Ghrelin was negatively related to body fat percentage (r = -0.46, P < 0.05) and BMI (r = -0.18, P < 0.02), but not body weight (r = -0.16, P > 0.45). There were no significant body weight changes during the seven day observation periods (2)(data not shown, P > 0.40).

The composition of the treatment beverages and their contribution to daily food intake is listed in Table 1. Overall, macronutrient intake during the seven day observation periods was primarily determined by the composition of the treatment beverages (Table 2).

Average 24 hour energy expenditure (24EE; uncorrected activity log alone) was 13.9 ± 1.9 MJ/d compared to 12.6 ± 1.6 MJ/d for total energy expenditure (TEE; doubly labeled water), which is an average over-reporting of energy expenditure of 11%. Thus, our assumption that subjects would likely misreport energy expenditure and the values would require adjustment was valid.

The mean preprandial ghrelin concentrations during the last week of each treatment period were 2501.4 ± 438.0 pg·mL^-1 for Hi-CHO, 2869.5 ± 817.3 pg·mL^-1 for Hi-PRO, and 2688.2 ± 755.5 pg·mL^-1 for Hi-FAT (Figure 1). These values are higher than reported in similar investigations. This discrepancy is explained by the use of the Linco Research Total Ghrelin RIA kit, which produces values that are approximately 10-fold higher than the most commonly used kit (Phoenix Pharmaceuticals)9. In a side-by-side comparison, both kits have been found to be analytically acceptable despite the differences in values obtained 9. Furthermore, the ghrelin concentrations of at least two studies using the same kit were very similar to those we measured 1011. The within- and between-subject coefficients of variation for the two observation periods (seven days per period) were 12.9% and 23.0%, respectively.

Preprandial ghrelin was not influenced by treatment, 24EE, macronutrient composition, and selected (without treatment beverages) and total (including treatment beverages) energy intake (breakfast or entire day), or the interactions between these variables (previous or same day)(all P between 0.40 to 0.80). As a further test, we included energy intake for seven days prior to- and two days after each ghrelin value. None of these days were significant (all P between 0.40 to 0.90). Individual day and mean 24EE up to the prior 4 four days before each ghrelin measurement was also not significant (all P between 0.10 to 0.90).

Selected and total energy intake for the entire day were significantly influenced by treatment period (P < 0.02), Monday/Friday effect (P < 0.003), Sunday effect (P < 0.03), and 24EE (P < 0.008) (Table 3a). Classifying energy intake into the three macronutrients, the only macronutrient influenced by 24EE was total and selected carbohydrate intake (P < 0.03, and P < 0.02, respectively) (Table 3b). There was no significant effect of ghrelin on total or selected energy intake for breakfast or entire day (all P between 0.80 to 0.90).

The partial correlation between breakfast energy intake and ghrelin was 0.07. At 80% power, we could have detected a ghrelin effect if the true partial correlation was a small as 0.36. For powers of 90% and 95%, the true partial correlations would have had to be 0.40 and 0.43, respectively. The partial correlation between total energy intake and ghrelin was even lower than that with breakfast energy intake (r = 0.003). Note that, for a partial correlation of 0.40, ghrelin would have only been explaining about 16% (0.40²) of the variation in energy intake, still a relatively small percentage of explained variation for a hormone purported to exert a large influence on intake.

Twelve healthy, non-smoking men were recruited from the Beltsville, MD area to participate in this study (Table 4). All subjects were weight-stable, and not using any medications known to affect food intake, appetite or water balance. The John Hopkins Bloomberg School of Public Health Committee on Human Research approved the study protocol. Subjects provided written informed consent and received a medical evaluation by a physician that included measurement of blood pressure and analysis of fasting blood and urine samples to screen for presence of metabolic disease.

Voluntary food intake was studied continuously for 16 weeks, whereby subjects consumed only foods provided by the Human Studies Facility (HSF) at the Beltsville Human Nutrition Research Center (BHNRC). Subjects choose foods ad libitum from the menus, and could consume any part or all of a food item, then return the remaining portion to be weighed. BHNRC staff that came into contact with the subjects provided no guidance as to the quantities and/or types of food items chosen. During weekdays, subjects reported to the BHNRC in the morning to eat breakfast, pack selected food items for lunch, then return again in the evening for dinner. Any food taken from the HSF that was subsequently not eaten (all or partial quantities), was returned the next day, and weighed and recorded. On Friday evenings, subjects were provided with coolers packed with a large amount of food for weekend meals. The weekend coolers provided a wide variety of foods in excess quantities, and subjects were allowed to request additional food items be included. Weekend food could be consumed on either day as long as the subjects logged which day each food item was eaten. All uneaten weekend food was returned on Monday, and weighed and recorded. Although subjects were instructed to consume only food items provided by HSF, they were allowed free access to beverages including caloric, noncaloric and alcoholic beverages. Detailed records of the amount, composition and name brand of beverages was submitted daily. In addition to beverages provided on the menu (milk and juice), both regular and decaffeinated coffee and tea were available at meals.

