Validity of FFQ Estimates of Total Sugars, Added Sugars, Sucrose and Fructose Compared to Repeated 24-h Recalls in Adventist Health Study-2 Participants

Sugar intake is a potentially important aspect of diet which has not previously been validated in the Adventist Health Study-2 (AHS-2). We sought to validate the food frequency questionnaire (FFQ) measurement of total sugars, added sugars, sucrose, and fructose against multiple 24-h dietary recalls (recalls) in AHS-2 participants. Food consumption data from a self-administered FFQ and six recalls from 904 participants were combined with nutrient profile data to estimate daily sugar intake. Validity was evaluated among all participants and by race. FFQ and recall means were compared and correlation coefficients (Spearman’s, energy-adjusted log-transformed Pearson’s, deattenuated Pearson’s) were calculated. Mean total energy, total sugars, and fructose intake were higher in the FFQ, whereas added sugars and sucrose were higher in recalls. The energy-adjusted (log-transformed) deattenuated correlations among all participants were: total sugars (r = 0.42, 95% CI 0.32–0.52), added sugars (r = 0.50, 95% CI 0.36–0.59), sucrose (r = 0.32, 95% CI 0.23–0.42), and fructose (r = 0.50, 95% CI 0.40–0.59). We observed moderate validity for added sugars and fructose and low-moderate validity for total sugars and sucrose measured by the AHS-2 FFQ in this population. Dietary sugar estimates from this FFQ may be useful in assessing possible associations of sugars intake with health outcomes.

Validation of estimated glycaemic index and glycaemic load, stratified by race, in the Adventist Health Study-2 (AHS-2)

Objective: Few studies have validated FFQ estimates of dietary glycaemic index (GI) and load (GL). We investigated how well our estimates of overall GI and GL from FFQs correlate with estimates from repeated 24 h recall data to validate overall GI and GL in the Adventist Health Study-2 (AHS-2). Design: The AHS-2 is a prospective population-based cohort of 95 873 Seventh-day Adventist adult church members enrolled from 2002 to 2007 to investigate diet, cancer and mortality. Setting: A 204-item FFQ was used to assess race- and gender-specific validity of GI and GL and 24 h recall data, from the calibration sub-study, were used as the reference. Participants: The 734 calibration study participants were randomly selected by church and included approximately equal numbers of blacks and whites but were otherwise similar to the whole cohort with respect to gender, age, education and vegetarian status. Results: The deattenuated correlation coefficients for overall GI ranged from 0·19 (95 % CI −0·06, 0·53) in black men to 0·46 (95 % CI 0·40, 0·60) in black women, with both non-black men and women falling between those values (0·45 (95 % CI 0·35, 0·65) and 0·38 (95 % CI 0·27, 0·57), respectively). GL correlations were somewhat higher for all study participants. When looking at the entire cohort, the deattenuated validity correlation value for overall GI was (r 0·38, 95 % CI 0·36, 0·47) and GL was (r 0·39, 95 % CI 0·34, 0·49). Conclusions: Our findings support the cautious use of our FFQ in epidemiological studies when assessing associations of overall GI and GL with disease risk. However, observed differences by race should be considered when interpreting results.

Environmental Impacts of Foods in the Adventist Health Study-2 Dietary Questionnaire

The objective of this study was to use life cycle assessment to estimate the environmental impacts (from farm to factory gate) of the 198 hard-coded line-items included in the food frequency questionnaire of the Adventist Health Study-2 survey and to assess differences among food groups. Life cycle inventories were created using existing data sources and primary data, and their global warming potential (GWP), land use, and water consumption impacts were assessed using the ReCiPe 2016 methodology. In addition to presenting the impacts according to weight and protein content across food groups, we include the novel addition of presenting impacts according to the NOVA classification indicating various levels of processing. Food categories were compared based on one kilogram of edible food, protein food sources were compared based on one kilogram of protein, and NOVA comparisons were based on one serving. In general, meats had the highest environmental impacts per both weight and protein content, while the lowest overall impacts per kilogram came from fruits. Meat analogs had the lowest overall impacts per kilogram of protein, contrary to expectations that additional processing would result in higher environmental impacts when compared to whole plant-based foods. Per serving, ultra-processed foods had the highest GWP, processed foods the highest land use, and minimally processed foods the highest water consumption. Results from this analysis were consistent with other studies. Results from this study suggest that meat and ultra-processed foods have the overall worst environmental impacts, but high water consumption in some minimally processed foods means that those should be carefully considered as well.

