The effect of cereal Β‐glucan on body weight and adiposity: A review of efficacy and mechanism of action

Abstract

The current review examines the totality of the evidence to determine if there exists a relationship between β‐glucan and body weight and adiposity and whether such a relationship is a consistent, causal and plausible one. Observational studies suggest an association between oat (i.e., β‐glucan) intake and reduced body weight, waist circumference and adiposity. High and moderate quality randomized controlled trials that were specifically designed to evaluate the efficacy of β‐glucan on anthropometric outcomes were given the highest weight. Several of these studies indicated a causal relationship between β‐glucan consumption and reduction in body weight, BMI, and at least one measure of body fat within diets that were not calorie-restricted. A review of additional animal and human evidence suggests multiple plausible mechanisms by which β‐glucan may impact satiety perception, gastric emptying, gut hormones, gut microbiota and short chain fatty acids in the complex interplay of appetite and energy regulation.

Supplemental data for this article is available online at http://dx.doi.org/10.1080/10408398.2021.1994523

Keywords:

• Oat
• barley
• β‐glucan
• body weight
• adiposity
• satiety
• gastric emptying
• gut hormones
• gut microbiota
• short chain fatty acids

Introduction

From 1975 to 2016 worldwide obesity nearly tripled, with more than 1.9 billion adults, 18 years or older being overweight, of which over 650 million were obese (World Health Organization (WHO) 2020). During the same period, 340 million children aged 5–19 years were overweight or obese (WHO, 2020). The primary cause for the global rise in obesity is a decline in physical activity and the increased intake of energy dense foods (WHO, 2020). Overweight and obesity are associated with several co-morbidities, including type 2 diabetes, cardiovascular diseases, musculoskeletal disorders, and certain cancers (WHO, 2020; Chan and Woo 2010). Even weight loss as low as 5% of body weight can reduce or eliminate disorders associated with obesity (Blackburn 1995; Pasanisi et al. 2001). Prevention of overweight and obesity can be achieved by engaging in regular physical activity and limiting the intake of higher calorie, nutrient poor foods and increasing the consumption of fruits, vegetables, legumes, nuts, and whole grains (WHO, 2020).

Increasing evidence suggests that whole grains and dietary fiber may play a specific role in weight management. Large prospective studies of cereal fiber (Koh-Banerjee et al. 2004; Du et al. 2010) and whole grain (Koh-Banerjee et al. 2004; Bazzano et al. 2005) intake consistently report an inverse relationship with body weight, BMI, and risk of obesity. Meta-analyses of randomized controlled trials indicate lower body weight for both dietary fiber (27 studies) and whole grain (11 studies), when consumers with the highest consumption levels were compared to the lowest (Reynolds et al. 2019). Isolated soluble fiber supplementation in a meta-analysis of 12 randomized controlled studies was shown to significantly reduce BMI, body weight, and body fat, independent of energy restriction (Thompson et al. 2017). A more extensive meta-analysis of 62 randomized controlled trials of viscous fiber in whole foods or isolated sources demonstrated a reduction in body weight, BMI and waist circumference, with no change in body fat, in subjects consuming a diet with no energy restriction (Jovanovski et al. 2020). About half of the viscous fiber studies were studies of β‐glucan from oats or barley (Jovanovski et al. 2020). These researches conducted a second meta-analyis of viscous fiber supplementation in subjects consuming calorie-restricted diets and found body weight, BMI, and body fat were significantly decreased (Jovanovski et al. 2021). Among the 15 randomized controlled trials included in the meta-analyses, 7 were oat and barley β‐glucan studies (Jovanovski et al. 2021).

The richest sources of β‐glucan soluble fiber are from oats and barley (Wood 2011). The non-digestible mixed-linkage polysaccharides (1→3) (1→4)- β-D-glucans are primarily found in the endosperm cell walls or bran of whole grain oats or barley and is measurable by standard methods (Wood 2011). Β- Glucan’s highly branched molecular structure and molecular weight contributes to its ability to form highly viscous solutions (Wood 2011). In 2009, the European Food Safety Authority offered a scientific opinion on the substantiation of a health claim between β‐glucan and the maintenance or achievement of normal body weight (EFSA Panel on Dietetic Products and Nutrition and Allergies (EFSA) 2009). The authority reported that based on the 77 references provided by the petitioner, a few measured satiety, but none addressed the effects of β‐glucan on body weight, and therefore, a cause-and-effect relationship was not established (EFSA Panel on Dietetic Products and Nutrition and Allergies (EFSA) 2009). A review published in 2012 reported that most intervention studies that aimed to achieve an isocaloric diet did not find significant effects of β‐glucan on body weight, BMI, or waist circumference (Cloetens et al. 2012). Six studies were included in this review, but the primary objectives of five of these were not explicitly designed to assess anthropometric changes. Another review reported mixed results of oats and β‐glucan on body weight changes, but noted that most of the studies examined were designed to measure blood lipids or blood pressure and some studies asked subjects to maintain constant body weight (Cook, Rains, and Maki 2013).

Although β‐glucan is a component of cereal fiber and whole grain intake, prospective studies did not specifically delineate the contribution of β‐glucan. Similarly, it is unclear what the specific contribution of β‐glucan was on body weight and obesity in the meta-analyses of clinical trials conducted on isolated soluble fiber supplements (Thompson et al. 2017) and viscous fiber in non-energy-restricted diets (Jovanovski et al. 2020) and in calorie-restricted diets (Jovanovski et al. 2021). A recent meta-analysis of 11 randomized controlled studies did evaluate the effect of β‐glucan consumption and reported lowered body weight by 0.77 kg and BMI by 0.62 kg/cm², with no significant impact on waist circumference (Rahmani et al. 2019). However, this meta-analysis did not include the full breadth of β‐glucan clinical studies that have investigated its effect on body weight and/or adiposity measurements and it conflicts with the findings of earlier reviews (Cloetens et al. 2012; Cook, Rains, and Maki 2013). Thus, it is the aim of this review to summarize the totality of the evidence related to the effect of β‐glucan on body weight and adiposity from observational studies and clinical intervention studies in overweight adults and children. Potential mechanisms of action are also reviewed to determine if the data supporting a relationship between β‐glucan and body weight and adiposity is a plausible one.

