The prevalence of obesity has increased dramatically over the past few decades afflicting people of all ages, races/ethnicities and both sexes [1
]. Accordingly, obesity research efforts have focused on identifying key factors contributing to obesity development and reduction. Processes of satiation and satiety and their impact on food intake have been areas of research concentration. The macronutrient profile of a meal, including fiber, influences satiation and satiety [7
]; however, the optimal composition to maximize satiety remains controversial.
Fats are considered less satiating than carbohydrate and protein; although in energy replacement and energy density controlled paradigms this is not necessarily the case [12
]. Characteristic properties of fats, such as fatty acid chain length and saturation also have been studied, yielding conflicting results on satiety outcomes [15
]. Similarly, inconclusive results relative to fibers’ properties on satiety have been reported [16
]. Studying nutrient components out of their natural matrix may be one factor that influences satiety results, but population characteristics is an important consideration. Satiety research on fats or fibers has been conducted primarily in healthy weight adults [8
] limiting extrapolation to effects in overweight and obese individuals that may experience greater benefit from certain dietary manipulations [8
Fats and some fibers slow gastric emptying, delay nutrient absorption, modulate glucose and insulin responses, and alter gut hormones involved in satiety [7
]. Fats are potent stimulators of certain satiety peptides, including cholecystokinin (CCK), peptide YY (PYY) and glucagon-like peptide (GLP)-1 [20
]. Delaying absorption of fat with fibers could result in increased release of these peptides and enhance satiety, due to similarities in their physiological effects. Previous work suggests that under energy equivalent conditions, increasing the fat content of a meal or incorporating foods inherently containing fiber into a low-fat meal can enhance satiety over an extended post-meal time period, an effect related to the CCK response post-meal [24
]. Few studies have focused on the role of nutrient combinations (particularly fat and fiber) to enhance satiety.
Avocados are a unique fruit that contains both fats and dietary fibers. One medium size, fresh Hass avocado (~136 g) is about 72% water and contains ~13.3 g monounsaturated fats, 10 g fiber, and a variety of carotenoids and other bioactive components [25
]. Wien and colleagues (2013) reported that adding approximately half an avocado to a lunch meal suppressed the desire to eat and increased how satisfied participants felt over 5 h compared to a meal without avocado [26
]. However, adding avocado increased the energy content of the meal. In the same study, postprandial insulin concentrations were reduced after replacing some fat and carbohydrate energy (salad dressing and cookie portions) with avocado [26
]. These data suggest that strategic manipulation of meals with avocado, a source of fat and fiber, could promote both satiety and metabolic benefits. Achieving both benefits without increasing energy would be ideal, particularly for individuals with weight or glucose control concerns.
Therefore, the present study examined the satiety effects of replacing carbohydrate energy in a meal with a half or a whole avocado compared to a control meal without avocado in overweight/obese individuals. The primary research outcome was the change in subjective measures of satiety in response to meals over 6 h. Secondary outcomes were changes in variables such as tiredness and alertness, and characterization of the association between satiety/appetite related hormones (PYY, GLP-1, ghrelin and insulin) and subjective satiety after the whole avocado vs. control meals. The working hypothesis is that under energy equivalent conditions, replacing carbohydrate with fat-fiber combination from avocados will enhance satiety and the effects will be related to the meal responsive changes in satiety-mediating hormones.
2. Materials and Methods
2.1. Ethics and Study Design
This study was approved by the Institutional Review Board (IRB) of Illinois Institute of Technology (IIT) in Chicago, IL and registered with ClinicalTrials.gov
(NCT02479048). The study was conducted according to the guidelines laid down in the Declaration of Helsinki and the International Conference on Harmonization-Good Clinical Practice (ICH-GCP). All subjects provided written informed consent. This study was part of a larger study examining the effects of avocado on metabolic endpoints, vascular reactivity and lipoprotein metabolism [27
]. The study was a randomized, three-arm, single-blinded, acute, crossover design conducted from 2015–2016 at the Clinical Nutrition Research Center (CNRC) at the Illinois Institute of Technology (Illinois Tech, Chicago, IL, USA).
