Our data provide first evidence that postmenopausal patients with higher levels of the estimated enterolignans, enterolactone and enterodiol, may have a significantly increased OS, which is independent of ER status of the tumour. Higher dietary fibre intake was also associated with a significantly better survival after breast cancer. In addition, the lower mortality associated with higher enterolignan levels in women with lower than median fibre intake provides some evidence for an effect of enterolignans, which is independent of fibre intake.
The estimation of enterolignan – the bioactive forms in the human body – using the method by Thompson et al (1991
), which is based on in vitro
fermentation of various plant foods with human faecal bacteria, has several strengths. It indirectly accounts for all enterolignan precursors and covers the long-term exposure to enterolignans. Nevertheless, the results must be interpreted cautiously as the estimation does not account for intra- and inter-individual variation of gut bacteria and metabolism. Also, the estimation is based on data collected in the early 1990s but, to our knowledge, there are no other data available to estimate the intestinally produced amounts of enterolactone and enterodiol, and this approach was also used by a previous study (Touillaud et al, 2007
Dietary intake of bread, vegetables, and fruits, which are important sources of lignans, was not individually associated with breast cancer prognosis. Intake of seeds, which are rich sources of lignans (Penalvo et al, 2005
), was associated with a nonsignificantly decreased overall mortality, but there was no significant dose-response relationship. Of note is that seeds contain other food constituents, like α
-linoleic acids (present in flaxseed), that have been shown to have anti-carcinogenic and anti-proliferative properties, which could render a protective effect (Wang et al, 2005
; Bozan and Temelli, 2008
So far, no other study has assessed the association between estimated enterolignans and breast cancer prognosis. Comparison of our findings with results on dietary lignan intake may not be appropriate. Nevertheless, in two smaller studies (each with ~800 postmenopausal women) conducted in the United States, higher dietary lignan intake was associated with lower mortality after breast cancer in one (McCann et al, 2009
) but not in the other (Fink et al, 2007
). Lignan intake was very different in these studies – one reporting a median intake of 245μ
(McCann et al, 2009
) and the other reporting 10 times higher intake levels (0 to >9
) (Fink et al, 2007
). In line with our results, a recent Danish study observed a significantly increased survival with higher plasma enterolactone levels (Olsen et al, 2011
Lignans have been found in high amounts in fibre-rich foods (e.g., cereals, wholegrains), and enterolactone concentrations were primarily explained by vegetable, rye, and fibre intake in Finnish men (Nurmi et al, 2010
). In our study, the primary sources of estimated enterolignans were intake of bread (white, brown, and wholegrain), vegetables, and fruits, which are also rich sources of fibre. Fibre intake could affect breast cancer risk by several mechanisms, including stimulation of the intestinal microflora, reduction of the enterohepatic oestrogen circulation, and thereby reduction of oestrogen concentrations in the body (Adlercreutz, 2007
; Sonestedt and Wirfält, 2010
). Additionally, increased fibre intake was found to be related to lower serum oestradiol levels in women with breast cancer (Rock et al, 2004
To account for potential confounding by fibre, we adjusted the multivariable regression models for estimated enterolignans and OS by fibre intake. High levels of estimated enterolignans were no longer significantly associated with overall mortality, probably because of their high correlation with fibre, which may have introduced collinearity in the multivariable models. Also, the association between dietary fibre and overall survival was no longer significant after adjustment for enterolactone. Adjustment for fibre intake was not found to affect the association of lignan intake with postmenopausal breast cancer risk in a study conducted in France (Touillaud et al, 2007
). However, the observation of a significant association of estimated enterolignans with a better prognosis in those with low dietary fibre intake strengthens the hypothesis that enterolignans may influence the prognosis of postmenopausal breast cancer independent of fibre intake. Furthermore, it has been suggested that high lignan intake might be an indicator of a healthy lifestyle (Sonestedt and Wirfält, 2010
). However, adjustment for lifestyle factors including physical activity, BMI, alcohol, and smoking did not affect the associations observed for enterolignans.
No significant effect heterogeneity by ER status of the tumours and menopausal hormone therapy use at diagnosis was observed for the association of estimated enterolignans and fibre with overall mortality, which is in line with the results of another study on dietary lignans and prognosis (Fink et al, 2007
). Therefore, hormone-dependent and -independent mechanisms of action might be involved in the observed association with enterolignans. Enterolignans are able to bind to ERs, thereby preventing endogenous oestrogens from binding and inhibiting tumour growth stimulation (Adlercreutz and Mazur, 1997
). In animal studies, several hormone-independent mechanisms have been suggested, such as inhibition of angiogenesis, tumour growth, and metastasis, as well as stimulation of apoptosis (Power et al, 2006
; Saarinen et al, 2008
). Moreover, an intervention with flaxseed in postmenopausal breast cancer patients resulted in an increased apoptosis and a decrease of tumour biological markers in tumour tissues (Thompson et al, 2005
The strength of our study was the large population-based patient sample with complete follow-up and verification of causes of death using death certificates as well as the consideration of all relevant prognostic factors. Sensitivity analyses were performed, which showed that results were not affected by time of diagnosis and FFQ completion. Additionally, fibre intake, enterolignan levels, and lignan-rich foods (except for the seeds) were estimated using a validated dietary questionnaire. Thereby, a broad spectrum of dietary lignan sources was addressed. The FFQ assessed dietary habits over the entire 12 months before diagnosis and, thus, captures the long-term situation of enterolignan exposure.
In the interpretation of our results, we also need to address some limitations. The assessment of dietary intake may be subject to measurement errors that are not only due to recall bias but also to the estimation using food-composition databases, which may not be complete for the whole range of foods consumed. Dietary changes after diagnosis, which could have contributed to survival, were not accounted for in this analysis. Although we collected comprehensive data and assessed many possible confounding factors, residual confounding could not be entirely ruled out. Also, sample size was limited in subgroups and therefore results must be interpreted cautiously.
In conclusion, we found some evidence for a better prognosis in postmenopausal breast cancer patients who have high estimated enterolignan and dietary fibre exposures. The associations of estimated enterolignans with survival were independent of ER status of the tumour and may, in part, be independent of dietary fibre intake. However, further investigations on enterolignans measured in serum or urine and breast cancer prognosis are warranted to confirm our results and to disentangle the possible mechanisms involved.