Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

The incidence of prostate cancer varies substantially between different countries, suggesting that there may be modifiable risk factors associated with lifestyle and environment. Currently, the only well-established risk factors for prostate cancer are age, family history, ethnic origin, family history and various genetic factors (Chan et al. 2005; Eeles et al. 2013). Over the last 30 years, many studies have investigated whether dietary and nutritional factors may influence risk, but the results of these studies have been generally inconsistent and inconclusive. Here, we summarise the results from a large prospective study in Europe, in which prostate cancer risk has been examined in relation to the intake of a range of foods and macronutrients, and to circulating biomarkers of nutritional factors. We also discuss the findings from EPIC and other studies on the relationship of prostate cancer risk with circulating concentrations of endogenous hormones.

2 Methods

The European Prospective Investigation into Cancer and Nutrition (EPIC) is a prospective study of approximately 500,000 men and women in ten European countries, recruited between 1992 and 2000. Full details of the study design have been published (Riboli 2002). Briefly, 150,000 of the participants recruited were men, living in eight countries: Denmark, Germany, Greece, Italy, the Netherlands, Spain, Sweden and the United Kingdom. At recruitment participants completed questionnaires on their diet and lifestyle, and most also provided blood samples from which serum, plasma, red blood cells and white blood cells were separated and stored at −80 °C or below. Participants have been followed to ascertain incident cancers and death; in most countries follow-up for cancer was through cancer registries, whereas for Germany and Greece follow-up for cancer was based on self-reported cancer diagnosis followed by confirmation through review of medical records.

For prostate cancer, analyses in EPIC have been planned to examine a range of hypotheses concerning the possible roles of nutritional, lifestyle and hormonal factors (Key et al. 2002). For potential risk factors assessed by questionnaire, such as dietary and lifestyle factors, analyses were conducted using Cox regression for the full cohort of men, with analyses stratified by recruitment centre (so that men who developed prostate cancer were compared with men recruited in the same place in Europe, who had not developed prostate cancer during the same follow-up period). For risk factors measured in the blood samples, such as nutritional biomarkers and endogenous hormones, analyses were conducted using a nested case–control design, such that for each case one control was selected matched on study centre, age at blood collection and follow-up period; laboratory analyses were conducted blind to case–control status and were planned so that, in general, cases and their matched controls were assayed in the same batch, thus largely eliminating inter-assay variation from the case–control comparisons. Statistical analysis of these nested case–control studies was by conditional logistic regression on matched sets, so that men who developed prostate cancer were compared with men recruited in the same place in Europe, who had not developed prostate cancer during the same follow-up period.

3 Results

Analyses of the relationships of intake of major food groups with prostate cancer risk in EPIC have been generally null (Table 1); the only significant association observed was a positive association with intake of yogurt, with an odds ratio of 1.17 (1.04–1.31) for men with a high intake of yogurt compared to men with a low intake of yogurt (Allen et al. 2008a). Sub-group analyses showed no significant differences in these associations by stage or grade of disease at diagnosis.

Table 1 Food groups and prostate cancer risk (Allen et al. 2008a; Key et al. 2004)
Full size table

For macronutrients, no association was observed between prostate cancer risk and intakes of total protein, fat, fibre or alcohol, but risk was higher in men with a high intake of dairy protein than in men with a low intake (odds ratio 1.22 (1.07–1.41); Allen et al. 2008a, Table 2). Sub-group analyses showed no significant differences in these associations by stage or grade of disease at diagnosis.

