Disclosure: The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Army or the Department of Defense. Any citations of commercial organizations and trade names in this report do not constitute an official Department of the Army endorsement of approval of the products or services of these organizations. JPK, AJY, and JCR have no conflicts of interest. SJM reports having received an honorarium from the Nestle Corporation.
Version of Record online: 16 APR 2013
Copyright © 2013 The Obesity Society
Volume 21, Issue 3, pages E244–E252, March 2013
How to Cite
Karl, J. P., Young, A. J., Rood, J. C. and Montain, S. J. (2013), Independent and combined effects of eating rate and energy density on energy intake, appetite, and gut hormones. Obesity, 21: E244–E252. doi: 10.1002/oby.20075
Funding agencies: This study is funded by the US Army Medical Research and Materiel Command.
- Issue online: 16 APR 2013
- Version of Record online: 16 APR 2013
- Accepted manuscript online: 18 OCT 2012 12:59PM EST
- Manuscript Accepted: 27 AUG 2012
- Manuscript Revised: 2 JUN 2012
- Manuscript Received: 13 FEB 2012
- US Army Medical Research and Materiel Command
Energy density (ED) and eating rate (ER) influence energy intake; their combined effects on intake and on postprandial pancreatic and gut hormone responses are undetermined. To determine the combined effects of ED and ER manipulation on voluntary food intake, subjective appetite, and postprandial pancreatic and gut hormone responses.
Design and Methods:
Twenty nonobese volunteers each consumed high (1.6 kcal g−1; HED) and low (1.2 kcal g−1; LED) ED breakfasts slowly (20 g min−1; SR) and quickly (80 g min−1; FR) ad libitum to satiation. Appetite, and pancreatic and gut hormone concentrations were measured periodically over 3 h. Ad libitum energy intake during the subsequent lunch was then measured.
Main effects of ED and ER on energy intake and a main effect of ER, but not ED, on mass of food consumed were observed, FR and HED being associated with increased intake (P < 0.05). Across all conditions, energy intake was highest during FR-HED (P ≤ 0.01). Area under the curve (AUC) of appetite ratings was not different between meals. Main effects of ED and ER on insulin, peptide-YY, and glucagon-like peptide-1 AUC (P < 0.05) were observed, FR and HED being associated with larger AUC. No effects on active or total ghrelin AUC were documented. Total energy intake over both meals was highest during the FR-HED trial with the greatest difference between FR-HED and SR-LED trials (P ≤ 0.01).
Consuming an energy dense meal quickly compounds independent effects of ER and ED on energy intake. Energy compensation at the following meal may not occur despite altered gut hormone responses.
Obesity and associated comorbidities persist as predominant public health concerns. Rising rates of obesity have coincided with an increased availability of a variety of palatable, highly processed energy dense foods suggesting a relationship between weight gain and the modern food environment (1, 2). Within this environment, food-related cues along with social, cognitive, and behavioral factors may promote over consumption and weight gain by overriding physiologic signals mediating energy balance (3, 4).
Epidemiological studies have identified dietary energy density (ED), defined as the metabolizable energy per gram of food (kcal g−1), as an environmental factor potentially associated with obesity (5-7). Clinical trials support this association as covert manipulation of dietary ED does not affect the mass of food consumed (8-10) and substituting high-ED for low-ED foods in the diet results in increased energy intake in excess of energy needs and weight gain (11). Further, energy dense foods are often highly processed, and marketed and packaged in a way that facilitates eating quickly (e.g., fast foods and portable, “ready-to-eat” meals) (4). This may compound the independent effect of ED on energy intake as eating fast has also been shown to increase energy intake during single meals (12-14), and has been associated with higher BMI and being overweight (15-17). Together these findings suggest that energy dense foods that are consumed at a fast rate may facilitate over-consumption.
Appetite is mediated by a variety of neural and endocrine signals including several pancreatic and gut hormones such as insulin, glucagon-like peptide-1 (GLP-1), peptide-YY (PYY), pancreatic polypeptide (PP) and ghrelin (18). The combined effects of eating rate (ER) and ED on energy intake and on the pancreatic and gut hormones that influence energy intake and appetite are undetermined. In this study, we experimentally manipulated ER and covertly increased the ED of a meal consumed ad libitum, while measuring perceived appetite, postprandial pancreatic and gut hormone responses, and ad libitum energy intake. The primary objective was to determine the combined effects of ER and ED manipulation on subjective appetite and energy intake measured during consecutive meals. A secondary objective was to examine associated effects on postprandial pancreatic and gut hormone responses. We hypothesized that energy intake would be greater both during high- ED compared to low-ED meals and meals consumed at a fast compared to slow rate, and that these independent effects would be additive. Further, we hypothesized that differences in energy intake resulting from ER and ED manipulation would not affect subjective and objective indices of appetite despite altered postprandial hormone responses.