Methods

Literature search and study selection criteria

In order to investigate the efficacy of cereal β‐glucan, multiple literature searches in PubMed, Google Scholar and Scopus using the terms “oat”, “barley”, “β‐glucan” and “body weight”, “obesity” and “adiposity”, as well as hand searching recent reviews were performed. Only studies published in English were included and those only available in abstract form were excluded. Studies that examined fiber blends from grains other than oats or barley were excluded as the independent effect of oats/β‐glucan could not be determined. The focus of the searches for observational studies were those that reported intake of oats, barley or β‐glucan from either source and at least one measure of body weight or adiposity. Measures of overweight and adiposity included BMI, waist circumference, or body fat. The criteria for the selection of human intervention trials are summarized in Table 1. Additional human and animal studies, as well as reviews, were identified to understand potential mechanisms.

Table 1. Criteria for study selection for human intervention studies.

Population	Overweight adults or children above the age of 2 years
Intervention	Consumption of β‐glucan from oats or barley
Comparator	Placebo, foods, or background diet without β‐glucan
Outcome	Changes in body weight and/or measures of adiposity
Time	At least 4 weeks duration each for intervention and control periods
Study design	Parallel or crossover randomized controlled trials within background diets that were either energy-restricted or not (i.e., habitual or modified but not hypocaloric).

Identification of studies and assessment of potential bias

Four observational studies were identified that investigated the relationship between oat intake and body weight and adiposity parameters. A total of 72 human intervention studies were identified that comprised two categories of β‐glucan studies: A) studies whose primary or secondary aim was to evaluate the effect on anthropometric and adiposity measures as reported in the abstract or introduction of the article; B) studies that were designed to investigate the cholesterol-lowering properties or hypoglycemic or other health effect, but included body weight and/or adiposity measures. These two categories were analyzed separately as the latter category of studies are subject to potentially significant bias. The hypocholesterolemic and glycemic effects of β‐glucan have been examined in numerous studies, many with the aim of maintaining stable body weight during the intervention in order not to confound the treatment effect of β‐glucan on blood lipid, blood glucose, or other metabolic changes sensitive to alterations in body weight. Subjects were often told to maintain their body weight and therefore these studies would be biased against body weight changes. Hence, studies where subjects were instructed to maintain their bodyweight were excluded from our analysis. A total of 16 studies were initially identified in category A and 56 studies in category B. After implementation of inclusion/exclusion criteria, 5 studies were excluded in category A and 32 studies were excluded in category B, leaving 11 and 24 randomized controlled trials for quality evaluation in each of the categories, respectively. The quality assessment involved an assessment of potential biases that would impede an accurate assessment of anthropometric and adiposity changes following oat, barley or β‐glucan supplementation. In category A, 6 studies were deemed high to moderate quality and 5 studies were identified as low quality or high bias. In category B, 10 studies were considered moderate quality and 14 as low quality or higher bias studies. Supplementary Table 1 summarizes the quality criteria used and Supplementary Table 2 outlines the application of inclusion/exclusion criteria and quality assessment of studies in category A. Supplementary Table 3 provides similar information for studies in category B. The quality methodology used in this review is similar to the process that regulatory agencies such as the U.S. FDA use in evaluating health claims for food ingredients.

Table 2. Observational studies of oat intake and body weight.

Reference	Subjects	Study Design	Diet Assessment	Significant Results	Control for Confounders
Damsgaard et al. 2017	713 children from 9 Danish schools, 8–11 y	Cross-sectional	7-d food records	WG oats inversely associated with fat mass index^1(β = 0.06, p = 0.03)	Linear mixed model adjusted for whole grain rye, wheat, school, class, age, sex, height, puberty, parental education, physical activity, blood EPA + DHA, intakes of E, fruits, vegetables, and saturated fats.
O’Neil et al. 2015	14,690 nationally representative US children, 2–18 y (NHANES 2001–2010)	Cross-sectional	24-hr dietary recall	Cooked oatmeal consumers vs. non-consumers had lower WC (66.2 cm vs. 67.9 cm, p = 0.003), but no significant differences in BW, BMI, BMI z-score or BMI-for-age percentile. Cooked oatmeal consumers vs. non-consumers had a 40% lower risk of being obese (OR = 0.60, p = 0.026), 33% lower risk of being overweight or obese (OR= 0.67, p = 0.026) and a 64% lower risk of having an elevated WC (OR= 0.36, p = 0.04)	Covariate-adjusted regression analysis and logistic regression analysis for differences in odds ratio adjusted for sex, race, ethnicity, poverty income ratio, physical activity, smoking status, and alcohol consumption. No significant differences in E intake between oatmeal consumers and consumers.
Fulgoni et al. 2015	22,823 nationally representative US adults, ≥ 19 y (NHANES 2001–2010)	Cross-sectional	24-hr dietary recall	Cooked oatmeal consumers vs. non-consumers had lower BW (79.1 kg vs. 81.7 kg, p < 0.01), BMI (27.5 kg/m^2 vs 28.5 kg/m^2, p < 0.01), WC (94.9 cm vs. 97.5 cm, p < 0.01).	Least squares mean regression analysis adjusted for sex, race/ethnicity, age, poverty income ratio, physical activity, smoking status and alcohol consumption. No significant differences in E intake between oatmeal consumers and consumers.
Musa-Veloso et al. 2016	18,273 nationally representative US children, 0–18 y and adults ≥ 19 y (NHANES 2009–2012)	Cross-sectional	24-hr dietary recall	Among cooked oatmeal consumers, underweight (β = 3.01, p < 0.0001), normal weight (β = 1.30, p < 0.0001), and overweight (β = 0.51, p = 0.0001) individuals consumed significantly more oatmeal than obese individuals.	Multiple linear regression model not adjusted for potential confounders.

¹Fat mass index calculated as total fat mass per kg/m² based on DXA scan.; Whole body composition, including fat mass was measured by DXA scan.

BW: body weight; BMI: body mass index; E: energy; OR: odds ratio; WC: waist circumference.

Table 3. Randomized controlled studies whose primary or secondary outcomes included body weight and/or adiposity measures.

