Effects of Meals High in Carbohydrate, Protein, and Fat on Ghrelin and Peptide YY Secretion in Prepubertal Children

Abstract

Context: Ghrelin and peptide YY (PYY) are two hormones produced by the gastrointestinal tract that have effects on appetite. However, little is known about their secretion in response to meals high in individual macronutrients in prepubertal children.

Objective: We sought to understand how meals high in carbohydrate, protein, and fat affect serum concentrations of total ghrelin and total PYY, hypothesizing that these macronutrients would exert differential effects on their secretion.

Design and Setting: This was a cross-sectional study at one tertiary care center.

Subjects: Subjects were 7- to 11-yr-old healthy normal-weight (NW) and obese (OB) volunteers recruited from local advertisements.

Interventions: After an overnight fast, the subjects were given a breakfast high in carbohydrate, protein, or fat at 0800 h. Blood samples for total ghrelin and total PYY were taken at baseline, 30 min, and hourly from 0900 to 1200 h.

Main Outcome Measure: We assessed postprandial ghrelin suppression and PYY elevation, as well as changes in reported hunger and satiety, after the three test meals.

Results: After the high-protein meal, ghrelin declined gradually in both groups over the study period without subsequent increase, whereas ghrelin suppressed more rapidly to a nadir at 60 min after the high-carbohydrate meal in both NW and OB children, followed by rebound in ghrelin levels. Similarly, after the high-protein meal, PYY concentrations increased steadily over the course of the morning in both groups without decline, whereas PYY levels peaked 30 min after the high-carbohydrate meal in both NW and OB subjects with significant decline thereafter. Ghrelin and PYY responses to the high-fat meal were somewhat intermediate between that observed with high carbohydrate and high protein. The OB children reported higher hunger and lower satiety after the high-carbohydrate meal compared to the NW subjects, whereas appetite ratings were similar between the groups after the high-protein and high-fat meals. Additionally, within the OB group, area under the curve (AUC) analysis revealed significantly greater PYY response, as well as lower AUC hunger and higher AUC satiety, to the high-protein meal than the high-carbohydrate and high-fat meals.

Conclusions: The patterns of secretion of ghrelin and PYY in our study of prepubertal children suggest that they may play a role in the effectiveness of high-protein/low-carbohydrate diets in promoting weight loss.

The macronutrient content of foods has differential effects on ghrelin and peptide YY secretion and appetite ratings in prepubertal children.

Childhood obesity is a major health problem in the United States and elsewhere (1). The many negative health consequences of obesity have led to efforts to better understand the physiological signals that regulate appetite and energy balance. Ghrelin and peptide YY (PYY) are two important hormones produced by the gut that affect hunger and satiety via their interactions with the central nervous system. Ghrelin is an appetite-stimulating hormone produced mainly by the oxyntic cells of the stomach (2,3). In adults, serum concentrations of ghrelin have been shown to rise during a period of fasting and then decline after consuming a meal (4,5). However, we (6) and others (7) have shown that ghrelin is not suppressed in prepubertal children after a mixed meal. PYY is an anorexigenic hormone secreted primarily by the endocrine L cells that line the distal small bowel and colon (3,8). Serum PYY levels have been shown in adult studies to increase after eating with a peak at 1–2 h (9,10). The PYY response to feeding is less well understood in young children, although we found normal-weight (NW) prepubertal children to have a similar pattern of PYY secretion as adults, with obese (OB) children having a more variable response (6).

Several studies in adults have evaluated the ghrelin response to meals high in individual macronutrients. In general, these have shown that a glucose tolerance test (11,12) or meals high in carbohydrate content (12,13,14,15,16,17) have the most suppressive effect on ghrelin, with the nadir occurring between 60 and 120 min, and rebound to baseline concentrations by 3–4 h. Most reports of high-protein meals show that ghrelin is suppressed less than after high-carbohydrate intake, but concentrations tend to remain below baseline for longer periods, resulting in a larger decremental area under the curve (AUC) (13,14,16,18,19,20), although some studies have found ghrelin to actually increase after a high-protein test meal (12,17). Reports of high-fat test meals have shown slightly more varied results but generally indicate a reduction in ghrelin concentrations from baseline and subsequent rebound that are between that of protein and carbohydrate (12,13,14,15), although one study showed no change (19) and another showed an increase in ghrelin levels (17). Less has been published about changes in PYY concentrations after individual macronutrients, but these data show fat intake to raise PYY levels the most, with lesser increases after carbohydrate and protein-rich meals (21,22).

Only sparse data exist on the effects of individual macronutrients on ghrelin secretion in children. These reports show that a glucose tolerance test results in suppression of ghrelin to a nadir at approximately 60 min (23,24). However, we are not aware of any reports that examine ghrelin responses to macronutrient-specific test meals or PYY responses to a macronutrient challenge in prepubertal children. The objective of our study was to assess the pattern of secretion of ghrelin and PYY after consumption of meals high in different macronutrients in NW and OB prepubertal children. We hypothesized that carbohydrate, protein, and fat would exert differential effects on their secretion.