Methods and Procedures
Nonobese (BMI 18.0-27.0 kg m−2) healthy men and women, age 18-55 years, were recruited from the U.S. Army Soldier Systems Center, Natick, MA and surrounding area. Exclusion criteria included previous diagnosis with any disease known to affect metabolism or current use of medications affecting metabolism and/or appetite, ≥2.2 kg body mass change during the 3-month preceding study participation, recent pregnancy, allergies to or dislike of the test foods, and a score of ≥20 on the Eating Attitudes Test (19). The study was approved by the human use review committee at the U.S. Army Research Institute of Environmental Medicine. Volunteers participated in these studies after providing their free and informed voluntary consent. Investigators adhered to US Army Regulation 70-25 and US Army Medical Research and Materiel Command regulation 70-25 on the use of volunteers in research.
A repeated measures crossover design was used with each participant completing a screening session and four experimental visits. Anthropometric characteristics, dietary restraint and disinhibition, and acceptability of the breakfast meal were assessed during screening. Body mass index was calculated from measured height and weight, and body fat percentage was measured by DXA (Lunar iDXA, GE Healthcare, Madison, WI). Acceptability of the breakfast meal was confirmed using a nine-point hedonic scale. Dietary restraint and disinhibition were measured with the Three Factor Eating Questionnaire (20).
Testing was conducted on non-consecutive days with no more than two weekly visits. Participants were instructed to adhere to their normal diet and physical activity patterns on the day preceding each study visit, fast for 12 h prior to each visit, and to abstain from strenuous physical exercise, consuming products containing caffeine, and smoking during the fast. Compliance was monitored by 24-h diet and activity recall.
Two ad libitum meals, breakfast and lunch, were provided during each experimental visit. The breakfast meal was a low-ED (LED; 1.2 kcal g−1) or high-ED (HED; 1.6 kcal g−1) oatmeal consumed at a prescribed eating rate; 20 g min−1 (SR) or 80 g min−1 (FR). The order of trials was random and determined using a Latin-square design. Approximately 15 min prior to beginning breakfast, an indwelling venous catheter was placed and a fasting blood sample collected. Blood was then sampled at 15, 30, 45, 60, 90, 120, and 180 min (T180) after the first bite of food was taken. An additional blood sample was collected immediately after finishing breakfast if meal completion was ≥3 min from a scheduled blood draw (Tmc). Appetite was rated concomitant with each blood sample collection. Lunch was served following the blood sample at T180. Meal acceptability was rated after both breakfast and lunch. Between meals activity was restricted to seated activities such as watching television or reading.
Breakfast consisted of an oatmeal entrée (Table 1) and water. To create the LED oatmeal, the ED of the HED oatmeal was lowered by increasing the water content and decreasing the amounts of energy-containing ingredients. To maintain texture, a soluble gum was added to the LED oatmeal and cook time was increased. The energy content of each recipe was calculated by summing the metabolizable energy of each ingredient based on product label information, and ED calculated by dividing total metabolizable energy by the mass of the oatmeal after cooking.
|Energy density, kcal g−1b||1.20 ± 0.01||1.63 ± 0.02|
|Instant oatmeal, flavoredd||10.4||16.8|
|Instant oatmeal, plaine||2.7||3.5|
|Heavy whipping creame||13.7||19.2|
|Whey protein powdere||2.3||3.0|
|Composition per 100 g|
|Carbohydrate, g [%]||11.5 ||16.4 |
|Fat, g [%]||5.9 ||8.3 |
|Protein, g [%]||3.3 ||4.6 |
As served, the HED and LED oatmeal were similar in macronutrient proportion, taste, appearance and texture, and the mass (1,034 ± 176 g) was constant across trials for each volunteer. Volume of the served portion was not measured, but visually no differences could be detected. Volunteers were instructed to eat until “comfortably full” while maintaining the prescribed ER. More food than could be consumed was served and no restrictions on meal duration were imposed. Total energy intake was calculated by multiplying the weighed difference between the uneaten and served portions of oatmeal by the ED. Water intake during experimental meals was fixed at 60 g every 3 min. During the postprandial period, 360 g of water was provided for participant comfort and to ensure adequate blood volumes could be obtained. The water was consumed at the discretion of the volunteer, but finished prior to lunch, and no additional water was permitted.