Thirty-nine subjects (21 men and 18 women) were enrolled and randomized in the study; 31 subjects (15 men and 16 women) completed all protocol specified procedures (Figure 1
]. Eligible subjects were men and women 20–65 years of age and BMI between 25 and 35 kg/m2
with elevated fasting glucose (5.0–6.4 mmol/L) and insulin (no greater than 90.3 pmol/L) concentrations. Participants also had to be nonsmokers and be in relatively good health with no previous history or current clinical evidence of cardiovascular, metabolic, respiratory, renal, gastrointestinal or hepatic diseases. Individuals taking medications or dietary supplements that would interfere with study outcomes, anyone reporting allergies or sensitivity to study products, and women who were pregnant or lactating were excluded. Female participants were studied during the follicular phase of their menstrual cycle. Each subject was studied once.
2.3. Study Meals and Intervention
There were three breakfast test meals: one was a low fat, high carbohydrate meal (control meal, CON) and two meals were similar in energy and energy density to CON but contained either a half (HA) or a whole (WA) of a fresh, medium-sized Hass avocado (Supplement Table S1
). Meals consisted of a bagel sandwich (with or without avocado), fresh honeydew melon, oatmeal, and a lemonade-flavored drink. Nutrient and energy content for each meal appear in Table 1
. Avocado-derived fat and fiber increased with increasing avocado content in the meal, with WA containing > 2/3 avocado-derived fat and fiber. HA was supplemented with butter fat to adjust total fat content between the two experimental (avocado) meals. Bagel sandwiches were used for manipulating components by adding avocado into hollowed bagels and using cream cheese and butter to manage desired fat levels and textural quality. Green leafy lettuce was used in all bagel sandwiches to help mask substitutions and control for color, visual appeal, and texture.
All meals were prepared in the metabolic kitchen at the CNRC under the supervision of the registered dietitian following strict food safety standards. Subjects came to the laboratory on three separate occasions and consumed each meal once based on a randomly assigned sequence by computer-generated randomization allocation list.
2.4. Assessments: Visual Analog Scale (VAS)
The VAS tool measured subjective satiety [28
] and other variables at baseline and at seven time-specific intervals post meal [24
]. The VAS questions included assessment of subjective feelings of fullness, hunger, desire to eat, prospective consumption, and how satisfied subjects felt at the moment of VAS collection. Other variables, including thirstiness, feelings of nausea, tiredness, and alertness also were assessed. Subjects were instructed to read each question carefully every time and indicate how they felt at the exact moment by drawing a straight vertical line on a 100 mm horizontal line scale anchored with “not at all” to “extremely.”
2.5. Assessments: Blood Collection and Analysis
Whole blood samples were collected into a 4 mL ethylenediaminetetraacetic acid (EDTA)-coated vacutainer tubes, inverted gently and immediately placed on ice. Within 30 min of collection, all blood samples were centrifuged at 453× g for 15 min at 4 °C. Plasma aliquots then were collected and stored at –80 °C until analysis. Prior to centrifugation, the following were added: 230 μL of aprotinin (Fisher Scientific, Fair Lawn, NJ, USA) and 40 μL of dipeptidyl peptidase IV inhibitor (EMD Millipore, Temecula, CA, USA) for GLP-17-36 and PYY3-36 and 30 μL of 4-hydroxymercuribenzoic acid for ghrelin. Glucose and insulin were analyzed using Randox Daytona Auto Clinical Analyzer (Randox, Antrim, UK) by enzyme-based assay kits (cat. no. GL3815, Randox, Antrim, UK) and immunoturbidimetry assay (cat. no. KAI071, Kamiya Biomedicals, Tukwila, WA, USA) respectively. Enzyme-linked immunoassay kits (Phoenix Pharmaceuticals, Inc. Burlingame, CA, USA) were used to analyze total ghrelin, GLP-17-36, and PYY3-36. Quality controls suggested by the manufacturers were used in all assays.
2.6. Summary of Study Procedures
Study procedures are described in detail elsewhere [27
]. Briefly, subjects were asked to follow a strictly limited avocado, olive oil, nuts and polyphenolic diet for at least three days prior to the study day and throughout the study duration, while maintaining their usual eating and physical activity patterns. In addition, subjects were asked to restrict intake of alcohol, coffee, tea, and other caffeinated beverages and limit their physical activity 24 h prior to a testing day.
