Cancer Incidence in Europe: An Ecological Analysis of Nutritional and Other Environmental Factors

Abstract

The aim of this work was to offer an ecological alternative to conventional observational studies and identify factors potentially associated with cancer incidence in Europe. The incidence of 24 types of cancer in 39 European countries (2012) was compared with a long-term mean supply of 68 food items from the FAOSTAT database (1993-2011) and some other variables such as smoking, body mass index, raised cholesterol, and socioeconomic indicators. In addition to simple Pearson linear correlations, the data were analyzed via factor analyses and penalized regression methods. This comparison identified two main groups of cancers that are characteristically associated with the same variables. The first group consists of cancers of the prostate, breast, white blood cells, and melanoma. Their incidence increases with rising gross domestic product (GDP) per capita, a prevalence of raised cholesterol and a high intake of animal products. The second group includes primarily cancers of the digestive tract and is most consistently correlated with alcoholic beverages, lard, and eggs. In addition, we found specific correlations between certain variables and some other types of cancer (smoking-lung and larynx cancer; low GDP per capita and high carbohydrate consumption-stomach and cervical cancer; tea drinking-esophageal cancer; maize consumption and wine drinking-liver cancer). The documented findings often remarkably agree with the current scientific consensus, and when combined with evidence based on different methodologies, they can further extend our knowledge of the etiology of cancer. In addition, our study also identifies several foods with possible preventive effects and indicates that various dairy products may markedly differ in their relationship to cancer incidence. All these data can potentially be of fundamental importance for clinical practice and the survival of cancer patients.

Keywords: Europe; cancer; epidemiology; food consumption; nutrition; raised cholesterol; smoking.

Figure 1

(A–F) Relationship between major independent variables and cancers of the prostate, breast, and white blood cells (non-Hodgkin lymphoma).

Figure 2

(A–F) Relationship between major independent variables and colorectal cancer.

Figure 3

(A–D) Relationship between the prevalence of smoking, and lung and larynx cancer.

Figure 4

(A–D) Relationship between major independent variables and stomach cancer.

Figure 5

(A–D) Relationship between some types of cancer and highly correlated independent variables.

Figure 6

(A–F) Relationship between major independent variables and total cancer incidence.

Figure 7

Factor analysis: a plot of Factor 1 and Factor 2 explaining 49.5% variability. For better clarity, only the main indicators of cancer incidence were selected (103 variables total). Abbreviations: NHL, non-Hodgkin lymphoma; % PC CARB energy, the proportion of energy from potato and cereal carbohydrates (as % of total energy intake); % CA energy, the proportion of energy from carbohydrates and alcoholic beverages (as % of total energy intake); Protein index, the ratio between milk and wheat proteins (an indicator of a high dietary protein quality).

Figure 8

Projection of 39 countries on the factor plane of Figure 7.

Figure 9

Factor analysis: a plot of Factor 1 and Factor 3 explaining 44.2% variability. For better clarity, only the main indicators of cancer incidence were selected (103 variables total). Abbreviations: NHL, non-Hodgkin lymphoma; % PC CARB energy, the proportion of energy from potato and cereal carbohydrates (as % of total energy intake); % CA energy, the proportion of energy from carbohydrates and alcoholic beverages (as % of total energy intake); Protein index, the ratio between milk and wheat proteins (an indicator of a high dietary protein quality).

Figure 10

Projection of 39 countries on the factor plane of Figure 9.

Figure 11

(A,B) Relationship between various dairy products (milk, cheese) and total cancer incidence.

Figure 12

(A–C) Relationship between various variables associated with dairy consumption and the incidence of prostate cancer. Note: “Protein index” is an indicator of protein quality, and expresses the mean ratio between dairy and wheat proteins.