The lunch meal was Stouffer's® Lasagna (1.4 kcal g−1, 41% energy from carbohydrate, 36% energy from fat, 23% energy from protein). Participants were served more food than they could likely consume and portion size (812 ± 139 g) was kept constant during all trials. Participants were instructed to eat until “comfortably full,” could request more food, and no restrictions on meal duration were imposed. Participants were provided with no more or less than 240 g of water, and finished the water before meal completion.
Eating rate manipulation
Eating rate during breakfast was manipulated using a portable, calibrated scale interfaced with a small computer monitor (Mandometer, AB Mando, Huddinge, Sweden) (14, 21). Participants maintained a constant ER during the experimental meal by following a pre-programmed eating curve graphed on the computer monitor. The curve was programmed to be a straight line with a slope of 20 or 80 g min−1, intersecting the y-axis at the mass of food served. As the participant ate, the amount of food removed from the plate was plotted in real-time. By matching the real-time eating curve to the preprogrammed eating curve, participants complied with the prescribed eating rate.
Perceived appetite was measured using the Satiety Labeled Intensity Magnitude Scale (SLIM). The SLIM is a vertical, 100 mm, bidirectional hunger-fullness scale that incorporates phrases describing different levels of hunger/fullness along the length of the scale with the phrases “greatest imaginable fullness” and “greatest imaginable hunger” serving as anchors (22). The sensitivity and reliability of the SLIM is comparable to that of visual analog scales used for measuring perceived appetite (22). Negative ratings are indicative of hunger whereas positive ratings indicate fullness. Negative and positive SLIM values are denoted throughout the manuscript by the terms “hunger” and “fullness,” respectively.
Participants were naïve to the study purpose. As the ER manipulation could not be disguised, participants were told they were participating in a trial examining the effects of ER on nutrient absorption and reaction time. Following each postprandial blood sample, participants were required to complete a 5 min vigilance task presented on a personal-digital assistant. No participant indicated that they suspected meal composition was being altered or food intake measured.
Biological sample collection and analysis
Venous blood samples were collected from an indwelling catheter placed in the antecubital space. Aprotinin and dipeptidyl peptidase-IV inhibitor were added to chilled EDTA tubes for measurement of PYY and GLP-1. For ghrelin measurements, blood was collected into chilled EDTA tubes containing phenylmethylsulfonyl fluoride and plasma was treated with 1 N hydrochloric acid. All samples were frozen immediately after processing and placed in storage. Serum glucose was measured using the Beckman Coulter DXC 600 Pro (Fullerton, CA). Serum insulin was measured by RIA (Siemens Medical Solutions, Los Angeles, CA). The sensitivity of the assay was 13.9 pmol L−1 and the inter-assay CV < 8.6%. Plasma PYY3-36, and active (acylated; Ghrelinact) and total ghrelin (Ghrelintot) were measured by RIA (Linco Research, St. Charles, MO). The sensitivity and inter-assay CV for the PYY, Ghrelinact, and Ghrelintot assays were 2.5 pmol L−1 and <12.2%, 7.8 ng/L and <16.9%, and 100 ng L−1 and <12.2%, respectively. Plasma active GLP-1 (GLP-17-36 amide and GLP-17-37) was measured by ELISA (Linco Research). The sensitivity of the assay was 13.7 ng L−1 and the inter-assay CV <17.0%. Plasma PP was measured using a multiplex assay (Milliplex MAP; Millipore, Billerica, MA). The sensitivity of the assay was 3.3 pmol L−1 and the inter-assay CV <16.5%.
Analyses were completed using commercially available statistical software (SPSS 18.0; SPSS, Chicago, IL). Sample size calculations assuming a SD of 168 kcal (21), an expected within-subjects correlation of 0.80, and α = 0.01 to account for multiple comparisons, indicated that 17 volunteers were needed to detect a 100 kcal between meal difference in energy intake at a power = 0.80.