Table 2 Macronutrients and prostate cancer risk (Allen et al. 2008a; Crowe et al. 2008a; Suzuki et al. 2009; Rohrmann et al. 2008)
Full size table

For nutritional biomarkers, no association was observed between prostate cancer risk and blood levels of carotenoids, B vitamins, vitamin D, selenium or genistein (Table 3). There was a positive association between prostate cancer risk and the percentage of palmitic acid in plasma phospholipids and an inverse association between prostate cancer risk and the percentage of stearic acid in plasma phospholipids (Crowe et al. 2008b), but no association with the branched chain fatty acid phytanic acid (Price et al. 2010, Table 3) or other fatty acids (results not shown). Sub-group analyses showed some evidence of heterogeneity for the association of fatty acids with risk by grade of disease, such that high levels of myristic acid and α-linolenic acid were associated with a higher risk for high grade disease but not for low grade disease (Crowe et al. 2008b). There was evidence of heterogeneity for the association of lycopene with risk by stage of disease, such that high levels of lycopene were associated with a lower risk for advanced disease but not for localised disease (Key et al. 2007). There was also some evidence of heterogeneity by stage for vitamin B12, such that high levels of vitamin B12 were associated with a higher risk for advanced disease but not for localised disease (Johansson et al. 2008). There was no evidence of heterogeneity by stage or grade for the other nutritional biomarkers.

Table 3 Nutritional biomarkers and prostate cancer risk (Crowe et al. 2008b; Price et al. 2010; Key et al. 2007; Johansson et al. 2008; Travis et al. 2009; Allen et al. 2008b; Travis et al. 2012)
Full size table

For endogenous hormones, prostate cancer risk was not associated with serum concentrations of testosterone or free testosterone (Travis et al. 2007). Risk was higher in men with high levels of IGF-I than in men with low levels (odds ratio 1.69 (1.35–2.13); Fig. 1), with no evidence of heterogeneity by stage or grade of disease at diagnosis (Price et al. 2012), and prostate cancer risk was not associated with IGF binding protein 3 (Allen et al. 2007).

Fig. 1
figure 1

Insulin-like growth factor-I and risk of prostate cancer in EPIC (figure adapted from results in Price et al. 2012)

Full size image

4 Discussion

In EPIC, we have examined a range of nutritional and hormonal factors in relation to the risk for prostate cancer. The strongest association we have observed is with IGF-I, and this finding is consistent with a pooled analysis of individual participant data from 12 prospective studies (including EPIC; Roddam et al. 2008). IGF-I is affected by nutritional factors such as energy and protein intake (Ketelslegers et al. 1995) and there is evidence that men with high intakes of animal protein (or particularly dairy protein) have relatively high circulating IGF-I (Giovannucci et al. 2003; Young et al. 2012), but more research is needed to better understand the nature of the effect of nutrition on IGF-I and on whether this effect is likely to have a material impact on the risk of prostate cancer (Key 2011). In EPIC, we observed a weak positive association of dairy protein with prostate cancer risk, and other studies have reported positive associations of risk with dairy foods (Qin et al. 2007; WCRF 2007), but this topic requires further study before firm conclusions can be drawn. In EPIC, we also observed that men with the lactase genotype associated with lactase persistence have a higher intake of dairy products than men with the wild-type genotype for lactase (Travis et al. 2013), and the possibility that the lactase genotype is associated with prostate cancer risk should be examined in very large datasets.

We did not observe an association of endogenous sex hormones with prostate cancer risk, and this is consistent with the results of a pooled analysis of individual participant data from 18 prospective studies (Endogenous Hormones and Prostate cancer Collaborative Group 2008). These results indicate that androgen levels in the normal range are not associated with prostate cancer risk, but more research is needed to determine whether there may be a reduction in risk in men with particularly low androgen levels.

For the other nutritional factors examined in EPIC we have not found any strong associations. This is compatible with other studies worldwide, which have suggested that several nutritional factors may increase or decrease the risk for prostate cancer, but which have not established any definite effects (Chan et al. 2005; WCRF 2007). Further research is needed, because there may be real effects of moderate magnitude which could have major implications for disease prevention. Future studies need to collect data for very large numbers of men with prostate cancer and to collect detailed information on the characteristics of the tumour such as stage and grade, and on survival, to enable analyses of risk factors for aggressive prostate cancer (Wilson et al. 2012).