomas (OR=1.9; 1.2–3.0) and cancers (OR=1.7; 1.1–2.8) compared with cluster 2. Cluster 4 diet was inversely associated with risk of small adenomas (OR=0.4; 0.2–1.0). There was no relationship between patterns and risk of hyperplastic polyps. Multiple adjustment decreased the strength of the relationships with cluster 1, which remained significantly associated with adenomas, but not cancer.”
there should have been:
“Logistic regression identified cluster 2 as significantly associated with risk of small adenomas (OR=1.7; 95% confidence interval 1.0–2.7), large adenomas (OR=1.9; 1.2–3.0) and cancers (OR=1.7; 1.1–2.8) compared with cluster 1. Cluster 4 diet was inversely associated with risk of small adenomas (OR=0.4; 0.2–1.0). There was no relationship between patterns and risk of hyperplastic polyps. Multiple adjustment decreased the strength of the relationships with cluster 2, which remained significantly associated with adenomas, but not cancer.”
The publisher apologises for any inconvenience caused by this mistake.