Descriptive statistics are presented as mean ± SD or geometric mean ± geometric SD if log-transformed prior to analysis (insulin peak and insulin Tmc). Fasting appetite, glucose, insulin, GLP-1, PP, PYY, Ghrelinact, and Ghrelintot were compared across meals using 1-way repeated measures ANOVA with Bonferroni corrections. Time courses of appetite, glucose and hormone responses during each trial were plotted separately for each volunteer. The following summary measures were used to characterize these responses: peak or nadir of response, rating or concentration at Tmc, rating or concentration at T180, and area under the curve (AUC) with respect to increase. Two-factor repeated measures ANOVA was used to examine main and interactive effects of ER and ED on these summary measures. Energy intake, mass of food and total mass consumed (food + water) during breakfast, energy intake during lunch, and total cumulative energy intake (breakfast + lunch) were also examined using two-factor repeated measures ANOVA. Sex was included as a between-subjects factor in models in which energy intake was the outcome to examine potential sex differences. For all two-factor repeated measures ANOVAs, when a significant ER-by-ED interaction was observed post-hoc comparisons were made using all possible t tests with Bonferroni corrections; because the comparison between the SR-HED and FR-LED trial was not of interest, only five post-hoc comparisons were examined and a P value of ≤ 0.01 used to determine significance.
Linear regression models were used to examine associations between energy intake, mass of food and total mass consumed with SLIM, glucose and hormone AUCs, and SLIM AUC with glucose and hormone AUCs. Multiple linear regression models were used to further examine the effects of energy intake and mass consumed on appetite, glucose and hormone responses. Glucose, hormone or SLIM AUC was entered as the dependent variable, energy intake and total mass or mass of food consumed at breakfast were entered as predictors. Subject was entered as a categorical factor using dummy variables in all regression models (23). Excepting the post-hoc comparisons noted above, significance was set at P ≤ 0.05 for all tests.
During analysis, large inter-individual variability in mean Ghrelinact, and Ghrelintot values were noted. Further analysis indicated much of this variability was associated with a specific week of data collection in which Ghrelinact, and Ghrelintot values of all volunteers tested prior to that week were noticeably lower than all volunteers tested thereafter. No obvious differences in the sample collection, handling or analysis protocols were noted, and none of the other measured hormones were affected. Nonetheless, the reliability of the absolute Ghrelinact and Ghrelintot values is questionable; therefore, ghrelin responses were analyzed as percent change from baseline rather than in absolute terms.
Twenty participants (12 male, 8 female), 30 ± 11 years, BMI 24 ± 2 kg m−2, 26% ± 8% body fat were enrolled and completed all four experimental trials. Mean dietary restraint (6.5 ± 3.3) and disinhibition (2.6 ± 1.5) were low. All participants complied with the prescribed ERs. Meal duration was 15 min (95% CI: 11-20 min) longer when breakfast was eaten at the slow compared to fast rate (P effect ≤ 0.05), while ED-manipulation did not affect meal duration (Table 2). Perceptually, the manipulated breakfast meals were indistinguishable as no differences in ratings of appearance, odor, flavor, texture, or overall acceptability were noted as a result of altering ER and ED (data not shown). Likewise, acceptability ratings of the lunch meal did not differ across trials (data not shown).
|Slow (20 g min−1)||Fast (80 g min−1)|
|LED (1.2 kcal g−1)||HED (1.6 kcal g−1)||LED (1.2 kcal g−1)||HED (1.6 kcal g−1)||Effectsb|
|Meal duration, min||25 ± 12||22 ± 13||8 ± 3||8 ± 4||ER|
|Food mass, g||498 ± 2351||451 ± 2611||582 ± 2891,2||615 ± 2852||ER, ERxED|
|Total massc, g||981 ± 473||895 ± 519||753 ± 358||792 ± 340||ER|
|Energy, kcal||601 ± 2831||733 ± 4192||697 ± 3361,2||999 ± 4593||ER, ED, ERxED|
Ad libitum intake at breakfast
The mass of food consumed during breakfast was not affected by ED, but was higher on average when meals were eaten at the fast rate (P effect ≤ 0.05, Table 2). The effect of ER; however, was limited to the HED meal as eating at the fast compared to slow rate resulted in a greater mass of food being consumed during the HED meals but not the LED meals (P interaction ≤ 0.05). The total mass (food + water) consumed at breakfast was on average higher when meals were eaten at the slow rate (P effect ≤ 0.05, Table 1), reflecting a greater water intake during the SR meals due to the longer meal duration. The total mass consumed at breakfast was not affected by ED.