Food items offered in the morning (breakfast and lunch) were presented in a cafeteria-style setting as three different rotating menus, each lasting seven days (Table 5). Some food items remained on all three menus (e.g. milk and orange juice). In the evening, breakfast and lunch items were also available. A typical dinner was presented cafeteria-style as one or two entrée selections with optional gravies or sauces, and a minimum of three vegetables and side dishes. A garden salad with a variety of additional toppings and dressings was also available. Fifteen different dinner menus were rotated daily (Table 5).

The goals of the menu design were to allow detection of macronutrient selection by offering a wide range of carbohydrate, fat- and/or protein-rich foods, and to provide a variety of commonly available foods typical of what many Americans eat. In a research setting it is impossible to duplicate the degree of food choice available in real life. However, more than 300 food items were used to develop menus for this study, and specific requests for food items were incorporated into the menus whenever possible.

After each subject selected his desired foods, he presented them to a staff member that recorded the identity and weight of each food item by hand and on a computer (combination of bar code recognition of the food item and hand-entering of the weight). Upon termination of feeding, each subject presented his tray to a staff member that weighed any uneaten food. The accuracy of the food item recording process was verified by comparing the information on the computer with the hand-entered logs. This verification procedure was followed daily, and repeated at the end of the study with all food records. Energy and macronutrient composition were determined by consultation with the USDA Nutrient Database for Standard Reference 29.

During the 16 weeks of ad libitum intake, subjects were randomly assigned to two of three treatments. Each treatment lasted 8 weeks with no break between the periods. The treatments consisted of a daily beverage that contained ≈ 2 MJ/day of predominantly carbohydrate (Hi-CHO), fat (Hi-FAT), or a combination of protein and carbohydrate (Hi-PRO) (Table 1). The daily beverage was divided into three equal portions, and subjects consumed them with each of the three primary meals. The protein drink was designed to provide half the daily Recommended Daily Allowance (RDA) 30 of protein, with the balance carbohydrate. The drinks were formulated using sucrose, heavy whipping cream, and egg white as the principle source of carbohydrate, fat, and protein, respectively. Water, fat free non-dairy creamer, and aspartame were used to provide volume, adjust texture and add sweetness. Cocoa was added to all drinks to provide a uniform taste and appearance. Subjects were blinded to the treatments and the three drinks were judged to be indistinguishable by a taste panel conducted in our laboratory.

Each morning for the last seven days of each treatment period, subjects reported to the laboratory after a 10–12 hr fast, provided a blood sample, then reported to the HSF to eat breakfast. Blood was collected in tubes containing EDTA, centrifuged, and stored at -80°C until analysis. Plasma ghrelin was analyzed using a commercially available radioimmunoassay kit (Total Ghrelin, Linco Research, Inc.). The intra- and interassay coefficients of variation (CV) were 5.6% and 7.3%, respectively.

Before breakfast and after voiding, body weight was determined weekly on an electronic balance to the nearest 0.01 kg. Body composition was measured by Dual-energy X-ray Absorptiometry (DEXA; QDR 4500, Hologic, Inc, Waltham, MA).

To "capture" daily variations in energy expenditure, we combined a self-reported activity log 31 and doubly labeled water measurements. Although doubly labeled water is the "gold standard" measure of free-living energy expenditure, its use is limited by the production of a single value that is assumed to represent average energy expenditure over the course of the dosing period (seven days in this study). This seven day value for energy expenditure is not useful to compare with daily variation in ghrelin and food intake (macronutrient composition and energy intake). Since self-reported measures of energy expenditure (that can provide a daily energy expenditure value) may be misreported by subjects 3233, we adjusted the daily numbers using doubly labeled water measurements (see below).

Twenty-four hour energy expenditure (24EE) was estimated using a daily recording log method, modified from Bouchard et al. 31. Briefly, subjects recorded their daily activities in a log every 15 min over the course of the last seven days of each treatment period. Activities were entered in as a number (1–9), corresponding to example activities listed in the log. Each activity assumed a pre-determined energy expenditure score, thus energy expenditure was calculated as time spent in that activity times the energy expenditure rate.

Total energy expenditure (TEE) was concurrently measured by the doubly labeled water method as described by Speakman 34, which provided an estimate of energy expenditure during the last seven days of each treatment period. Subjects reported to the BHNRC between 6:30 and 9:00 a.m., at which time they received an oral dose of H₂ ¹⁸O (0.16 g/kg body weight) and ²H₂O (0.30 g/kg body weight). Urine samples were collected immediately before the dose and on every morning (second void) for the last seven days of the treatment period. The first sample was collected approximately 24 hr after the dose. Enrichments of ²H and ¹⁸O in urine samples were measured by infrared spectroscopy and isotope ratio mass spectrometry, respectively. TEE was calculated using the equations of Weir 35.