Reference	Primary/ Secondary Outcomes	Subjects, Country	BMI (kg/m^2)	Study design, Duration	Background Diet	Comparator	Daily BG Dose, Source	↓Total E Intake^1	↓BW^1	↓BMI^1	↓WC^1	↓Other Adiposity Measure^1	Study Quality^2
STUDIES GIVEN THE HIGHEST WEIGHT (HIGH AND MODERATE QUALITY)
Aoe et al. 2017	Visceral fat, BMI	100 OW Japan	27.9 ± 2.7 27.2 ± 2.8	RCT, P, DB, 12 wk	Habitual	BG-free pearled barley	4.4 g Pearled barley	No	Yes	Yes	No	No (VFA, SCUFA) Yes (VF for subgroup VF ≥ 100 cm^2)	HI
Chang et al. 2013	Obesity, fat deposition, WC, serum parameters, hepatic steatosis	40 OW Taiwan	29.1 ± 2.3 29.5 ± 2.5	RCT, P, DB, 12 wk	Habitual	Hot placebo cereal	3 g Hot oat cereal	NR	Yes	Yes	NR	Yes (Body fat, WHR)	MO
Raimondi de Souza et al. 2016	Blood lipids blood glucose, anthropo-metric parameters	132 OW HC Brazil	28.5 29.4	P, DB, 12 wk	Healthy	Cornstarch, rice flour porridge	3 g OB porridge	No	No	No	No	No (neck perimeter)	MO
Reyna-Villasmil et al. 2007	Blood lipids, BW	38M, HC, OW Venezuela	28.4 ± 0.8 28.2 ± 0.8	RCT, P, 8 wk	AHA Step II	whole wheat bread	6 g Nutrim-OB bread	No	Yes	Yes	NR	NR	MO
Smith et al. 2008	Compare HMW vs. LMW barley BG on CVD endpoints, BW.	90 HC, OW United States	26.7 ± 4.2 26.0 ± 3.2	RCT, P,DB, 6 wk	Habitual	6 g LMW BG barley	6 g HMW barley	No	Yes	NA	NA	NA	MO
Shimizu et al. 2008	Blood lipids, BMI, WC, VF, SCUF	44M, HC, OW Japan	26.2 ± 2.4 24.5 ± 2.8	RCT, P, DB, 12 wk	Habitual	Rice	7 g Barley	No	NR	Yes	Yes	Yes (VFA) No (SCUFA_)	MO
STUDIES GIVEN THE LOWEST WEIGHT (LOW QUALITY)
Beck et al. 2010	BW, satiety, blood lipids, glucose, appetite hormones	56 F OW Australia	29.3 ± 2.2 29.3 ± 2.1 29.2 ± 2.2	RCT, P, SB, 12 wk	Hypocaloric	Wheat porridge, RTE, snacks (relatively high fiber)	5–6 g 8–9 g OB porridge, RTE, snacks	No	No	NR	No	NR	LO
Geliebter et al. 2014	BW, subjective appetite, CVD risk factors	46 OBS United States	32.9 ± 4.7	P, 4 wk	Habitual	Water	4 g Oat porridge	NR	No	NR	No	No (WHR, fat mass)	LO
Li et al. 2016	Weight management, blood glucose, blood lipids.	238 T2D, OW China	26.9 ± 2.7 27.4 ± 2.4 27.2 ± 2.8	P, 30 d, 1 wk run-in baseline	LF, HF diet T2D	LF, HF diet T2D	2.65 g WG oats	No	No	No	No	No (VF)	LO
							5.3 g WG oats	No	No	No	No	No (VF)
				1 y follow-up			2.65 g WG oats	NR	No	No	No	No (VF)	LO
							5.3 g WG oats	NR	Yes	No	No	No (VF
Reyna et al. 2003	Metabolic profile, BW, BMI	16M, OW, T2D Venezuela	28.5 ± 1.7 28.9 ± 2.0	P, 4 wk	Am. Diabetic	Am. Diabetic Diet	NR Oat- Oatrim bread, Oatrim-artif. Sw. cookies (low E)	(Yes) Part of design	Yes	Yes	NR	NR	LO
Saltzman et al. 2001	BW, BP, blood lipids, insulin sensitivity	43 OW United States	26.1 ± 3.4 2.67 ± 3.2	P, 2 wk BW maintenance, 6 wk BW loss	Hypocaloric	Hypocaloric diet	2.2 g Oatmeal	NC (stats NR)	No	NR	NR	NR	LO

¹Significant difference for BG group vs. comparator, p < 0.05. ²Study quality was assessed in relation to potential biases for BW and/or adiposity measures only; HI: high; MO: moderate; LO: Low. See Supplementary Tables 1 and 2 for details of assessment.

AHA: American Heart Association; BG: beta-glucan; BP: blood pressure; BW: body weight; DB: double-blind; E: energy; HF: high fiber; HC: hypercholesterolemic; HMW: high molecular weight; LF: low fat; LMW: low molecular weight; NC: not clear; NR: not reported; OB; oat bran; OBS: obese; OW: overweight; P: parallel; RTE: ready-to-eat; SB: single-blind; SCUFA: subcutaneous fat area; Ss: subjects; Sw: sweetened; TDF: type 2 diabetes; Tx: treatment; VF: visceral fat; VFA: visceral fat area; WC: waist circumference; WG: whole grain; WHR: waist-hip-ratio.

Observational studies of oat intake and anthropometric measures

Table 2 summarizes four cross-sectional studies that investigated the relationship between oat intake and body weight and adiposity parameters (Dammsgaard et al. 2017; O’Neil et al. 2015; Fulgoni et al. 2015; Musa-Veloso et al. 2016). Similar studies for barley intake were not found. The beneficial effect of oats was reported in a study of 713 Danish children, aged 8–11 years (Damsgaard et al. 2017). Regression analysis indicated that whole grain oat intake (g/MJ) was inversely associated with the fat mass index, after adjustment for several potential confounding factors, including age, gender, puberty, physical activity, and energy intake. A nationally representative sample of U.S. children, aged 2–18 y from 2001 to 2010 National Health and Nutrition Examination Survey (NHANES) reported that children that consumed cooked oatmeal had a smaller waist circumference or lower central adiposity than children who did not eat cooked oatmeal (O’Neil et al. 2015). The oatmeal eaters also had a significantly lower risk of being obese, obese/overweight, or having central adiposity. Analysis of a nationally representative sample of U.S. adults, 19 years or older from 2001 to 2010 NHANES data also revealed body weight, BMI and waist circumference were significantly lower among cooked oatmeal consumers versus non-consumers (Fulgoni et al. 2015). Potential confounding factors such as physical activity, gender, race, and income were controlled in the regression analyses and energy intake was not significantly different between oatmeal consumers and non-consumers in both NHANES studies of children and adults (O’Neil et al. 2015; Fulgoni et al. 2015).