Main and interactive effects of ER and ED on energy intake at breakfast were observed (Table 2). Energy intake was higher on average during the FR (P effect ≤ 0.05) and HED (P effect ≤ 0.05) meals. The effect of ER on energy intake; however, was only observed during the HED meals while the effect of ED was observed at both the slow and fast eating rates (P interaction ≤ 0.05). Further, the effect of ED on energy intake was potentiated by eating at the fast rate (P interaction ≤ 0.05) as energy intake was 22% higher during the SR-HED compared to the SR-LED meal (P ≤ 0.01) but 43% higher during the FR-HED compared to FR-LED meal (P ≤ 0.01). Across all meals, energy intake during the FR-HED meal was greater than that observed during any other meal with the greatest difference between the FR-HED and SR-LED meal [398 kcal (99% CI: 238 kcal, 558 kcal) P ≤ 0.01]. This additive effect of ER and ED was observed in both men and women. The absolute difference in energy intake between the FR-HED and SR-LED meals was greater in men [496 kcal (95% CI: 360, 632 kcal)] than in women [250 kcal (95% CI: 84, 417 kcal)] (P interaction ≤ 0.05); however, when expressed as a percent difference no effect of sex was observed [−10% (95% CI: −42%, 62%), P = 0.69].
Glucose and hormone responses
Differences in the mass of food and energy consumed following ER- and ED-manipulation of the breakfast meal resulted in altered pancreatic and gut hormone responses (Figure 1). Main effects of both ER and ED on insulin, GLP-1 and PYY AUC were observed with AUC being higher during the FR (P effect ≤ 0.05) and HED trials (P effect ≤ 0.05). Though not significant, PP AUC tended to be higher during the FR trials (P effect = 0.06). Glucose, Ghrelintot, and Ghrelinact AUC were not affected by ER or ED.
Main effects of both ER and ED on glucose, insulin, and GLP-1 concentrations at the completion of the breakfast meal were observed (Table 3). Glucose, insulin, and GLP-1 concentrations were higher at the completion of the SR (P effect ≤ 0.05) and HED meals (P effect ≤ 0.05). Neither ER nor ED had any significant effect on PYY, PP, Ghrelinact or Ghrelintot concentrations at meal completion.
|Slow (20 g min−1)||Fast (80 g min−1)|
|LED (1.2 kcal g−1)||HED (1.6 kcal g−1)||LED (1.2 kcal g−1)||HED (1.6 kcal g−1)||Effectsb|
|Glucose, mmol L−1|
|Tmc||5.1 ± 0.51||5.6 ± 0.72||4.8 ± 0.41||5.0 ± 0.41||ER, ED, ERxED|
|Peak||5.5 ± 0.6||6.2 ± 1.0||5.7 ± 0.8||6.0 ± 0.8||ED|
|T180||4.8 ± 0.4||4.9 ± 0.5||4.8 ± 0.5||4.9 ± 0.7|
|Insulin, pmol L−1c|
|Tmc||95.1 ± 16.7||154.2 ± 16.7||47.9 ± 13.9||59.0 ± 13.9||ER, ED|
|Peak||199.3 ± 11.1||305.6 ± 11.1||218.0 ± 10.4||339.6 ± 10.4||ED|
|T180||75.0 ± 44.4||109.7 ± 63.2||90.3 ± 52.8||152.1 ± 87.5||ER, ED|
|GLP-1, ng L−1|
|Tmc||50.2 ± 31.1||75.3 ± 55.0||46.0 ± 34.7||50.6 ± 43.8||ER, ED|
|Peak||61.4 ± 29.7||94.3 ± 54.2||77.9 ± 40.5||95.7 ± 58.2||ER|
|T180||32.0 ± 9.6||40.6 ± 21.1||38.6 ± 15.4||52.0 ± 27.9||ER, ED|
|PYY, pmol L−1|
|Tmc||35.9 ± 15.0||37.5 ± 11.0||33.9 ± 14.9||32.4 ± 12.6|
|Peak||46.0 ± 17.2||50.8 ± 13.8||48.8 ± 18.6||51.2 ± 14.0|
|T180||40.0 ± 13.0||44.2 ± 13.5||41.6 ± 18.6||51.2 ± 14.0||ED|
|PP, pmol L−1|
|Tmc||90.8 ± 67.1||86.9 ± 63.3||104.4 ± 65.3||109.4 ± 69.4|
|Peak||98.7 ± 63.4||98.6 ± 67.7||108.4 ± 62.8||116.7 ± 69.4|
|T180||58.2 ± 40.81||41.4 ± 26.32||59.4 ± 42.81||65.1 ± 36.81||ER, ERxED|
|Total ghrelin (%Δ)d|
|Tmc||8 ± 22||−1 ± 19||−2 ± 14||−5 ± 13|
|Nadir||−11 ± 27||−13 ± 20||−11 ± 11||−18 ± 24|
|T180||10 ± 59||−3 ± 51||2 ± 22||6 ± 52|
|Active ghrelin (%Δ)e|
|Tmc||50 ± 222||3 ± 36||−2 ± 27||−14 ± 50|