Subjects were asked to consume a standardized dinner meal on the night before each study day to control variance from prior food and beverage intake. Subjects arrived at the research center in a fasted state (8–12 h fasting), well-hydrated and well-rested (+/− 1 h usual sleep pattern). Vigorous exercise, alcohol and caffeine intake were restricted 24 h prior to the visit. After reviewing subjects’ compliance and study readiness, participants completed their first set of VAS subjective satiety questions followed by placement of an intravenous catheter inserted into their antecubital vein for a fasting/baseline blood sample (0 h). Participants then were given one of the three test meals based on their assigned computer generated randomization sequence. At 0.5, 1, 2, 3, 4, 5, and 6 h after the breakfast meal, VAS again were administered. Immediately after VAS completion, blood samples were collected via the catheter. The study day visit procedures are shown in Figure 2
. Each study visit occurred at least one week apart (7–10 days).
2.7. Statistical Analysis
Demographic characteristics of study participants were analyzed using descriptive statistics. Subjective data from VAS and biological samples were analyzed by mixed-model analysis of repeated measures using PROC MIXED [29
]. All models included meal, time and meal- by-time interaction as main factors with subject as the blocking variable. VAS data were normalized to subject’s own baseline, which also corrected for baseline variance. Biological sample data were analyzed using all values with baseline values as a covariate and are presented as mean ± standard error (SE) scores over time using the Statistical Analysis System (SAS) generated least square means ± SE of the response. Area under the 6 h response curves using the trapezoidal method after controlling for baseline (iAUC) also were calculated for biological sample data. Data not conforming to expected distributional assumptions were log-transformed. For non-normally distributed data, such as ghrelin iAUC, the Wilcoxon rank sum test was performed in SAS.
Hormone data are presented as time by concentration curves. Stepwise multiple regression analysis was performed using SPSS [30
] to investigate the association of gut hormones and insulin and subjective satiety responses using only CON and WA data for each gut peptide, insulin, and VAS change from baseline. The probability of F
-to-enter is 0.05, probability of F
-to-remove is 0.10. A two-tailed p
-value < 0.05 was considered significant. Power calculations based on paired t-test indicated that a sample size of thirty subjects would provide >80% power to detect differences of 25% or greater with SD of 18 in VAS or gut hormone variables between avocado-containing meals and CON meal.
The purpose of this study was to investigate the satiety effects of avocados, a unique food source containing both fats and fibers, in an isocaloric replacement paradigm, substituting carbohydrate energy in a low-fat, high carbohydrate (Control, CON) meal with half or whole avocado. The potential impact of the fat-fiber combination derived partly or predominantly from avocados, independent of fat content, was assessed by adjusting total fat in the avocado meals. Study results indicated that replacing carbohydrate with fats and fibers derived from avocados without increasing energy or energy-density enhanced the satiety value of meals in overweight and obese individuals as evidenced by greater hunger suppression after WA compared to CON (WA > CON = HA) and increased feelings of how satisfied subjects felt over 6 h (WA = HA > CON). Satiety induced by WA was associated mostly with PYY responses. In contrast, satiety induced by CON was related mostly to insulin responses. These findings suggest isocaloric dietary manipulation with a whole avocado promotes favorable metabolic responses in addition to enhancing satiety and reducing motivation to eat.