The third observational study also reported a beneficial effect of oat intake, but clear conclusions could not be made due to study limitations (Musa-Veloso et al. 2016). Evaluation of NHANES data from 2003 to 2012 demonstrated that among cooked oatmeal consumers, those that were classified as underweight, normal weight or overweight based on BMI, consumed more oatmeal than obese oatmeal consumers (Musa-Veloso et al. 2016). However, the regression analysis was limited to only oatmeal consumers, and within this group, only obese consumers were compared to the other BMI categories. Additionally, the regression analysis did not adjust for potential confounding factors such as physical activity and energy intake.

Overall, three observational studies (Damsgaard et al. 2017; O’Neil et al. 2015; Fulgoni et al. 2015) provide evidence for a beneficial relationship between oat intake and body weight and adiposity measures. Two of the studies used large, heterogeneous samples. However, a limitation of cross-sectional studies is that while an association can be established, causal inference cannot be made.

Randomized controlled intervention studies of oat and barley Β‐glucan and body weight and/or adiposity

Category A: Studies with body weight, BMI or adiposity as primary/secondary endpoints

Cause-and-effect is determined by well-designed randomized controlled intervention studies. The findings from 11 randomized controlled studies whose primary or secondary objectives included anthropometric measures are summarized in Table 3 ranked by study quality. Nine of the studies were conducted within the context of background diets that were not energy-restricted. One study was deemed high quality (Aoe et al. 2017) and five studies were considered moderate quality (Chang et al. 2013; Raimondi de Souza et al. 2016; Reyna-Villasmil et al. 2007; Smith et al. 2008; Shimizu et al. 2008). The highest weight was given to these six studies.

Aoe et al. (2017) randomized 100 overweight Japanese subjects to consume 4.4 g/d of barley β‐glucan in pearled barley or pearled barley with no β‐glucan in a 12 week, parallel-group, double-blind study. The primary aim of this study was to investigate whether high β‐glucan barley substituted for rice in their habitual diet would reduce visceral fat. There was no difference in energy intake between the two groups. Body weight and BMI were significantly reduced in the β‐glucan group compared to the control. Waist circumference, visceral fat area, and subcutaneous fat area decreased in both groups and the changes were not significantly different from each other. However, a subgroup of subjects with visceral fat area ≥ 100 cm² had a significantly greater decrease in visceral fat area in the β‐glucan group compared to the control group.

A randomized, parallel-group, double-blind study was conducted by Shimizu et al. (2008) in 44 overweight Japanese males who consumed 7 g/d of β‐glucan from barley or a rice control for 12 weeks as part of their usual diet. Although total energy intake in the β‐glucan group was not significantly different, BMI, waist circumference, and visceral fat area were significantly reduced compared to the control group. There was no significant difference between the two groups for change in the subcutaneous fat area.

Smith et al. (2008) randomly assigned 90 overweight US men and women to treatments of low molecular weight (LMW) or high molecular weight (HMW) concentrated barley containing 6 g of β‐glucan within the context of their usual diets in a double-blind, parallel-group, 6-week study. Following the intervention, body weight decreased in the HMW group, whereas body weight increased in the LMW group. Hunger ratings were also significantly lower in the HMW group compared to the LMW group. Changes in energy intake from baseline did not significantly differ between the two groups.

Chang et al. (2013) performed a randomized, parallel-group, double-blind study in 40 overweight Taiwanese subjects. Subjects consumed 3 g of β‐glucan in a hot oat cereal or a hot placebo cereal for 12 weeks as part of their habitual diet. Although the two study cereals were isocaloric, it is unclear if total daily energy intakes of the subjects were similar as this information was not reported even though subjects kept diet records. At the end of the intervention, the reduction in body weight, BMI, % body fat, and waist-to-hip-ratio was significantly greater in the β‐glucan group than the group that consumed the placebo.

In an 8 week randomized, parallel-group study, Reyna-Villasmil et al. (2007) investigated the effect of 6 g of β‐glucan from soluble oat fiber in bread versus a whole wheat bread control in 38 overweight male Venezuelan subjects who were consuming the American Heart Association Step II diet. Energy intake during the two intervention periods were similar. Both groups experienced significant weight loss and a decrease in BMI, but the β‐glucan group had a greater reduction.

Raimondi de Souza et al. (2016) randomized 132 overweight Brazilian men and women into two groups that consumed 3 g β‐glucan in oat porridge or cornstarch, rice flour control porridge in a 12 week, parallel-group, double-blind study. Subjects followed a healthy diet as prescribed by the Ministry of Health that did not differ in energy intake between the two groups. Results indicated both the β‐glucan and the control group significantly reduced body weight, BMI, waist circumference and neck perimeter with no significant differences between them.

Five studies were given low quality ratings for increased bias toward body weight measurements and therefore were not part of the evidence in assessing the effect (Beck et al. 2010; Geliebter et al. 2014; Li et al. 2016; Reyna et al. 2003; Saltzman et al. 2001). Reasons for the low ratings are outlined in Supplementary Table 2. Some of the reasons included lower compliance (75%) of some of the β‐glucan products used (Beck et al. 2010), inappropriate control (Geliebter et al. 2014), body weight of subjects was not stabilized on lower calorie, low fat, high fiber diet compared to usual diet prior to intervention in a study of relatively short duration (Li et al. 2016), intervention diet was lower in calorie than the control diet (Reyna et al. 2003), and body weight of subjects were not stabilized on a highly restricted calorie diet of 1000 kcal/d prior to intervention (Saltzman et al. 2001).

In summary, the majority of the 6 high and moderate quality studies given the highest weight demonstrated a significant effect of β‐glucan on body weight and measures of adiposity. All 6 were parallel-group, randomized studies of overweight subjects whose baseline BMI did not differ between treatment and comparator groups. Five of the six studies were double-blind. Three were oat studies and three were barley studies. The β‐glucan dose ranged from 3 to 7 g with study durations of 6–12 weeks, with most having a duration of 12 weeks. Four studies employed habitual background diets, while one study utilized a defined healthy diet and another the American Heart Association Step II diet. Five of the studies that reported on background energy intake did not observe significant differences between the treatment and control groups. Among 5 studies that measured bodyweight and BMI, 4 or 80% observed a significant reduction for both parameters. Waist circumference was measured in three studies, but only one study or 33% reported a significant decrease. One or more adiposity-related outcomes measures (visceral fat, body fat, or waist-hip-ratio) were measured in 4 studies and 3 or 75% observed a significant effect in the total sample or a subgroup that had higher visceral adiposity.

