Evaluation of a modified food frequency questionnaire to measure lignans in Australian men and women

Phytoestrogens are plant compounds that possess estrogenic and biological properties that have been postulated to protect against chronic diseases. Isoflavonoids and lignans are two main classes of phytoestrogen that have been investigated for their estrogenic efficacy and occurrence in the human diet. Isoflavonoids are found in soy and related products, whereas lignans are found in a wider range of plant-based foods, such as cereals, vegetables, fruits, legumes, nuts and seeds; and in beverages such as tea, coffee and wine. In Western populations with low dietary intake of soy products, compared to the Asian counterparts, lignans could be a more important and consistent source of phytoestrogens from the diet. Data for the phytoestrogen content in foods are now available, as more recent research has been conducted to quantify content in commonly consumed food choices in Western populations. The collation of these published values has led to the desire to adequately assess lignan intake. The aims of the research were to evaluate the validity and reliability of a phytoestrogen food frequency questionnaire (FFQ) tool, with a reference method, the weighed food record (WFR), and urinary biomarkers, to measure phytoestrogen intake in the Australian context. The phytoestrogen FFQ was updated and refined to align with food groups and dietary patterns in the current Australian Dietary Guidelines, in particular to optimise measurement of lignans from the FFQ, and utilise current databases of lignan content available from direct measurement of lignans in foods. Intake level and contributing food sources of each class of phytoestrogen, and the associations between social and lifestyle characteristics and phytoestrogen intake and urinary biomarker were also explored. This was a cross-sectional study that recruited 59 Australian men and women aged 18 to 67 years at Joondalup campus, Edith Cowan University. Intake of lignans, isoflavonoids and enterolignans from foods was assessed using the 277-item phytoestrogen FFQ and 3-day WFR, and excretion was assessed with urinary biomarkers. Published values of phytoestrogen content in foods were utilised to measure the intakes. Subjects collected three 24-hour urine samples and phytoestrogen concentration was analysed using a liquid chromatography-mass spectrometry (LC-MS) technique for four lignan subclasses, five isoflavonoids and two enterolignans. Statistical analyses were conducted using IBM SPSS for Windows (SPSS Inc., Version 22 Chicago, IL). Median intake comparisons were assessed with the Wilcoxon signed rank test. Associations between the intake and excretion measurements of two dietary assessment methods were assessed using Spearman’s Rho correlations. Level of agreement between methods was assessed with cross-classification analysis and Bland Altman plots. A triangular comparison between the three methods was conducted with the Method of Triads (MOT) using the software R. The Mann-Whitney U test and Kruskal-Wallis one-way ANOVA were used to compare the median intakes and excretion across categories of social and lifestyle factors. The FFQ had acceptable convergent validity for intake of total lignans and enterolignans when compared to a WFR, in terms of median intakes (lignans: 3914 versus 4302 μg/day, p=0.09; enterolignans: 54 versus 65 μg/day, p=0.81, respectively); and associations between the two methods (lignans ρ=0.42, p Top contributing food sources of lignans were from the nuts and seeds group (30%), nonalcoholic beverages (19%), and breads and cereals (19%); for enterolignans, from dairy products (86%) followed by nonalcoholic beverages (11%). Soy and related food products were the major contributors (78%) to total isoflavones, followed by breads and cereal products (17%). Female subjects who were Caucasian, were at, or had achieved university education level and took regular commercial dietary supplements, were more likely to have a higher lignan and enterolignan intake and excretion level than subjects with different characteristics. Based on these findings, we conclude that the modified phytoestrogen FFQ is highly reliable. It would be a useful assessment tool for example to rank usual intake of phytoestrogen classes for individuals within a group, or quantify mean intakes between different population groups. It is not acceptably valid or accurate for estimation of individual phytoestrogen status, for example for use in experimental studies or to investigate associations with chronic diseases. The lack of associations between measurement of the FFQ and biomarkers could partly be due to limitations of the FFQ tool, such as recall bias or inaccuracies in the estimation of frequency of intake or portion sizes. They also suggest that urinary biomarkers alone are not sufficient for estimation of phytoestrogen status and that additional biomarkers obtained from faecal and plasma samples should be considered for a more complete picture of phytoestrogen status