Morning ghrelin concentrations are not affected by short-term overfeeding and do not predict ad libitum food intake in humans

Abstract

Background: Ghrelin has a short-term orexigenic effect but may also be a marker of food intake over time. We previously found an inverse association between ghrelin concentrations and food intake.

Objectives: The objectives were to determine whether the fasting plasma ghrelin concentration is related to food intake and whether the previous day's intake predicts the suppression of ghrelin.

Design: Sixty-nine nondiabetic adults (40 men) aged 33 ± 9 y were studied as inpatients at a Clinical Research Center. After 6 d of consuming a maintenance diet, the subjects self-selected their food from our vending machine system for 3 d. Total plasma ghrelin concentrations were measured every morning during the vending machine period.

Results: The fasting ghrelin concentration was negatively correlated with body mass index (r = −0.31, P = 0.016) and weight (r = −0.26, P = 0.044). Mean morning ghrelin concentrations remained constant (149 ± 59, 152 ± 60, 148 ± 61, and 145 ± 59 pg/mL on days 1, 2, 3, and 4, respectively) even though the subjects overate while using the vending machines (160 ± 42% of weight-maintenance needs). No associations were found between daily ghrelin concentrations and subsequent food intake on any day (day 1: r = −0.04, P = 0.76; day 2: r = −0.01, P = 0.95; day 3: r = −0.11, P = 0.38). Suppression of total ghrelin concentrations was not associated with the previous day's intake or with subsequent food intake.

Conclusion: Morning plasma ghrelin concentrations do not affect acute increases in food intake. This trial was registered at clinicaltrials.gov as NCT00342732.

INTRODUCTION

Obesity is a growing problem throughout the world and is associated with a myriad of health-related problems. The development of obesity is most certainly multifactorial, resulting from environmental changes (decreased activity and increased food intake) and genetic predispositions. Whereas much research has been devoted to changes in physical activity (1–3), it has been more difficult to assess food intake and changes in food intake over time, mostly because of an insufficient means of accurately measuring food intake (4–6).

Ghrelin is reported to be an orexigenic hormone secreted primarily by the stomach and the duodenum (7). The hormone was first identified as an endogenous ligand of the growth hormone secretagogue receptor (8) but has since been implicated in a wide array of biological functions (9). A putative role for ghrelin in the regulation of gastric secretions and emptying (10), insulin secretion (11), glucose metabolism (12), energy expenditure (13), and food intake and energy homeostasis (14, 15) has been described.

Evidence in humans indicates that plasma ghrelin acts as a short-term hunger signal. Ghrelin has been shown to peak before meal initiation (16) and to decrease postprandially in proportion to the caloric content of the meal (17, 18). Furthermore, ghrelin concentrations are lower in obese than in lean individuals (19) and are suppressed to a lesser degree in obese subjects after meals (20).

Despite the consistent finding of elevated ghrelin concentrations before meal initiation, there is little evidence that elevated fasting ghrelin concentrations as representative of elevated overall 24-h ghrelin concentrations or hyperghrelinemia are related to food intake and subsequent weight gain. In fact, a small study from our group found that, independent of body weight (21), the fasting plasma ghrelin concentration is a significant negative predictor of ad libitum energy intake. The purpose of this study, therefore, was to determine in a larger cohort whether fasting ghrelin concentrations predict food intake from a cafeteria style diet and whether ghrelin concentrations responded to or predicted the subsequent days' intake during a 3-d period of ad libitum feeding.

SUBJECTS AND METHODS

Subjects

A total of 155 subjects were enrolled in this study to characterize a food intake phenotype; however, plasma was only drawn on a daily basis during ad libitum eating from a subset of subjects. As a result, sufficient plasma for measuring ghrelin concentrations was only available in 69 of these individuals (29 women, 40 men). Of the 69 subjects, 58 had 4 available morning plasma ghrelin samples drawn during the ad libitum feeding period, whereas 11 subjects were missing ≥1 sample (total sample count for each time point: day 1, 61; day 2, 65; day 3, 65; and day 4, 66). Also, of the 69 study subjects, 17 were white and 52 were Pima Indians. All volunteers were fully informed of the purpose of the study and written informed consent was obtained. All subjects were free of disease as determined by a physical examination, medical history, and laboratory testing. The experimental protocol was approved by the Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases.