Category B (studies with body weight/BMI or adiposity measurements but not as primary/secondary endpoints)

Ten studies out of 24 studies whose primary or secondary outcomes were not body weight and/or adiposity measures were considered moderate quality (Biӧrklund et al. 2005; Braaten et al. 1994; Chen et al. 2006; Connolly et al. 2016; Gulati, Misra, and Pandey 2017; Katz et al. 2005; Maki et al. 2010; Uusitupa et al. 1992; Wolever et al. 2010; Zhang et al. 2012) (Supplementary Table 3). Only one study utilized an energy-restricted background diet (Maki et al. 2010). Among 7 studies that assessed body weight and/or adiposity changes, none of the studies observed a significant effect for both outcomes, however, one study reported a trend toward a decrease in body weight (P = 0.07) (Chen et al. 2006), and another indicated a trend toward a reduction in BMI (p = 0.06) (Connolly et al. 2016). Two out of 4 studies noted a decrease in waist circumference (Maki et al. 2010; Zhang et al. 2012). Other adiposity measures were evaluated in 3 studies and none reported a significant change, although one study reported a trend toward a reduction in fat mass (p = 0.06) Connolly et al. 2016). It is unclear why these studies did not observe significant anthropometric changes. In most instances, the studies were designed to investigate blood cholesterol, blood glucose or blood pressure outcomes that would be confounded with significant changes in body weight, hence, even though it was not specifically reported, the possibility that subjects were instructed to maintain their body weight cannot be ruled out. Studies may also have lacked statistical power as the sample size was not geared toward assessing significant changes in body weight. The remaining 14 studies were considered low quality within this category. Reasons for the low ratings are summarized in Supplementary Table 3.

Mechanism of action

Our review demonstrates that well-designed studies of oat or barley β‐glucan significantly reduce body weight, BMI and adiposity, independent of caloric restriction. The ability of oat β‐glucan to impact body weight and adiposity measures is biologically plausible as evidenced from animal and human intervention studies that have examined changes in satiety perception, gastric emptying, gut hormones, and gut microbiota. It is also the property of viscous and soluble fibers in general (Jovanovski et al. 2020; Jovanovski et al. 2021; Thompson et al. 2017).

Satiety perception

During a meal when eating is inhibited by the feeling of “fullness” is termed satiety (Amin and Mercer 2016). Subjective satiety is often measured by ratings on visual analogue scales. Several acute studies have evaluated the acute effects of oat and barley β‐glucan on subjective satiety with mixed results. It is beyond the scope of this review to summarize each of these, however, a recent review of whole grain β‐glucan containing foods and β‐glucan extracts has been published (Rebello, O’Neil, and Greenway 2016). In their review, Rebello, O’Neil, and Greenway (2016) evaluated 16 human oat β‐glucan intervention studies and reported that although some studies of β‐glucan supplementation did not show an effect on satiety, the majority of studies demonstrated that oat β‐glucan increases perceptions of satiety. Among studies that reported on the physicochemical properties, viscosity was found to be an important factor in enhancing satiety. Increasing doses of β‐glucan from 2.2 g to 5.5 g was shown to increase satiety, however, the effects were not always consistent. This may have been due to other factors that are also important, such as solubility, extractability, and molecular weight. These factors are also impacted by processing, cooking, and storage treatments.

Gastric emptying

Due to their ability to absorb large quantities of water, it has been hypothesized that viscous fibers promote gastric distention and greater volume in the stomach, leading to a delay in gastric emptying and a sense of fullness. In a randomized crossover trial, Geliebter et al. (2015) compared energy-equivalent breakfasts of oatmeal and cornflakes in 18 lean and 18 obese men and women and observed that gastric emptying was the most rapid for water, followed by cornflakes and then oatmeal, which reached a peak the latest. The slower gastric emptying observed for oatmeal is likely attributed to the dietary fiber content of oatmeal (8 g total, 4 g soluble), whereas cornflakes, contained no dietary fiber. Consumption of a liquid formula lunch meal and feelings of hunger 180 minutes after cereal consumption was significantly less following the intake of oatmeal versus cornflakes or water.

The molecular weight of β‐glucan is an influencing factor. Thondre, Shafat, and Clegg (2013) observed that the high molecular weight (HMW, i.e., 650 kDA) β‐glucan soup containing 12.9 g β‐glucan significantly delayed gastric emptying compared to the low molecular weight (LMW, i.e., 150 kDa) β‐glucan soup containing 3.6 g β‐glucan compared to the energy and carbohydrate matched, reduced fiber control. The higher viscosity of the HMW β‐glucan soup likely contributing to the slower transit of nutrients into the duodenum. Similarly, Wolever et al. (2020) found oatmeal with added oat bran (4 g β‐glucan) had a significantly longer gastric emptying half-time and lag time than reduced molecular weight and low viscosity oatmeal at the same dose of β‐glucan. However, ad libitum intake of pizza 3 hours after breakfast did not significantly differ between the two. Juvonen et al. (2009) also observed slower gastric emptying following the intake of a high viscosity oat bran beverage compared to a low viscosity oat bran beverage that were identical in volume, energy, macronutrient, and β‐glucan content. Contrary to expectation 3 hours after the beverages were consumed, the low viscosity oat bran beverage had a significantly higher subjective satiety rating than the high viscosity oat beverage, and energy consumption of an ad libitum lunch meal of soup and cheese sandwich did not differ between the two. However, when energy intake of the ad libitum lunch was combined with energy intake for the rest of the day, the total energy intake of the high viscosity oat bran beverage was significantly lower than that of the low viscosity oat bran beverage. Particle size may also impact gastric emptying rates. Magnetic resonance imaging (MRI) revealed little difference initially in a study that compared oat flakes and oat flour meals with the same mass, macronutrient and b-glucan meals, but toward the end of the gastric cycle there was evidence of more flakes being retained in the stomach (i.e., slowed gastric emptying) (Mackie et al. 2017). During the initial phase, the liquid layer appeared to be emptied and not the oat flakes themselves.