|Nadir||−27 ± 55||−29 ± 26||−38 ± 26||−50 ± 36||ER|
|T180||74 ± 180||−5 ± 66||36 ± 159||60 ± 224|
Main effects of ER on peak GLP-1 concentrations, and Ghrelinact concentrations at nadir were observed with GLP-1 concentrations being higher (P effect ≤ 0.05) and Ghrelinact lower following FR meals (P effect ≤ 0.05, Table 3). Eating rate had no effect on the peak concentration of any other hormone or glucose. A main effect of ED on peak glucose and insulin concentrations was observed with peak concentrations of both being higher following the HED meals (P effect ≤ 0.05, Table 3). Energy density had no effect on the peak concentration of any other hormone.
Main effects of both ER and ED on insulin and GLP-1 concentrations at T180 were observed, with concentrations being higher during the FR (P effect ≤ 0.05) and HED trials (P effect ≤ 0.05, Table 3). PP concentrations were higher at T180 during the FR trials (P effect ≤ 0.05), which was attributable to the T180 concentration being lower during the SR-LED trial than during any other trial (P ≤ 0.01). PYY concentrations at T180 were higher during the HED trials (P effect ≤ 0.05). Glucose, Ghrelintot, and Ghrelinact concentrations at T180 were not affected by ER or ED.
Fasting SLIM ratings were lower before the SR-LED meal than before any other meal (P ≤ 0.05, Figure 2). Though an ER-by-ED interaction was observed for SLIM AUC, post-hoc comparisons did not reveal any significant differences in SLIM AUC across trials (Figure 2). Further, this interaction was not observed when adjusting the model for fasting SLIM rating suggesting that the observed ER-by-ED interaction was attributable to the low fasting SLIM rating before the SR-LED meal and not to any effects resulting from ER- or ED-manipulation.
SLIM ratings were higher on average, indicative of greater fullness, at meal completion and at peak (P effect ≤ 0.05) when meals were consumed at the fast rate (P effect ≤ 0.05, Figure 2). Fullness at completion of the breakfast meal and peak fullness were not; however, affected by ED. By T180 no differences in SLIM ratings were observed across trials.
Ad libitum energy intake at lunch and total energy intake
Energy intake during lunch was slightly lower on average [55 kcal (95% CI: 20 kcal, 89 kcal), P effect ≤ 0.05] following HED breakfast meals but was not affected by ER (Figure 3a). When total energy intake, defined as the sum of the energy consumed at the breakfast and lunch meals, was compared across trials, total energy intake during the FR-HED trial was greater than that observed during any other trial (P interaction ≤ 0.05, Figure 3b). The largest between-meal difference in total energy intake was observed when comparing the FR-HED trial to the SR-LED trial [348 kcal (99% CI: 174 kcal, 521 kcal), P ≤ 0.01].
Associations of energy intake and mass consumed with appetite, glucose and hormone responses
Energy intake during the breakfast meal was positively correlated with log10 insulin (r = 0.70), GLP-1 (r = 0.51) and PYY (r = 0.40) AUC (P ≤ 0.05). Similarly, the mass of food consumed during breakfast was positively correlated with log10 insulin (r = 0.45), GLP-1 (r = 0.43), PYY (r = 0.30), and PP (r = 0.34) AUC (P ≤ 0.05). The total mass (food + water) consumed at breakfast was not correlated with the AUC of any hormone, but was positively correlated with SLIM AUC (r = 0.25, P ≤ 0.05). Glucose, Ghrelintot and Ghrelinact AUC were not correlated with energy intake, the mass of food consumed or the total mass consumed at breakfast. SLIM AUC was not correlated with the AUC of glucose or any hormone.