Subjects and Methods

Subjects

Children were recruited by way of local advertisements. The study population consisted of 32 healthy subjects 7 to 11 yr of age, including 13 NW children [body mass index (BMI), 15th to 85th percentile for age and sex] and 19 OB children (BMI ≥95th percentile for age and sex). Children who completed one test meal were allowed to participate in others, but this was not required. We enrolled 9–10 NW and OB subjects (matched for sex and race within each phase of the study) for each test meal to achieve adequate power (see Statistical Analysis). All subjects had a complete history and physical exam done by a single pediatrician (J.P.L.) and were prepubertal by clinical assessment (Tanner stage 1 breast development in females, testicular size ≤3 ml in males). The children had no health problems and were not taking any medications. Clinically, obese subjects were not felt to have a secondary cause of their obesity, including genetic disorders (e.g. Prader-Willi syndrome) and endocrinopathies (e.g. hypothyroidism or GH deficiency). All parents/guardians gave informed consent, and subjects gave assent. The study was approved by the Institutional Review Board at the University of Kentucky.

Protocol

Subjects were asked to eat and drink and have similar levels of activity as they normally would for 2 d before testing. The children had the test meals on three separate days (assuming they chose to participate in multiple phases), with the second and third visit at least 2 wk later than the preceding visit. They were admitted to the General Clinical Research Center (GCRC) at the University of Kentucky on the day of the study at 0730 h, having fasted since midnight. Height and weight were determined without shoes and in light clothing. BMI, height, and weight percentiles and Z-scores were determined by Epi Info, version 3.5.1 (Centers for Disease Control and Prevention, Atlanta, GA). A venous cannula was placed by an experienced nurse for phlebotomy, and baseline serum glucose, insulin, and leptin were obtained at 0800 h. Serum concentrations of total ghrelin and total PYY were obtained at 0800, 0830, and then hourly from 0900 to 1200 h. Appetite was assessed at these same time points using two 100-mm visual analog scales modified from Barkeling et al. (25,26). The scale for hunger was anchored with the phrases “I am not hungry at all. I don’t want anything to eat.” at 0 mm, and “I am starving! I want to eat now!!” at 100 mm. The scale for satiety was anchored with the phrases “I am not full at all. I could eat a lot more.” at 0 mm, and “I am stuffed! I can’t eat anything else!!” at 100 mm.

The high-carbohydrate meal consisted of one Fruit Roll-up, one low-fat Pop-tart, one package of Sunkist fruit chews, and one juice box (430 kcal, 88% carbohydrate, 2% protein, 10% fat). The high-protein meal consisted of one whole wheat tortilla with six slices of chopped Healthy Choice oven-roasted turkey, one half cup melted shredded Kraft fat-free cheddar cheese, and one carton Mighty milk (440 kcal, 36% carbohydrate, 44% protein, 20% fat). The high-fat meal consisted of half a packet of Carnation instant breakfast mixed with 4 ounces of heavy whipping cream and chilled to milkshake consistency (417 kcal, 17% carbohydrate, 2% protein, 81% fat). The meal was given at 0800 h, immediately after baseline labs and appetite ratings were taken. The subjects had 30 min to finish the meal, and all subjects consumed the meal in its entirety. Afterward, they were not allowed to eat or drink until the end of the study. A parent or guardian was required to be present with the subject throughout the study period. The children were restricted from vigorous physical activity but were allowed to play in their room or about the GCRC between blood draws. Body composition was determined by dual-energy x-ray absorptiometry (Prodigy; GE Lunar, Madison, WI).

Assays

Serum glucose was determined immediately after collection by modified glucose oxidase method using a polarographic oxygen electrode (Beckman Coulter, Inc., Brea, CA). All other studies were collected on ice, centrifuged immediately at 4 C, and stored at −80 C until assayed. Serum total ghrelin, total PYY, insulin, and leptin were measured by RIA (Linco Research, Inc., St. Charles, MO). Total ghrelin RIA measures both octanoylated and des-octanoylated ghrelin and has intra- and interassay coefficients of variation of 3.3–10.0 and 14.7–17.8%, respectively. Total PYY RIA measures both PYY_1-36 and PYY_3-36 but not neuropeptide Y or pancreatic polypeptide and has intra- and interassay coefficients of variation of 2.9–9.4 and 5.5–8.5%, respectively. Intra- and interassay coefficients of variation for insulin RIA are 3.1–4.6 and 2.9–5.0%, respectively. Intra- and interassay coefficients of variation for leptin RIA are 3.4–8.3 and 3.0–6.2%, respectively. Conversion factors from metric to SI units are: ghrelin, pg/ml × 0.3 = pmol/liter; PYY, pg/ml × 0.25 = pmol/liter; and leptin, ng/ml × 1 = μg/liter. Homeostasis model of assessment of insulin resistance was determined using the formula [insulin (μU/ml) × glucose (mmol/liter)]/22.5 (27).