Two studies reported no effect of two oat products on gastric emptying and satiety perception. In one study that utilized 12 healthy volunteers, 50 g oat flakes and 50 g wheat bran flakes were compared to 50 g cornflakes (control) and the wheat bran flakes was shown to have the slowest gastric emptying rate compared to oat flakes (Hlebowicz et al., 2007). However the gastric emptying rates of both wheat bran flakes and oat flakes were not significantly different from the control. These results may have been due to the high dietary fiber content of the bran flakes (7.5 g) compared to the oat flakes (4 g) which contained only 0.5 g β‐glucan (Hlebowicz et al. 2007). The lack of a significant effect in a second study that compared isocaloric vanilla yogurt with oat bran muesli (9 g TDF, 4 g β‐glucan) to the same yogurt with cornflake muesli is unclear (Hlebowicz et al. 2008).

Although there are some discrepancies, overall, the data shows that β‐glucan significantly impacts the rate of gastric emptying. The degree to which this is impacted is based on a variety of factors that include molecular weight, viscosity, and particle size.

Gut hormones

The ability of viscous fibers to prolong the presence of nutrients in the GI tract, increases the potential interaction between nutrients and the intestinal mucosa to stimulate the release of appetite-regulating hormones, many of which influence energy intake (Rebello, O’Neil, and Greenway 2016). Several randomized controlled trials have assessed the potential effects of β‐glucan intake on appetite-regulating hormones. An acute meal study reported a 16% higher plasma peptide YY (PYY) response and a 23% lower plasma ghrelin response following a breakfast of barley β‐glucan enriched bread versus a control bread in healthy volunteers (Vitaglione et al. 2009). PYY has been shown to reduce gut motility and reduce food intake, whereas ghrelin regulates the opposite of these functions (Chaudhri, Small, and Bloom 2006; Rebello, O’Neil, and Greenway 2016). Barone Lumaga et al. (2012) compared 3 g barley β‐glucan beverage with a control beverage with no fiber consumed at breakfast and noted satiety perception was significantly increased 3 hours post-breakfast with energy intake reduced by 18% at lunch and 40% over the rest of the day. Ghrelin was suppressed by 8.1%, whereas the food intake inhibitor, pancreatic polypeptide was increased by 34.6%. Beck et al. (2009a) reported a positive dose response relationship between increasing doses of oat β‐glucan (2.2 g, 3.8 g, 5.5 g) and increasing concentrations of plasma, PYY 2 to 4 hours post meal in 14 overweight adults (Beck et al. 2009a). Oat β‐glucan was also shown to increase postprandial cholecystokinin (CCK) levels in a dose-dependent manner in overweight subjects (Beck et al. 2009b). CCK is widely distributed in the gastrointestinal tract and central nervous system and has several functions, including slowing gastric emptying and suppression of food intake (Little, Horowitz, and Feinle-Bisset 2005).

Of particular interest was the observation that Beck et al. (2009a) found PYY levels were greatest at 4 hours for the highest β‐glucan dose, demonstrating that longer time frames may be required to show full satiety effects of β‐glucan. The shorter time frame of 3 hours for satiety and gut hormone measurements in the Juvonen et al. (2009) study may explain why the low viscosity oat beverage was shown to increase satiety compared to the high viscosity oat beverage. The faster transit of the low viscosity oat beverage caused higher levels of PYY and other gut hormones to be released initially, but the delayed effect of the high viscosity oat beverage resulted in a lower total caloric intake over the course of the day (Juvonen et al. 2009; Beck et al. 2009a). A shorter study time frame (i.e., ≤ 3 hours) may also be a contributing factor in other studies (Juvonen et al. 2011; Wolever et al., 2020; Zaremba et al. 2018) that did not observe significant satiety, gut hormone changes and/or energy intake effects at lunch. Evidence of a potential second meal effect would also not be uncovered.

One long-term study evaluated the impact of β‐glucan on satiety, gut hormone concentrations, and body weight. Beck et al. (2010) tested control and two doses (5–6 g, 8–9 g) of β‐glucan from oat bran formulated in breakfast cereal and snacks in 56 overweight women on an energy-restricted diet for 3 months. All groups lost weight and waist circumference, with no significant difference between the groups. Hormonal changes were contrary to expectations. Leptin, a mediator of long-term energy balance and suppressor of food intake was significantly decreased, as were GLP and PYY levels across all groups. Even with the same body weight change compared to the control, the greatest decrease in PYY was observed in the 5–6 g β‐glucan group and not in the 8–9 g β‐glucan group. The findings are difficult to interpret and are limited by the reduced compliance of the β‐glucan snacks, the use of a high fiber control and the large standard deviations observed, particularly in relation to hormonal changes.

Rodent models provide additional support by demonstrating that consumption of oat β‐glucan reduce energy intake and body weight and increase gut hormones associated with appetite regulation. Young male rats fed oat β‐glucan for four weeks had 10% lower food intake, 37% lower body weight, 26% less body fat, and higher concentrations of plasma GLP-1 and PYY than the controls (Adam et al. 2014). A dose-dependent relationship between oat β‐glucan intake, energy intake and body weight has been observed in diet-induced obese mice, which has been associated with increased plasma levels of PYY (Lin et al. 2013; Huang et al. 2011). Miyamoto et al. (2018) demonstrated similar results in their study when barley flour containing 2% or 5% β‐glucan was fed to mice consuming a high-fat diet for 12 weeks. These mice reduced their food intake and improved insulin sensitivity, had increased secretions of gut hormones PYY and GLP-1, gained less weight and had a lower fat mass than similar control mice, not fed barley flour.

The human and animal data collectively demonstrate that β‐glucan positively impacts gut hormones in appetite regulation. In the human studies where short-term energy intake was not impacted by β‐glucan, it is possible that the study design was too short as there is some data in support for a second meal effect.

Gut microbiota, short chain fatty acids and energy regulation

There is increasing evidence that gut microbiota changes may be part of the mechanism through which β‐glucan and other fibers impact body weight and adiposity. Transplantation of microbiota from diet-induced obese mice into lean germ-free mice results in a marked increase in adiposity compared to microbiota transplanted from lean mice donors, suggesting that gut microbiota is a key component of energy regulation (Ley et al. 2008). Two phyla that are the most predominant in the gut are Bacteroidetes and Firmicutes and the ratio or relative abundance of the two is frequently associated with obesity (Dreyer and Liebl 2018; Amabebe et al. 2020). Both phyla affect the production of short chain fatty acids (SCFAs) (Dreyer and Liebl 2018). Firmicutes, found primarily in the upper region of the small intestine, is efficient in extracting energy from the diet and storing it as adipose tissue (Dreyer and Liebl 2018). Bacteroidetes, commonly found in the colon, produce SCFAs in response to fiber-rich diets (Dreyer and Liebl 2018).