Protocol

The participants were admitted to the metabolic ward and were immediately started on a weight-maintaining diet (20% protein, 30% fat, 50% carbohydrate). Weight-maintenance energy needs (WMEN) were based on equations developed on the ward: WMEN for men = [9.5 × weight (kg)] + 1973; WMEN for women = [9.5 × weight (kg)] + 1745) (22). On the second day after admission, body composition was assessed by using dual-energy X-ray absorptiometry (Lunar Corp, Madison, WI) as previously described (23). Glucose tolerance was assessed with a 75-g oral-glucose-tolerance test according to the criteria of the World Health Organization (24), and only nondiabetic subjects were included for study participation.

After 6 d of consuming the weight-maintaining diet, the subjects were allowed to eat ad libitum from our computerized vending machine system for 3 d, as previously described (25). On the first day of admission, the subjects were asked to complete a food-preferences questionnaire that listed 80 items in random order and were asked to assign each food a hedonic rating using a 9-point Likert scale with 1 = dislike extremely and 9 = like extremely (26). On the basis of the responses to the food-preference questionnaire, 40 food items were made available to the subjects each day in the vending machines: 8 breakfast items, 22 lunch or dinner items, and 10 snack items. The subjects were given only foods that they rated from 4 to 8 on a Likert scale on the food-preferences questionnaire; half of the items were low in fat and half were high in fat. At breakfast, 2 items were high in complex carbohydrates, 1 item was high in protein, and 1 item was high in simple sugar. Ten lunch or dinner items were high in complex carbohydrates and 12 were high in protein. Four snack items were high in complex carbohydrates and 6 were high in simple sugar.

In addition, a core group of condiments was provided each day, including butter, peanut butter, cream cheese, and jams; salad items and dressings; crackers, bread, tortillas, and Indian fry bread; spices and salsa; and, orange juice, apple juice, milk and a 6-pack of soda of the subject's choice. Each subject had a single, refrigerated vending machine assigned and had access to the machine for 23.5 each day. Subjects ate in a separate dining area that had no television. Food selection was monitored by the computerized vending system, and food wrappers and unconsumed food were returned to the machine to be weighed. Energy and macronutrient intakes during the ad libitum feeding period were calculated from the CBORD Professional Diet Analyzer Program (CBORD Inc, Ithaca, NY) as previously described (25).

Blood samples were drawn at ≈ 0530 on the morning of the first day of vending machine usage (day 7) and at the same time each morning thereafter until discharge (day 10) for a total of 4 times. Because of the unrestricted access to the vending machines, only the first blood draw (day 7) can be considered to have been taken after a true overnight fast for all subjects (see Results for further information). Plasma ghrelin concentrations were measured with a commercial radioimmunoassay (Phoenix Pharmaceuticals Inc, Belmont, CA) as previously described (19). Plasma samples were analyzed in duplicate, and concentrations are expressed in pg/mL. The intraassay CV for the ghrelin radioimmunoassay ranged from 5% to 7%, and the interassay CV was 12–15%.

Data analysis and statistics

All data are presented as means ± SDs unless stated otherwise and were analyzed by using SAS version 8.02 (SAS Institute, Cary, NC); a P value <0.05 was considered as significant. Any direct comparisons between groups were made by Student's t test. Plasma ghrelin concentrations and mean intakes were compared across days with a repeated-measures analysis of variance. Values were adjusted for confounders by using general linear regression models. Associations between variables were assessed by using Spearman correlation coefficients. With 69 individuals, we had a >90% power to detect a modest 0.4 association at an α of <0.05.

