Usual breakfast eating habits affect response to breakfast skipping in overweight women

Ethics statement

This study was conducted according to the principles expressed in the Declaration of Helsinki. The study was approved by the Colorado Multiple Institutional Review Board. All patients provided written informed consent for the collection of samples and subsequent analysis.

Subjects

Healthy women of all ethnic groups, ages 25−40, with BMI 27−35 kg/m², without eating disorders, and who were either habitual breakfast eaters (Eaters) or breakfast skippers (Skippers) were recruited. Subjects were asked to identify themselves as Eaters or Skippers by answering the question: “How many days per week do you eat breakfast?” Eaters reported eating breakfast ≥ 5 days per week and Skippers reported eating breakfast ≤2 days per week. Women without menses were excluded, and all women reported no active dieting or intensive exercise training. Additionally, subjects were excluded if they were identified as having night eating syndrome (at least 25% of food intake is consumed after the evening meal and/or at least two episodes of nocturnal eating per week) (21).

Prestudy procedures

Prestudy procedures included measurement of height, weight, vital signs, and resting metabolic rate (RMR). After 30 min of quiet rest, RMR was measured using standard indirect calorimetry with the ventilated hood technique (Parvo Medics Truemax 2400, Salt Lake City, UT). Subjects remained awake, supine, lightly clothed in a thermoneutral (68−74°F) quiet room. The following equation was used to estimate total energy expenditure (TEE): TEE = measured RMR (kcals) × activity factor. A standard activity factor of 1.4 was used for all subjects. Prior to the first study visit day, all subjects completed a 3-day diet diary in order to obtain usual dietary intake patterns and confirm reported breakfast eating pattern (Eater or Skipper).

Study visits

Each subject underwent 2 study conditions (Breakfast – B and No Breakfast – NB) with the order of study conditions determined using a computer-generated randomization table. Four Skippers and three Eaters underwent the NB condition first, and six Eaters and five Skippers underwent B first. Subjects were not aware of the order of study visits until the first study day. The study visits were timed to occur during the follicular phase of the menstrual cycle with one month between visits (22). Subjects were asked not to do purposeful exercise in the 48 h prior to each study day. Each subject's total daily energy intake (EI) for study visits was estimated based upon TEE as described above. Subjects were provided with a standard dinner meal (35% of daily EI; 15% protein, 30% fat, 55% carbohydrate) to be eaten the night before each test day between 6 and 8 PM and were asked not to eat anything after 8 PM. On the morning of the test day, subjects presented for baseline measurements at 7:00 AM. Their height, weight and vital signs were measured and then they were asked to lie quietly for 30 min. Subsequently, RMR was measured at baseline, as described above, and subjects were then asked to rate hunger and satiety using visual analog scales (VAS). Hunger was rated on a 100-mm line preceded by the question, "How hungry do you feel right now?" and anchored by "not at all hungry" on the left and "extremely hungry" on the right. Satiety was rated by VAS with the question, “How full do you feel right now?” with the anchors "not at all" and "extremely (23). When the appetite ratings were complete, an intravenous catheter was inserted into an antecubital vein and blood was drawn for fasting glucose, insulin, free fatty acids (FFA), ghrelin, PYY (peptide YY), GLP-1 (glucagon-like peptide 1), and triglycerides (TG). When all baseline measurements were complete, subjects were provided with breakfast along with 250mL water on the B test day, or 250mL water alone on the NB test day. The breakfast meal contained 25% of total daily EI (15% protein, 30% fat, 55% carbohydrate). The content of the breakfast was wheat flakes plus milk, scrambled eggs, and orange juice. Time zero was the time that the subject began eating (or drinking) and subjects were asked to consume the breakfast (or water) within 20 min. Blood sampling was performed prior to and every 30 min after the lunch meal for 3 h. Subjects rated hunger and satiety using VAS hourly prior to lunch, every 30 min for 3 h following lunch, and hourly until the end of the study day. Indirect calorimetry was performed hourly throughout each testing day (Figure 1).

At 240 min, subjects were provided with a standard lunch containing 35% of daily EI (15% protein, 30% fat, 55% carbohydrate). The lunch meal was comprised of the same foods in the same amounts on both test days, and was made up of a sandwich, chips, and fruit. Subjects were asked to consume all contents of the lunch meal within 20 min.

At 540 min, subjects were provided with an ad libitum dinner meal and told to eat as much or as little as they needed to feel full. The dinner meal was comprised of pepperoni pizza, fettucine alfredo, vanilla yogurt and chocolate pudding and provided a total of 2,600 kcal (15% protein, 30% fat, 55% carbohydrate). After the meal was finished the subjects were allowed to leave. They were provided with a box of snack foods to take home and were told that if they wanted to eat anything that evening, they should eat only items contained in the box (and no items from their home). The box contained soda (diet and regular), grapes, plain M&M's, cashews, vanilla ice cream, pretzels, and potato chips. They returned the uneaten portion of snacks the following day, and the amount eaten both at dinner and from snack foods was determined by weigh-back methods.