Multiple regression analyses indicated that both log10 energy intake during breakfast ([β ± SEM] β = 1.6 ± 0.2, P < 0.001) and log10 food mass consumed during breakfast (β = −0.7 ± 0.2, P = 0.001) were associated with log10 insulin AUC. In contrast, energy intake but not the mass of food consumed during breakfast was significantly associated with PYY AUC (β = 10.3 ± 3.2, P = 0.002) and GLP-1 AUC (β = 6.6 ± 2.0, P = 0.002), and the mass of food consumed but not energy intake during breakfast was associated with PP AUC (β = 56.3 ± 17.3, P = 0.06). The total mass consumed during breakfast but not energy intake was associated with SLIM AUC (β = 3.6 ± 1.8, P = 0.05). Neither energy intake nor the mass of food consumed at breakfast were associated with Ghrelintot or Ghrelinact AUC in multiple regression analyses.
Energy density and ER are modifiable factors that influence energy intake; however, prior to this investigation, their combined impact on energy intake and resulting effects on pancreatic and gut hormone responses were undetermined. We report that a covert increase in the ED of a meal increased ad libitum energy intake and that this effect was potentiated by eating at a fast rate. Moreover, eating at a fast rate had a greater effect on energy intake when a HED relative to a LED meal was consumed. Despite altered pancreatic and gut hormone responses expected to promote satiety, appetite was largely unaffected and energy compensation at the subsequent meal incomplete. These findings demonstrate that the independent effects of ED and ER on energy intake are additive and provide additional support for a role of ED and ER in facilitating overeating.
Independent and additive effects of ED and ER manipulation on ad libitum energy intake were observed. The mass of food consumed at breakfast did not differ in response to ED manipulation but increased with increasing ER; therefore, both a covert increase in ED and eating at a fast rate independently resulted in greater energy intakes. This effect of ED on energy intake is consistent with previous studies demonstrating that covert changes in dietary ED do not affect the mass of food consumed (8-10) and energy intake changes in parallel with ED (24). The effects of ER manipulation on energy intake are less clear than those of ED. The reported effects of slowing ER on ad libitum energy intake are equivocal (12-14,25,26), with some studies demonstrating effect modification by sex (13) or habitual eating pattern (14). Our findings suggest that the ED of the meal may also be an effect modifier as the effect of ER on energy intake reached statistical significance only during the HED meals. In two recent trials, experimentally increasing habitual ER was shown to increase ad libitum energy intake (12, 14). We extend these findings by demonstrating that eating quickly potentiates the effect of ED on energy intake. To the best of our knowledge, this is the first study to demonstrate that eating at a fast rate compounds the effects of ED on energy intake during a single meal.
The independent and additive effects of ER and ED on intake affected pancreatic and gut hormone responses. Observed increases in insulin, GLP-1, PYY and PP AUC during the FR trials and, excepting PP, during the HED trials as well, are consistent with an effect of either mass or energy consumed on hormone responses. These findings are in general agreement with the observation that postprandial insulin (27), GLP-1 (28), PYY (29), and PP (30) concentrations increase in proportion to meal size. As a consequence of the experimental dissociation between the mass of food consumed and energy intake; however, we were able to explore whether these hormones were more sensitive to the mass of food or energy consumed with the potential confounding effects of macronutrient proportion, texture, taste and palatability held constant. Multivariate models demonstrated that insulin, GLP-1 and PYY were sensitive to energy intake whereas PP appeared more sensitive to the mass of food consumed.
Despite robust differences in intake that resulted in altered pancreatic and gut hormone responses, appetite was relatively unaffected by ED and ER manipulations. These findings are in contrast to experiments that have demonstrated that exogenous administration of ghrelin increases (31), and exogenous GLP-1 (32), PYY (33), and PP (34) suppress appetite. However, infusion protocols often produce supraphysiologic hormone concentrations (32, 35) and fail to account for the postprandial physiologic milieu in which gut hormones act. Moreover, associations between gut hormone concentrations and appetite have not been consistently observed when hormone responses are manipulated by dietary means (27, 36). A partial explanation may be that the effects of physiologic concentrations of gut hormones on appetite are masked by the numerous environmental, sensory and cognitive cues that can affect the desire to eat (3, 4). Regardless, the observation that subjective appetite more closely approximated the total mass of food and water consumed rather than energy intake suggests that candidate gut hormone satiety signals should closely approximate the total mass of food and liquid consumed at a given meal. The associations documented in this study suggest insulin and PP may be somewhat stronger candidates than PYY, GLP-1 and ghrelin. However, confirmation of these findings in controlled experiments measuring gut hormone responses following covert ED manipulation of mass- and macronutrient-matched meals are needed.