Individual daily 24EE values were corrected using the ratio adjustment (notation denoting subjects is suppressed),

24EE_{dayx, corrected} = 24EE_dayx × (TEE/24EE_{day 1–7}), where

24EE_dayx is the uncorrected daily energy expenditure value from the activity log for one of the seven days (day X),

TEE is the daily mean energy expenditure estimate using doubly labeled water. Represents a single value during the seven days of measurement (of which 24EE_dayx is one), and

24EE_{day 1–7} is the mean of the seven days of uncorrected 24EE values corresponding to TEE, of which 24EE_dayx is one.

To simplify the notation, the 24EE_{dayx, corrected} value for day X will subsequently be referred to as 24EE.

To check the assumption of homogeneous variances necessary for valid F-tests and correct P-values, we used the standard technique of plotting the standard deviations (SD's) against the means for selected energy intake, grouping observations by subject and treatment period. The results of this scatter plot revealed a strong positive linear relationship (r = 0.67, P < 0.001). The relationships between the SD and mean for macronutrient and energy intake (total and selected), and 24EE were also positive and significant. This indicated that the SD's (variances) were a function of the mean and that the data needed to be transformed. We followed methods described by Draper and Smith 36, and used a family of transformations based on logarithms. For selected energy intake, this transformation was log (b₀ + b₁y_i), where b₀ and b₁ are the estimated coefficients of the line fit by regressing the SDs on the means, and y_i represents the energy intake data. The other variables were transformed using this same family of transformations. This procedure resulted in homogeneous variances for all variables once transformed, satisfying ANOVA assumptions. We present the data on the original scale in tables and figures for ease of interpretation (unless indicated otherwise).

Due to the free-living nature of the subjects, there were three observations (of 168) where (for unknown reasons) a subject's food intake differed greatly from habitual intake due to a skipped meal or meals with low energy intake. For this reason, these observations were not used in the analyses. Additionally, a preliminary sensitivity analysis and residual diagnostics (e.g., restricted likelihood distance, Cook's D; optional output of Proc Mixed, new in version 9.1, in 37) suggested they were outliers.

The experimental design was an incomplete block crossover design, with two of the three drink treatments given sequentially to each subject. Data were analyzed in the mixed linear models framework, using the Proc Mixed procedure in SAS (version 9.1)37. Subject-to-subject variation was modelled as a random effect. Repeatedly measuring each subject over the seven days induced an autoregressive covariance structure we modelled as AR(1). Other design effects we retained in our modelling were a two level period effect ((first 8 week treatment period (1) vs. the second 8 week period (2)), and two day-of-the-week variables, found in a preliminary analysis to account for day-of-the-week effects. Each of these day-of-the-week variables classify days into two groups: (1) Sunday (0 vs. 1 for other days of the week) and (2) Monday/Friday (0 vs. 1 for other days of the week). They allow for the major differences in food intake and energy expenditure due to day-of-the-week effects. Some subject-specific variables, such as body weight, were included as covariates as appropriate. The treatment effects (Hi-CHO, Hi-PRO, and Hi-FAT) were included in all models.

For models predicting ghrelin concentration, we included 24EE, energy intake, and the interaction between 24EE and energy intake. We also considered prior day (up to 7 days) and subsequent day (up to 2 days) values for energy intake and ghrelin, and their interactions as candidate covariates. Values for up to 4 prior days for 24EE were used to predict ghrelin. For models predicting daily energy intake, we included preprandial ghrelin concentrations, 24EE, the interaction between ghrelin and 24EE, and additionally considered as candidate covariates the prior (seven days) and subsequent (two days) days for these two variables and their interactions. We explored models that included other variables and interactions, but none of those variables appeared useful. Data are presented as total intake (intake including treatment drinks) and/or selected intake (intake without treatment drinks). Values are presented as means ± SD unless indicated otherwise.

Since a preliminary analysis suggested that the effect of ghrelin on energy intake was small or negligible, we conducted a power analysis to determine our ability to detect an effect of ghrelin if the effect was small. This was accomplished by Monte-Carlo simulation (creating simulated data sets based on the data we collected) and, starting with no effect of ghrelin (a true coefficient of zero for ghrelin in a regression context), determining how large the true coefficient needed to be to obtain significance for most of the simulations, at powers of 80%, 90%, and 95%, with 1000 simulations for each coefficient value. These results are most easily interpreted as how large a partial correlation between ghrelin and energy intake (adjusting for all other fixed and random effects, other than ghrelin) would be necessary for us to detect it. We conducted this analysis for both total energy intake and breakfast energy intake (the latter was the meal most likely to be influenced by preprandial ghrelin because of the timing of the blood draw).