RESULTS

Age, body weight, body mass index (BMI), and percentage body fat for all subjects are presented in Table 1. Twenty-seven of 69 subjects had at least one episode of nighttime eating (defined as at least one occasion of food intake between 2300 and 0500) during the ad libitum feeding period. This number was consistent with that found in the larger cohort, in which nighttime eaters have been found to eat more and gain more weight over time (27). Nighttime eaters and nonnighttime eaters did not differ significantly in age, body weight, BMI, or percentage body fat. In addition, exclusion of nighttime eaters from the analysis did not change the results.

TABLE 1.

Subject characteristics

	All (n = 69)	Women (n = 29)	Men (n = 40)	P value1
Age (y)	33.0 ± 9.42	33.2 ± 10.4	32.9 ± 8.7	0.875
Weight (kg)	95.6 ± 23.2	93.9 ± 25.5	96.8 ± 21.6	0.625
BMI (kg/m2)	33.4 ± 8.0	35.8 ± 9.4	31.8 ± 6.5	0.055
Body fat (%)	38.8 ± 10.0	47.2 ± 6.6	32.9 ± 7.4	<0.0001
Glucose (mg/dL)	92.0 ± 9.3	91.5 ± 8.7	92.4 ± 9.9	0.725
Log insulin	1.55 ± 0.13	1.56 ± 0.13	1.54 ± 0.13	0.502

¹Differences between men and women were analyzed by using Student's t test.

²Mean ± SD (all such values).

Total energy intake did not differ by day (P = 0.23) and was 4648 ± 1346, 4389 ± 1465, and 4439 ± 1572 kcal for days 1, 2, and 3, respectively, with the vending machines; the mean intake over 3 d was 4492 ± 1246 kcal. The mean energy intake over the entire vending machine period, as a percentage of weight-maintenance requirements, was 160 ± 42%. The mean ad libitum intake over the 3 d did not differ significantly by race.

Mean morning plasma ghrelin concentrations were unchanged on days 1, 2, 3, and 4: 149.2 ± 58.9, 152.2 ± 59.7, 147.8 ± 60.8, and 145.2 ± 59.1 pg/mL, respectively. The time of the last eating episode did not correlate significantly with the morning ghrelin concentration.

As expected, the fasting ghrelin concentration on day 1 was significantly negatively correlated with BMI before and after adjustment for race and sex (r = −0.31, P = 0.02; r = −0.301, P = 0.02, respectively). Additionally, the ghrelin concentration was correlated with body weight (unadjusted: r = −0.259, P = 0.04; adjusted for race and sex: r = −0.317, P = 0.01) and percentage body fat (unadjusted: r = −0.263, P = 0.04; adjusted for race and sex: P = 0.14). The fasting total ghrelin concentration on day 1 was also negatively correlated with the fasting log insulin concentration measured before the 3-h oral-glucose-tolerance test, before (r = −0.527, P = 0.0005) and after (r = −0.424, P = 0.039) adjustment for sex and race. Adjustment of the plasma ghrelin concentration for percentage body fat also did not alter the results.

As shown in Figure 1, we found no significant correlation between unadjusted morning ghrelin values and subsequent food intake (day 1: r = −0.04, P = 0.76; day 2: r = 0.01, P = 0.95; day 3: r = −0.11, P = 0.38). Adjustment of ghrelin values for age, sex, and race did not alter the findings (day 1: r = −0.074, P = 0.57; day 2: r = 0.013, P = 0.92; day 3: r = −0.129, P = 0.31), nor did the inclusion of BMI or percentage body fat in the model.

FIGURE 1.

Associations between morning total ghrelin concentrations and subsequent ad libitum food intake on the following day, as assessed by Spearman correlations. The associations were not significant (day 1: r = −0.040, P = 0.76, n = 61; day 2: r = −0.008, P = 0.95, n = 66; day 3: r = −0.110, P = 0.38, n = 65), even after ghrelin concentrations were adjusted for age, sex, race, and BMI. Data shown are unadjusted values.