Only a limited number of human studies have examined gut microbial shifts following β‐glucan supplementation, and only one that has reported on the relative changes in Bacteroidetes and Firmicutes in relation to anthropometric varaibles. In a randomized, controlled crossover study, Wang et al. (2016) compared the effects of consuming breakfasts containing 3 g HMW, 3 g LMW, 5 g LMW barley β‐glucan or a wheat and rice control on gut microbiota for 5-weeks in mildly hypercholesterolemic subjects. At the phylum level, the 3 g/d HMW β‐glucan increased Bacteroidetes and decreased Firmicutes compared to the control. No significant gut microbiota changes were observed for both the LMW β‐glucan doses. At the genus level, 3 g/d HMW β‐glucan increased Bacteroides, decreased Dorea and showed a trend toward an increase in Prevotella (p < 0.1). The shifts in Bacteroides, Prevotella, and Dorea significantly correlated with shifts in BMI and waist circumference. Bacteroides and Prevotella belong to the phylum Bacteroidetes and Dorea belongs to the phylum Firmicutes. On the other hand, another study reported a shift toward an increased Firmicutes and decreased Bacteroidetes pattern following oatmeal consumption, but correlational analysis with adiposity measures was not performed (Ye, Sun, and Chen 2020).

The risk of obesity has also been attributed to a relative decrease in Bifidobacterium spp. or Akkermansia muciniphila (Amabebe et al. 2020). Lower numbers of Bifidobacterium have been observed in obese people than in individuals that are lean (Schwiertz et al. 2010; Santacruz et al. 2010). Similarly, Akkermansia muciniphila inversely correlates with body weight and obesity in both humans and rodents (Santacruz et al. 2010; Everard et al. 2013). A limited number of human studies have examined the changes in Bifidobacteria and Akkermansia muciniphila following β‐glucan supplementation. A significant time by treatment interaction was observed for fecal bifidobacteria levels in a randomized crossover study by Connolly et al. (2016) where 32 hypercholesterolemic subjects consumed 45 g of whole grain oat granola (1.3 g β‐glucan) for 6 weeks. Velikonja et al. (2019) indicated in their randomized, parallel-group study of 6 g barley β‐glucan versus control that pre-intervention microbiota was important in the nature and size of dietary intervention. Hence differences in gut microbiota were assessed in the 51 subjects enrolled in the study who were divided into total cholesterol-responsive and total cholesterol-non responsive groups. Certain bacterial groups that were perceived as health-promoting, such as Bifidobacterium bifidum, Bifidobacterium fecale and Akkermansia muciniphilia were found to be higher in the total cholesterol responsive group. However, changes in these microbial species were not assessed in the two groups following β‐glucan supplementation. Although Wang et al. (2016) observed specific gut microbial anti-obesity shifts, changes in Bifidobacteria and Akkermansia were not significantly different between the oatmeal and control groups. Overall differences in gut microbial changes among studies after oatmeal consumption may be attributed to differences in microbial concentrations and diversity present in the popultations studied, diet, other host factors, as well as differences in the β‐glucan dose and the methodology used to assess microbial communities.

Short chain fatty acids (SCFAs) are the bacterial fermentation by-products of dietary fiber, including β‐glucan. SCFAs are a major energy source of colonic epithelial cells and SCFA oxidation is estimated to recover 600–750 kJ of energy per day from undigested carbohydrates (Blaak et al. 2020). In addition to being a source of energy harvest for colonocytes, SCFAs modulate metabolic pathways and receptor-mediated mechanisms, and activate gut-brain neural circuits in appetite regulation and energy homeostasis(Byrne et al. 2015; Sukkar et al. 2019; Chambers et al. 2015a). G-protein-coupled receptors (GPR43, GPR41), also known as free acid receptor 2 and 3 (FFAR2, FFAR3) expressed at numerous tissues sites, including the gut epithelium and adipose tissue are activated by SCFAs (Byrne et al. 2015; Sukkar et al. 2019; Chambers et al. 2015a). The expression of FFA2 and FFA3 in enteroendocrine cells that release GLP-1, PYY, and other anorectic hormones in the gastrointestinal tract and leptin from adipocytes suggest SCFAs may act as triggers, evidenced by FFAR2 and FFAR3 knockout mice that exhibit reduced levels of these hormones in response to SCFAs (Byrne et al. 2015; Chambers et al. 2015a). Further, rodent models demonstrate that acute and chronic administration of SCFAs reduces body weight by increasing energy expenditure via increases in lipid oxidation (Sukkar et al. 2019; Blaak et al. 2020; den Besten et al. 2015; Hattori et al. 2010; Gao et al. 2009). Three acute human studies and of SCFA supplementation demonstrate similar results (Sukkar et al. 2019; Van der Beek et al. 2015; Canfora et al. 2017; Chambers et al. 2018). Distal colonic infusions of acetate in 6 overweight/obese men increased fat oxidation and fasting PYY levels (Van der Beek et al. 2015). Colonic infusion of three SCFA mixtures high in either acetate, propionate and butyrate increased fasting fat oxidation, fasting and postprandial PYY concentrations in a study of normoglycemic men (Canfora et al. 2017). Resting energy expenditure was also increased after the acetate and propionate infusions compared to placebo (Canfora et al. 2017). Intake of oral sodium propionate by healthy volunteers during a 180-minute study revealed increased resting energy expenditure that was accompanied by elevated rates of whole body lipid oxidation, independent of changes in glucose and insulin concentrations (Chambers et al. 2018). In addition, a study of chronic supplementation of propionate for 24 weeks in overweight and obese subjects significantly reduced body weight compared to a control (Chambers et al. 2015b). Additional long-term human studies of SCFAs, either through direct administration or through the use of prebiotics are warranted to follow-up on these intriguing findings.

Another emerging concept in energy homeostasis involves SCFAs and intestinal gluconeogenesis(IGN) (Blaak et al. 2020). There is evidence that butyrate and propionate activate IGN in enterocytes, where propionate itself acts as a substrate and is converted into glucose (De Vadder et al. 2014; Blaak et al. 2020). The glucose that is released by IGN is detected by the portal neural system that sends a signal to the brain that aids in regulating energy homeostasis, including decreased fat storage and body weight (De Vadder et al. 2014; Blaak et al. 2020).

