Dietary Fiber Effects on the Microbiome and Satiety

Dietary patterns with high fiber content are linked to a lower risk for the development of cardiovascular disease [1-3], hypertension [4], type 2 diabetes [5] and increased body weight [4]. Potential biological mechanisms that may mediate these beneficial health effects include a slowing of the absorption of meal carbohydrate (CHO) [6-11], reduction in blood lipids [8,12] and an increase in the release of satiety hormones [10,13]. The PI has previously shown that compared to low-fiber (LowFi) meals, high-fiber (HiFi) meals reduced blood glucose concentrations postprandially by 11% [14]. Another potential mechanism is a postulated role for microbial fiber fermentation to improve health through production of the short chain fatty acids (SCFA) acetate, propionate, and butyrate [15-17]. In addition to promoting colon health, butyrate production may stimulate release of the gut hormones, glucagon-like peptide-1 GLP-1 and peptide YY (PYY) [18] resulting in improved appetite regulation [19]. Since the seminal paper of Gordon in 2004 [20], a large body of research has uncovered the critical role that gut microbes play in health. Importantly, much of these data, including findings supporting a beneficial role of SCFA [21-23] have been derived from animal studies. Human studies are now needed to advance the translational significance of rodent studies and the potential benefit of fiber on microbial metabolites and cardiometabolic health, glucose regulation, appetite and satiety. The current study will determine the effects of dietary fiber intake on appetite, intestinal metabolism, and the microbiome. We hypothesize that the mechanisms by which dietary fiber provides metabolic benefit include direct physical effects in the upper gastrointestinal (GI) tract to slow nutrient absorption and indirect effects to reduce food intake mediated by SCFA-induced secretion of GI hormones resulting in increased satiety. To test this hypothesis, we will conduct a randomized controlled trial of 4 weeks of HiFi or LowFi diets in 44 subjects (specific aim 1, SA1) and also leverage a screening colonoscopy to standardize baseline microbial populations for a 3-week, pre- and post HiFi intervention study in 26 subjects (SA2) with metabolic syndrome. We will assess the effects of the diets on appetite and satiety, cardiometabolic risk and intestinal metabolism at the beginning and at the end of the feeding interventions. The fiber chosen is derived from peas, as recent data suggest that legumes significantly improve glycemia [6,24-27], diabetes [28,29], heart disease risk [30], and risk for obesity [31]. These methods will be employed to accomplish two specific aims.

SA1a: Test the effect of a HiFi diet on appetite and satiety and whether SCFA production mediates improved satiety in HiFi feeding. Hypothesis (H) 1a: In adult men and women, the HiFi (n=22) compared to the LowFi (n=22) diet will significantly improve markers of satiety (GLP-1, PYY, subjective appetite ratings) and lower activation in brain regions that control food intake/reward/appetite while increasing activation in executive control regions during functional magnetic resonance imaging (fMRI) visual food cues. These changes will be related to higher postprandial SCFA concentrations and altered microbial populations as evidenced by greater bifidobacteria levels and low Firmicutes to Bacteroidetes ratio.

SA1b: Determine whether a HiFi diet improves cardiometabolic health. H1b: A HiFi diet will result in lower glycemia, blood lipids, blood pressure, and waist circumference compared to a LowFi diet.

SA2: Quantitate the changes in microbial composition and colonic SCFA production rate (using stable isotopic infusion techniques) on HiFi diet feeding (n=26) and whether any changes are potential mediators of observed benefits on satiety and cardiometabolic risk factors. H2: A significant microbial species reduction will follow colonoscopy bowel prep, and repopulation after HiFi will be characterized by greater bifidobacterial and low Firmicutes/Bacteroidetes ratio. An increase in SCFA flux following HiFi will be associated with improvements in microbial composition and postprandial markers of satiety and blood triglycerides and glucose excursions.

Sample size Based on our own published [14] and unpublished data, and that from others [32-35], a power analysis revealed that a sample size of between 10 to 20 subjects/group is needed to detect significant differences in key variables (alpha 0.05) and a power of 90% (15 to 18 subjects/group with 80% power). For specific aim 1, we will add 2 subjects/group to account for a 10% subject dropout and for specific aim 2, we will add an additional 6 subjects to account for 30% dropout. Thus, for specific aim 1 44 subjects (22/group) and for specific aim 2, 26 subjects are analyzed in a repeated-measures design. We believe any dietary fiber effect smaller than past, published treatments will be balanced by the relative 'clean' starting point of the colon after colonoscopy (specific aim 2) and also by the fact that we are providing all study meals and hence fully controlling the subject's intake

Data analysis:

Statistical analysis will be performed with SPSS software (version 25). Graphical methods are used to assess the appropriateness of assuming linear relationships and histograms and probability plots used to assess the normality of residuals. Transformation or non-parametric methods will be, employed as needed. Fasting glucose and hormones concentrations will be, obtained serially - both acutely after meals and in the fasting state before and after the diets. Changes over time (treated as a nominal factor so as not to assume a linear trend) and by diet in the composition of the microbiome will be assessed by grouping into the dominant bacterial phyla (i.e. Actinobacteria, Bacteroidetes, Firmicutes, Proteobacteria and Tenericutes) and genus. For SA1, a two-factor ANOVA will be used for each outcome, with the factors being Group, Time and the Group by Time interaction. The groups are constructed via matched-sample randomization, so we expect comparability at baseline. For SA2, a paired sample t-test will be used to compare outcomes of interest. Results will be reported as group means or medians, as most appropriate for the data along with 95% confidence intervals for the summary statistics. Analyses of the fMRI data during visual stimulation are performed using Statistical Parametric Mapping 12 software (www.fil.ion.ucl.ac.uk/spm). Data are preprocessed, beginning with slice timing and realignment of the images to the mean image. The anatomical T1-weighted image is co-registered to the mean functional image. Normalization into Montreal Neurological Institute (MNI) space and Gaussian spatial smoothing is then performed. For each participant (first-level analyses), a general linear model is applied for the high- and low-caloric food and non-food image conditions. For each condition, a separate regressor is modeled by using a canonical hemodynamic response function that includes time derivatives. Movement parameters are, modeled as confounders. For second level analysis, a mixed model ANOVA is used, with the within-factor, image condition (high calorie food, low calorie food, non-food|) and the between-factor group (HiFi vs LowFi). A priori regions-of-interest (ROIs) such as, insula, orbitofrontal cortex, amygdala, and prefrontal cortex are examined for potential group-by-food image interactions (the effect of most interest). Whole-brain analyses are also conducted (corrected for multiple comparisons) to identify other potential ROIs.