We assessed the suppression of ghrelin change (%) in morning total ghrelin values compared with day 1. As shown in Figure 2, mean morning ghrelin concentrations did not differ from day to day, although there was variability between subjects. The mean ± SD values for ghrelin suppression were 6.9% ± 41.3%, 1.3% ± 24.3%, and 0.3% ± 38.1% for days 2, 3, and 4, respectively. Daily ad libitum food intake during the vending machine period did not predict change in fasting ghrelin concentration the following day (Figure 3). Adjustment of ghrelin values for age, sex, race, and BMI or percentage body fat did not change any of the relations. Additionally, ghrelin suppression did not predict subsequent intake on day 2 (r = 0.03, P = 0.85) or day 3 (r = −0.11, P = 0.38), and adjustment of ghrelin for age, race, sex, and BMI or percentage body fat did not change the outcome.

FIGURE 2.

Suppression of ghrelin concentrations from day 1, calculated as a percentage of the baseline (day 1) value. Values were available for 60 subjects for both the day 2 and day 3 calculations and for 59 subjects for the day 4 calculations. Daily ghrelin suppression was not significant (repeated-measures ANOVA, P = 0.26). Positive values indicate an increased plasma ghrelin concentration compared with day 1.

FIGURE 3.

Associations between total daily ad libitum food intake and suppression of morning ghrelin concentrations on the following day. Data were analyzed by using Spearman correlations. No significant relations were found, regardless of whether ghrelin suppression was adjusted for sex, race, age, and BMI. Data shown are unadjusted values. Positive values indicate an increased plasma ghrelin concentration compared with day 1.

DISCUSSION

In our study paradigm for ad libitum feeding, we found that the total ghrelin concentration in the morning did not predict food intake throughout the day. Furthermore, a failure to suppress morning ghrelin concentrations, in relation to a baseline value, does not predict overall food intake. Indeed, 24-h food intake, even in the face of overeating, has no effect on morning ghrelin concentrations or suppression of those concentrations over a 3-d period.

Our group of study subjects did show the expected negative relation between fasting plasma ghrelin concentration and measures of body composition. This has been repeatedly reported in the literature (19, 21, 28). Moreover, we found ghrelin concentrations to be negatively associated with fasting plasma insulin, consistent with previous findings (19, 21, 29). Our main outcome regarding the association between fasting ghrelin concentration and food intake was negative and somewhat unexpected. We feel, however, that this was not a result of our study population being exceedingly different from the general population because we are able to reproduce the associations between ghrelin concentrations and body composition and insulin concentrations.

In lean individuals, ghrelin concentrations increase 1–2 h before meal initiation (16, 30) and subsequently decrease postprandially (16, 18) with a gradual return to preprandial concentrations. Additionally, peripheral infusion of ghrelin results in increased food intake in lean subjects (15, 31). Furthermore, ghrelin concentrations are higher in individuals with Prader-Willi syndrome—a genetic form of obesity with severe hyperphagia (32). Taken together, these observations have been interpreted to indicate that ghrelin, in the short term, is an orexigenic hormone that has a role in meal initiation. Paradoxically, however, ghrelin has also been implicated in long-term energy balance. Lean individuals actually have higher fasting plasma ghrelin concentrations than do obese individuals (19, 20). Pima Indians have lower ghrelin concentrations than do whites, and ghrelin concentrations are higher in individuals with anorexia than in controls. These data indicate that ghrelin concentrations may be related to food intake over the longer term.

If fasting ghrelin concentrations, which correlate well with the area-under-the-curve of 24-h ghrelin (16), are related to the development of obesity through increased food intake, one would expect a positive relation between the 2 variables. Yet previous work (21) in a smaller group of 30 individuals showed that fasting plasma ghrelin concentrations during a weight-maintenance period were actually negatively correlated with food intake during a subsequent 3-d period of ad libitum intake. In this study, we found no correlation between fasting ghrelin concentrations and food intake in a larger group. In fact, despite a mean increase in food intake of ≈160% of weight-maintenance energy requirements over 3 d during the ad libitum diet period, we saw no change in morning ghrelin concentrations (or the suppression of ghrelin). These results indicate that the role of ghrelin in short-term food intake and energy balance remains uncertain.