Optimal meal composition to maximize satiety remains controversial. In general, protein is considered to be the most satiating followed by carbohydrate and then fat [31
]. Importantly however is whether energy and volume is controlled in study designs comparing macronutrients and satiety. Fat has a clear disadvantage because it has twice the energy density of other macronutrients. Early evidence suggests that when palatability and energy density is held constant, subjective satiety, and energy intake as a measure of satiating effect was not different between carbohydrate and fat preloads [13
]. This has been observed when foods were ingested orally or infused intravenously or intragastrically [12
]. Likewise, texture, and other sensory cues must be considered when investigating satiety effects of macronutrients. In the present study, energy, energy density, palatability, and various sensory factors (including visual appeal) were matched closely. All meals looked and had texturally similar components. All meals were well liked (pleasantness mean score >60 for all meals and not significantly different) and all suppressed hunger and induced satiety. However, compared to CON, the WA meal produced a significantly greater reduction in hunger and increase in how satisfied participants felt after the WA meal compared to CON. Subjects also reported feeling more satisfied after the HA meal compared to the CON, while other subjective satiety responses were not significantly suppressed/enhanced by HA. The HA meal had similar total fat content to the WA meal, but less avocado-derived fat and fiber than WA meal. These data suggest that having a higher level of fat in the meal is associated with subjects feeling more satisfied, whereas increasing both the avocado-derived fat and fiber combination offers additional satiety-inducing benefits influencing several subjective variables. Whether it is the fat component or the fiber component driving the effect is the subject for subsequent research. Findings from this study suggest that replacing carbohydrate energy with fat energy in combination with a source of inherent (rather than added) fiber, as demonstrated here with avocados, can increase the satiety value of a meal compared to a low-fat high-carbohydrate meal.
The control of food intake, satiety and appetite involves complex processing of non-homeostatic and homeostatic factors, the latter including physiological status and the dynamics of fuel metabolism and episodic gut hormones in response to food intake. Flint et al. reported that insulin concentrations were inversely correlated with feelings of hunger, whereas glucose concentrations were not related to hunger or satiety [34
]. In this current study, feelings of hunger were significantly suppressed with the WA meal, yet insulin was also markedly reduced [27
]. GLP-1 and PYY are anorexigenic hormones and produce feelings of satiety and inhibit food intake [35
]. Fat is a potent stimulator of GLP-1 and PYY, explaining their significant elevation after WA meal compared to CON meal. PYY was identified as the primary contributor to the satiety response of WA, with GLP-1 and ghrelin contributing only modestly. In contrast, ghrelin suppression was greater after CON compared to the WA meal, consistent with carbohydrate being more potent than fat or protein to suppress ghrelin [37
]. However, analysis did not reveal ghrelin as playing a significant role in hunger or other subjective satiety responses to CON. Instead, insulin was consistently identified as explaining much of CON meal-associated satiety variance, except in the case of fullness, in which GLP-1 contributed 32% and insulin added an extra 8%. These findings suggest that how
satiety is achieved through biological signaling may have important implications. Achieving satiety through sufficient release of pre-absorptive satiety peptides may be ideal versus exaggerated insulin responses to meals.
Insulin is a master hormone in fuel trafficking and metabolism and has varied roles centrally, including food intake regulation. A dietary pattern that is associated with elevated insulin has been identified with increasing risk of hyperglycemia, metabolic syndrome and insulin resistance [39
], particularly in an overweight/obese population. Hyperinsulinemia and insulin resistance in the periphery can produce hypoinsulinemia and/or modified/dampened insulin signaling in the brain [42
]. Individuals in the current study were overweight or obese and had elevated fasting insulin concentrations with insulin resistance [27
]. Findings suggest that replacement strategies for carbohydrate with avocado fats and fibers reduced the post meal insulin demand, increased satiety gut hormone concentrations and improved the subjective profile of satiety. Whether the effects are due to improved insulin sensitivity peripherally or centrally to augment satiety signaling cannot be determined in the present research, but provides the basis for follow up mechanistic work in appropriate models. From a practical standpoint, the population studied represents a typical cohort of middle-aged people at risk for cardio-metabolic disease—a point when realistic dietary changes can make a significant impact on reversing the disease risk trajectory.
This study has strengths and limitations. An important strength was the application to real life. Isocaloric replacement of carbohydrate for fat is not easy yet may have advantages for appetite and post-meal metabolic control. Strategic manipulation of menu items maintaining consistent energy density and sensory variables was achieved using readily available foods in grocery stores. Avocados provided healthy fat and fibers, adding to the simplicity of the macronutrient exchanges. One study limitation was the lack of a test meal to measure subsequent food/energy intake 6 h after meals as the meal-related time course of subjective satiety (particularly later in the postprandial period) suggested an effect on subsequent meal intake may have been plausible. Not separating the specific fat type and fiber effect also may be a limitation. Finally, gut hormones were not measured after HA. This information may have provided insight on the observed differences in fullness ratings between HA meal and CON and WA meals.