DNA Methylation Profiles of Vegans and Non-Vegetarians in the Adventist Health Study-2 Cohort

We sought to determine if DNA methylation patterns differed between vegans and non-vegetarians in the Adventist Health Study-2 cohort. Genome-wide DNA methylation derived from buffy coat was profiled in 62 vegans and 142 non-vegetarians. Using linear regression, methylation of CpG sites and genes was categorized or summarized according to various genic/intergenic regions and CpG island-related regions, as well as the promoter. Methylation of genes was measured as the average methylation of available CpG’s annotated to the nominated region of the respective gene. A permutation method defining the null distribution adapted from Storey et al. was used to adjust for false discovery. Differences in methylation of several CpG sites and genes were detected at a false discovery rate < 0.05 in region-specific and overall analyses. A vegan diet was associated predominantly with hypomethylation of genes, most notably methyltransferase-like 1 (METTL1). Although a limited number of differentially methylated features were detected in the current study, the false discovery method revealed that a much larger proportion of differentially methylated genes and sites exist, and could be detected with a larger sample size. Our findings suggest modest differences in DNA methylation in vegans and non-vegetarians, with a much greater number of detectable significant differences expected with a larger sample.

The Association Between Phytosterol Intake and Breast Cancer Incidence in the Adventist Health Study-2 Cohort

Objectives Breast cancer is the most diagnosed form of cancer among American women. Worldwide, it is second only to lung cancer. Phytosterols are phytochemicals found in plant foods that have potential benefits for breast cancer. Research on phytosterols and cancer associations to date has been limited to breast cancer cell lines and animal studies, and the results have been promising. Our objective is to examine the association between breast cancer incidence and phytosterol intake in the Adventist Health Study-2, a large cohort in North America. Methods The present study estimated the association between phytosterol intake and breast cancer incidence in 52,734 females who were part of the Adventist Health Study 2 (AHS-2) cohort. Breast cancer cases (n = 1050) were ascertained with tumor registries from 2008 to 2014. Phytosterols content in foods was quantified by using USDA 17 and other published sources. These values were used to estimate phytosterol intake from food intake assessed by a self-administered food frequency questionnaires (FFQ). Results Hazard ratios were below the null, but statistically non-significant for β-sitosterol [HR = 0.77, 95%CI (0.44–1.36)], campesterol [HR = 0.84, 95%CI (0.46–1.55)], stigmasterol [HR = 0.76 (0.46–1.26)], and total phytosterol [HR = 0.77, 95%CI (0.43–1.40)]. In premenopausal women, HRs ranged between 0.95–1.72; in postmenopausal women, HRs were below the null, ranging between 0.67–0.83. In both premenopausal and postmenopausal women, HRs were statistically non-significant. Conclusions The inverse association between phytosterol consumption and breast cancer incidence appears uncertain. The uncertainty possibly could be due to lack of power or measurement error. Additional epidemiological studies with a larger number of breast cancer cases, improved phytosterol intake estimates, or both are needed. Funding Sources Unilever Research &Development, Vlaardingen, The Netherlands.

Associations of Circulating Methylmalonic Acid and Vitamin B-12 Biomarkers Are Modified by Vegan Dietary Pattern in Adult and Elderly Participants of the Adventist Health Study 2 Calibration Study

ABSTRACT Background Elevated plasma methylmalonic acid (MMA) is a functional biomarker of vitamin B-12 status but limited information is available on its prevalence in US vegetarians. Objectives The study examines the prevalence of plasma MMA ≥0.27 µmol/L in those consuming vegetarian diets, its associations with vitamin B-12 intake and biomarkers, and the modifying effect of vegetarian patterns on these associations. Methods In this cross-sectional study (n = 785), concentrations of MMA, vitamin B-12, holotranscobalamin (holoTC), and homocysteine (Hcy) were determined in participants of the calibration substudy of the Adventist Health Study 2 (AHS-2). Vitamin B-12 intake from food, fortified food, and supplements was assessed by six 24-h recalls. Regression models were used to estimate ORs of having high MMA as related to vitamin B-12 status biomarkers, vitamin B-12 intake, and dietary pattern. Results The prevalence of low vitamin B-12 status defined by serum vitamin B-12 &lt;148 pmol/L, holoTC &lt;35 pmol/L, MMA ≥0.27 and ≥0.37 μmol/L, or Hcy ≥15 μmol/L, and the OR of having high MMA did not differ by dietary pattern, possibly due to intake from fortified food and supplements. Total daily vitamin B-12 intake in the second tertile range of 4.4–14.5 μg/d reduced the likelihood of elevated MMA by 69%; and a doubling of vitamin B-12 intake was associated with a 4.3% decrease in plasma MMA. The association between log plasma MMA and biomarkers was modified by diet, with the vegan pattern showing an ∼3-fold stronger association with log serum vitamin B-12 and Hcy than did the nonvegetarian pattern. Conclusions The prevalence of vitamin B-12 intake &lt;2.0 μg/d was 15.2% in vegans, 10.6% in lacto-ovo-vegetarians, and 6.5% in nonvegetarians. Given the irreversible neurological consequences of vitamin B-12 inadequacy, the importance of regular supplemental vitamin B-12 intake in adult and elderly individuals is stressed.