Laboratory analyses

Blood samples were collected in EDTA-containing tubes, centrifuged, placed in aliquot tubes, and stored at −70 to −80^°C until analysis. All assays were run after both study days were complete for each subject. For GLP-1, 30ɥl of dipeptidyl peptidase IV inhibitor was added to the 4ml EDTA tube prior to collection. GLP-1 assays were performed with Alpco Diagnostics ELISA (43-GPTHU-E01). Insulin concentrations were measured using competitive radioimmunoassay (Millipore). Radioimmunoassay was used to analyze serum leptin (Millipore), serum PYY concentrations (Millipore Cat. #PYYT-66HK) and total serum ghrelin concentrations (Millipore Cat. #GHRT-89HK). All radioimmunoassays were performed with a Perkin Elmer Wallac Gamma counter using Maciel RIA-AID data reduction software. Assays for glucose, TG and FFA were performed on the Olympus AU400e Chemistry Analyzer (Beckman). Reagents were purchased from Beckman Coulter for glucose and TG and from WACO for FFA.

Statistical analysis

Sample size calculations were performed using SigmaPlot version 12 (San Jose, CA). Sample size was determined with the use of glucose and insulin data from a study of over- and underfeeding done by our group (24). Power was estimated for a range of possible differences for a paired t-test at the 0.05 level. A sample of 18 subjects would give us 80% power to detect a difference in insulin AUC of 3,000 µIU/mL × 180 min. Assuming a 25% dropout rate, a sample size of 24 women enrolled would provide about 18 subjects for analysis. Baseline characteristics of the two groups (Skippers vs. Eaters) and EI were analyzed with the use of Wilcoxon rank sum tests. Area under the curve (AUC) was calculated for all outcomes with the exception of energy intake (laboratory values, appetite ratings, energy expenditure, and respiratory quotient [RQ]) using the trapezoid method (25). For laboratory values, AUC for the 3-h period following lunch was used. For appetite ratings, energy expenditure and RQ, AUC was calculated for the prelunch period (time zero through 240 min) and postlunch period (270 through 540 min). Energy intake (dinner, evening snack, and total EI) and all AUC outcomes were modeled using mixed-model linear regression (SAS 9.3, SAS Institute, Cary, NC). Day, Skipper/Eater, and order were included as predictors in every model. The day by Skipper/Eater interaction was tested in every model. The interaction term was left in the final model in models where it was significant at the 0.05 alpha-level, but was removed from the final model if not significant. All models included random intercepts for subjects to account for the repeated measures on subjects. All data for all 18 subjects were included in analyses, with the exception of EI data for one subject in the Skipper group (due to failure to follow protocol in returning the snack box items) and glucose data for one subject in the Eater group (due to missing data).

Subjects and baseline characteristics

A total of 24 women were enrolled in the study. Two subjects completed one study day and then dropped out of the study due to scheduling conflicts, and 4 other subjects dropped out prior to the first study day, also due to scheduling conflicts. A total of 18 women completed the study, 9 Skippers and 9 Eaters. Subjects were 72% Caucasian, 22% African American, and 6% Asian. They were 88% Non-Hispanic and 22% Hispanic. There was no difference among groups in race or ethnicity, and no difference in age, BMI or RMR (Table 1).

Laboratory measures

There were significant interactions between day and breakfast pattern (Skipper/Eater) for insulin and FFA AUC, such that values were higher on the NB day for Eaters but similar on both days for Skippers (Figure 2 and Table 2). The AUC's for TG, GLP-1 and PYY were higher on the B day, whereas AUC was lower on the B day for glucose. Order effect was evaluated and was not found to be significant.

Appetite ratings

There were significant interactions between condition and breakfast pattern for prelunch hunger AUC and postlunch satiety AUC (Figure 3 and Table 3). Prelunch hunger AUC was higher on the NB day for Eaters as compared with Skippers. Overall prelunch hunger under both the B and NB conditions was higher in Eaters. Prelunch satiety under both conditions was lower in Eaters as compared to Skippers. Postlunch satiety AUC was similar on both days for Eaters, whereas it was higher on the B day for Skippers. There was no significant order effect.

Energy intake

EI for breakfast and lunch did not differ by day, per study design (Table 4). Dinner, snack, and total EI, as well as macronutrient composition, were examined with the use of the mixed model, and there were no interactions, differences between Eaters and Skippers, or order effect, so data presented are for the difference between B and NB only. Dinner, snack and total EI did not differ by day. Protein, carbohydrate and fiber intake were higher on the B day.

Energy expenditure

The prelunch EE AUC was higher on the B as compared with NB day (Table 5). Postlunch EE AUC showed a trend toward being higher on the B day, but did not reach significance (P = 0.051). There were no group differences, interactions, or order effect for EE.

Carbohydrate and fat oxidation

There was an interaction between prelunch RQ AUC and habitual breakfast pattern, such that the AUC was lower on the B day for Skippers as compared to Eaters (Table 5). Prelunch RQ AUC was higher overall on the B as compared with NB day. Postlunch RQ AUC was lower in Skippers as compared to Eaters under both conditions. There was a significant order effect for both prelunch and postlunch RQ AUC, with higher values for both if the NB day was first.