Several in vitro and animal studies have shown an increase in the production of SCFAs (acetate, propionate, and butyrate) by gut microbiota following the inclusion of oats, oat bran or β-glucan to batch culture or diets (Korczak, Kocher, and Swanson 2020). There have only been a limited number of human studies and the results are variable. Fecal SCFAs in human intervention studies are often used as a biomarker of gut-derived SCFA production, but these are not indicative of in vivo colonic production as the body rapidly and almost completely absorbs the SCFAs that have been produced (Blaak et al. 2020). Animal studies of β-glucan intake permit a more accurate assessment of the role SCFAs in energy and body weight regulation.

Fecal SCFAs were hardly detectable in germ free mice fed a high-fat diet supplemented with 2% β‐glucan barley flour for 12 weeks (Miyamoto et al. 2018). Plasma levels of PYY and GLP-1, body weight and fat mass were also not significantly different from the control, demonstrating that the metabolic effects of β‐glucan are dependent on the microbial fermentation and subsequent SCFA production. Another study evaluated whether the intake of 8% β‐glucan barley flour affects the expression of genes related to glucose and lipid metabolism in the ileum, liver, and adipose tissue in mice fed a high fat diet for 92 days (Mio et al. 2020). There were no significant differences in final body weight or food intake, but liver weight and retroperitoneal and mesenteric fat weights were significantly lower in the β‐glucan group versus the control group. Serum leptin and postprandial blood glucose concentrations were also significantly lower in the β‐glucan group. Concentrations of total SCFAs, and specifically acetic acid and propionic acid were higher in the β‐glucan group than in the control group on analysis of cecal contents. DNA microarray revealed 10% of genes involved in lipid metabolism were decreased in liver and adipose tissues. Furthermore, a negative correlation was observed between some SCFAs and expression of mRNA linked to lipid synthesis and degradation, which the authors concluded affect lipid accumulation. Arora et al. (2012) also examined the anti-obesity efficacy of mice fed a high fat diet containing 10% barley derived β‐glucan (∼80%) preparation for 8 weeks (Arora et al. 2012). Energy intake and body weight were significantly lowered in the β‐glucan group versus the control. Cecal and fecal contents of the group fed β‐glucan demonstrated an increase in Bifidobacterium and Lactobacillus-Enterococcus. Total levels of SCFAs, as well as acetate and propionate levels were significantly increased in the β‐glucan group compared to the control group. In addition, manganese enhanced magnetic resonance imaging (MEMRI) of β‐glucan fed mice showed marked suppression of neuronal signal intensities in the satiety centers of the brain (i.e., arcuate nucleus, ventromedial hypothalamus, paraventricular nucleus, periventricular nucleus and the nucleus of the tractus solitarius), suggesting a satiated state.

The currently available data suggests β‐glucan supplementation may be associated with anti-obesity shifts in gut microbiota and the production of SCFAs that may directly impact satiation and energy regulation. Additional human studies are need to confirm the potential anti-obesity shift in gut microbiota after β‐glucan intake.

Different mechanisms for HMW versus LMW β‐glucan?

A few human studies that have compared HMW to LMW β‐glucan have indicated delayed gastric emptying (Thondre, Shafat, and Clegg 2013) and decreased body weight and hunger ratings (Smith et al. 2008) with HMW versus LMW. One acute study observed a delay in gastric emptying with HMW β‐glucan but did not observe differences in satiety perception, ghrelin response or energy intake at an ad libitum lunch when intake of HMW and LMW oat β‐glucan were compared; however, as discussed previously it is unclear if the short study duration contributed to the null results (Wolever et al., 2020).

Aoe et al. (2020) investigated the effect of HMW (500 kDa, 94% TDF, 94% β‐glucan) and LMW (12 kDa, 45% TDF, 33% β‐glucan) barley β‐glucan in mice fed a moderate-fat diet for 61 days. Although food intake was significantly in lower both groups, body weight gain was significantly lower in the HMW group compared to the control. Liver, total abdominal fat, retroperitoneal fat, epididymal fat, and mesenteric fat were also significantly lower in the HMW group versus the control group. Fecal fat excretion was higher and apparent digestibility of fat was lower in the HMW group than in the control group. The LMW group did not significantly differ from the control for these indices. However, both groups significantly lowered serum leptin, total and LDL cholesterol, and mRNA expression of sterol regulatory element-binding protein-1c (SREBP-1c). High fermentation rates were observed in both groups, but the gut bacterial counts of Bifidobacterium and Bacteroides and cecal total SCFAs was smaller in the HMW group compared with LMW group. The authors suggest HMW β‐glucan exerts its effect by the inhibition of nutrient absorption due to its high viscosity in the gastrointestinal tract, whereas the effects of LMW are due to its prebiotic effects. Whether these conclusions are relevant for all comparisons of β‐glucan molecular weight and viscosity is still to be determined, as well as the potential effect of food matrix and processing.

Conclusions

Most oat and barley foods are typically high in fiber, whole grain, nutrient dense, and satiating, making them natural choices for inclusion in diets that promote weight maintenance. Furthermore, there is increasing consistent evidence that oat or barley β‐glucan intake prevents weight gain and reduces obesity. Three observational studies show an association between increased oat intake and lower body weight, waist circumference and reduced adiposity in both children and adults. Causal inference can be made from six well-designed randomized controlled intervention studies of β‐glucan supplementation that demonstrate a significant reduction in body weight, BMI and at least one measure of body fat. These effects were apparent in overweight subjects from diverse population groups that were consuming diets that were not energy-restricted. The ability of β‐glucan to impact body weight and adiposity measures is biologically plausible as evidenced from animal and human intervention studies that have examined changes in satiety, gastric emptying, gut hormones, and gut microbiota and their metabolites (i.e., SCFAs). The physiological mechanisms involve a complex interplay of multiple systems in appetite and energy regulation that are just beginning to be elucidated. The potential impact of the characteristics of the β‐glucan itself, such as molecular weight, viscosity, food matrix, and processing on these mechanisms still needs to be clearly elucidated. Additional research is warranted to confirm and extend our findings.

Disclosure statement

VS and YC are employed by PepsiCo, Inc. The views expressed in this manuscript are those of the authors and do not necessarily reflect the position or policy of PepsiCo, Inc.

Additional information

Funding

This work was funded by PepsiCo Inc.