Energy intake was slightly reduced at lunch following the HED meals providing some evidence for energy compensation; nonetheless, this compensation was incomplete. Possibly, energy compensation occurred following lunch or during the subsequent day. However, incomplete energy compensation following ED manipulation is a consistent finding. Previous studies have shown that increasing the ED of a portion of the diet does not affect the mass of food adults consume thereby promoting increases in energy intake that persist over periods of up to 2 weeks resulting in weight gain (11). To our knowledge, effects of ER on long-term energy compensation are undetermined.
The finding that total energy intake over two meals was highest in the FR-HED condition but not different across the other conditions suggests that ED may be more important than ER in determining short-term energy intake. However, a recent report documented an inverse association between ED and the rate at which a variety of commonly consumed foods and beverages were habitually eaten suggesting that long-term, the effects of ED on energy intake may be mitigated by habitually consuming energy dense foods at a slower rate than low-ED foods (37). Nonetheless, Viskaal-van Dongen et al. also noted that both ER and ED were independently, positively associated with energy intake across a variety of foods (37). In conjunction with our results, these findings suggest that adopting an eating pattern that includes frequent consumption of energy dense foods at a fast rate may promote overeating. Conversely, reducing ER and consuming low-ED foods may deter overeating. In support, reducing dietary ED has been shown to decrease ER concomitant to reductions in energy intake (38). Further, faster habitual ER (15-17) and higher dietary ED (6) have been documented in overweight and obese compared to lean adults suggesting that interventions aimed at reducing dietary ED and ER may have robust effects in this population. Though the findings of this study are not generalizable to overweight and obese populations, our group has recently noted that effects of ER on appetite and gut hormone responses appear to be independent of BMI (21). In addition, the tendency to consume a consistent mass of food irrespective of energy content has been demonstrated in both normal-weight and obese adults (9, 10). As such, an important question to address in future research is whether the observed effects of ED and ER manipulation lead to changes in body mass over time.
Strengths of the study design include the covert manipulation of ED, constancy of macronutrient proportion and palatability of the meals, standardization of served portions, and the low dietary restraint and disinhibition of the volunteers. Though a viscous fiber was added to LED meals, it is unlikely that this appreciably affected the experimental outcomes as differences in total fiber intake between LED and HED meals were only 0.2 g/100 g food. It should be noted however that the laboratory environment was artificial, and effects due to deviations from normal eating behaviors such as habitual ER, food availability or eating schedules cannot be determined. Also, standardizing the rate rather than the mass of water consumed at breakfast may have differentially affected appetite across meals though previous studies have not documented effects of smaller amounts water intake on appetite or gut hormone responses (39, 40). Decreasing the frequency of blood sampling following the first 60 min of measurement may have lessened our ability to accurately detect peak PYY and nadir ghrelin concentrations. Moreover, timing appetite and hormone measurements relative to meal initiation rather than meal termination may have influenced results as, relative to the FR meals, a shorter postprandial interval was captured following completion of SR meals and frequently measurements occurred during the course of eating the SR meals rather than just after meal termination. Conversely, timing measurements relative to meal termination would have resulted in an overall longer measurement period for SR relative to FR meals. Finally, the use of a single scale to measure perceived appetite may be a limitation. Though the reliability and sensitivity of the SLIM scale are established (22), using the SLIM in conjunction with visual analog scales may have enabled us to capture different dimensions of appetite.
In summary, we demonstrate that consuming a high ED meal quickly compounds the independent effects of ED and ER on energy intake. Moreover, compensation at the following meal did not occur despite altered pancreatic and gut hormone responses expected to suppress appetite. These findings suggest that frequent consumption of energy dense foods, especially those that are typically consumed at a fast rate may promote excess energy intake. Given the ubiquitous presence of these foods in the modern food environment, adoption of behavioral and environmental strategies that limit the accessibility of these types of foods may be efficacious for healthy weight management.
The authors wish to acknowledge the study volunteers. The authors also wish to acknowledge Christina Carvey, Michael Stanger, Jay O'Hara, Ryan Regalia, Bryan Wiley, and Matthew Dickson who all made significant technical contributions, Matt Ely for data management assistance, and the staff at Pennington Biomedical Research Center responsible for performing the assays. Research funded by the US Army Medical Research and Materiel Command.