Our findings were not wholly unexpected, however. Previous work by Ravussin et al (33) indicated that fasting ghrelin concentrations were not related to body weight change in the face of over- or underfeeding. In addition, whereas ghrelin concentrations in that study tended to decrease with overfeeding and increase with underfeeding, no significant relations were found. Moreover, Bunt et al (28), in a prospective study in children, found that fasting ghrelin concentrations, adjusted for baseline weight or BMI, did not predict future growth rates.

One possible explanation for our findings was that, on average, our study subjects were obese, although subjects had a wide range of BMIs (18.5–55.8). It was previously reported that ghrelin concentrations in obese individuals respond differently to meal challenges than do those in lean individuals. For example, a low-dose ghrelin infusion was shown to increase food intake only in lean but not in obese subjects (31). Furthermore, the suppression of ghrelin concentrations after a meal has been shown to be impaired in obese subjects (20). It is possible, therefore, that obese individuals are either insensitive to the effects of ghrelin and/or have ghrelin resistance (21, 31) and cannot be expected to suppress morning ghrelin concentrations in response to overeating. Additionally, it may be that the “cafeteria effect” during the ad libitum phase of the study masked any relations that may exist between ghrelin and short-term food intake. Yet, in other studies in which a “cafeteria effect” would be expected to exist, calorie intake during a free-choice buffet after peripheral ghrelin administration did increase food intake compared with saline (15), which indicated that a high food intake is not likely the reason for the lack of association. The macronutrient content of the diet may also play a role in plasma ghrelin concentrations (34), but our study participants were all eating a similar mixed diet before the ad libitum period began; therefore, this was probably not a factor in our results. Last, as recently suggested (29), despite the evidence cited above, there may not be as strong a causal link between ghrelin and feeding in humans and that factors such as habitual meal patterns (entrainment) are stronger predictors of 24-h ghrelin patterns.

Because ghrelin has been shown to increase 1–2 h before eating (16), it is possible that the timing of the ad libitum breakfast meals may affect ghrelin concentrations. Whereas some subjects ate by 0600, most subjects had their first meal completed by 0900. It is possible, of course, that we did not measure ghrelin at its potential peak, yet we do believe that we measured it well above the nadir of ghrelin values. Overall, however, we believe that the impact on our results would be minimal. One of our main analytic outcomes measured was whether increased food intake suppressed ghrelin over 3 d. Because our ghrelin measurements took place at the same time every day, we should have been able to see a suppression if the effect had been large enough.

Our study had several limitations that should be addressed. First, we reported total rather than active ghrelin, and there is some debate as to which of these is the most appropriate form of the peptide to measure. Yet, a recent report showed that, in terms of short-term suppression through food intake, total and active ghrelin behave in a similar manner (34). Additionally, some individuals had at least one episode of nighttime eating while using the vending machine system; therefore, their morning plasma ghrelin samples might not actually have been “fasting.” There were, however, no differences in morning ghrelin concentrations between the nighttime and nonnighttime eaters, despite a difference in 3-d mean energy intake. Furthermore, we combined the results of both whites and Pima Indians, who have previously been shown to have different mean fasting plasma ghrelin concentrations (19). However, despite finding differences in fasting ghrelin values by race, energy intake during the vending machine period did not differ between the groups. Another limitation of the study was that daily blood samples were not collected from all subjects in this study because the decision to draw blood samples daily was added as an amendment to the protocol after we had already begun recruiting and studying subjects. A t test comparison between the subset of subjects who had blood samples drawn daily and the remainder of the subjects who did not showed no significant differences in age, weight, BMI, percentage body fat, and intake during the vending machine period. Last, our associations might have been significant if the sample size had been larger; however, the biological relevance of such weak associations is difficult to interpret.

In conclusion, we found no relation between ad libitum energy intake during food consumption from a vending machine system and either total morning ghrelin concentrations or the suppression of morning ghrelin concentrations over a 3-d period. We believe that these data indicate that ghrelin alone, at least over an acute period of ad libitum intake, is not responsible for increases in food intake.