Statistical analysis

The primary outcome measure to be assessed was the postprandial change in ghrelin and PYY. Initial power analysis based on adult literature (11) found that a sample size of nine subjects per group per test meal would have 90% power to show a 20% decrease in ghrelin. Ghrelin and PYY levels were expressed as percentage of baseline values, whereas hunger and satiety ratings were expressed as millimeters change from baseline. Baseline characteristics of the two groups were compared by using two-sample t tests for means or Fisher’s exact test for proportions. Mean response was compared within groups across time and between groups at selected time points by using a mixed linear model for a repeated measures design with post hoc comparison of means based on least square means differences. AUC response represents the area enclosed by the preingestion baseline and the postingestion curve through 1200 h. Differences in AUC between the two groups were compared using two-sample t tests. To compare the AUC responses of the three meals within the same group, an ANOVA was done with post hoc comparison of means based on least square means differences. P < 0.05 was considered significant. Results are reported as mean ± se.

Results

Baseline characteristics

Clinical characteristics of the two groups of children are summarized in Table 1. Nine OB and 10 NW subjects completed the high-fat phase of the study, whereas there were 10 children per group in the high-carbohydrate and high-protein phases. The OB subjects had higher BMI and BMI Z-score than the NW subjects by design. Fasting ghrelin levels were significantly higher in the NW children than the OB, but fasting PYY concentrations did not differ between the groups.

Table 1.

Characteristics of subjects at baseline

	NW	OB	P value
n	13	19
Age (yr)	9.62 ± 0.42	9.68 ± 0.29	0.89
% Caucasian	77	68	0.71
% Male	77	79	0.93
Weight (kg)	32.9 ± 1.7	58.0 ± 3.8	<0.0001
Weight Z-score	0.26 ± 0.18	2.34 ± 0.15	<0.0001
Height (cm)	137.2 ± 2.5	142.6 ± 2.2	0.12
Height Z-score	0.15 ± 0.26	0.91 ± 0.27	0.06
BMI (kg/m2)	17.3 ± 0.4	27.9 ± 1.1	<0.0001
BMI Z-score	0.31 ± 0.15	2.26 ± 0.07	<0.0001
% Fat	18.1 ± 1.6	42.9 ± 1.4	<0.0001
Glucose (mg/dl)	90.8 ± 2.1	92.6 ± 1.5	0.48
Insulin (μU/ml)	6.20 ± 0.77	23.1 ± 4.5	0.002
HOMA	1.42 ± 0.19	5.74 ± 1.02	0.0005
Leptin (pg/ml)	3.52 ± 0.85	31.8 ± 4.0	<0.0001
Ghrelin (pg/ml)	835 ± 47	664 ± 37	0.007
PYY (pg/ml)	87.0 ± 7.9	111.1 ± 11.0	0.11

Data are expressed as mean ± se. 

High-carbohydrate meal

Ghrelin declined significantly from 0800 to 0900 h in both groups (Fig. 1A; P < 0.0001 NW; P = 0.008 OB) and then increased significantly from 0900 to 1200 h (P = 0.002 NW; P < 0.0001 OB). Comparison of ghrelin AUC revealed a trend toward a difference between the groups (−36% NW vs. −8% OB; P = 0.10). In both groups, hunger decreased significantly from baseline to 0830 h (Fig. 1B; P < 0.0001 NW and OB) and then increased from 0830 to 1200 h (P < 0.0001 NW and OB). The OB children reported significantly higher levels of hunger than the NW subjects at 1000 h (P = 0.004) and 1100 h (P = 0.007), although there was no difference in hunger AUC between the two groups.

Figure 1.

Mean ± se serum total ghrelin (A; mean absolute basal concentrations, 796 pg/ml NW, 561 pg/ml OB), hunger (B; mean basal values, 68 mm NW, 74 mm OB), total PYY (C; mean basal concentrations, 76 pg/ml NW, 123 pg/ml OB), and satiety (D; mean basal values, 27 mm NW, 38 mm OB) in NW (triangles and solid lines) and OB (squares and dashed lines) children after high-carbohydrate meal.

PYY increased significantly from 0800 to 0830 h in both the NW (Fig. 1C; P < 0.0001) and OB children (P = 0.04) and then declined from 0830 to 1200 h in both groups (P < 0.0001 NW and OB). However, AUC analysis showed a significant difference between the groups (96% NW vs. −20% OB; P = 0.01). Satiety increased from 0800 to 0830 h (Fig. 1D; P < 0.0001 NW; P = 0.0005 OB) and then decreased from 0830 to 1200 h (P < 0.0001 NW and OB) in both groups. As with hunger ratings, the OB children reported significantly lower satiety than the NW children at 1000 h (P = 0.008) and 1100 h (P = 0.002), and there was a difference in satiety AUC between the two groups (73% NW vs. −24% OB; P = 0.05).

High-protein meal

Ghrelin decreased significantly from 0800 to 1100 h (Fig. 2A; P < 0.0001) in the NW children but did not increase from 1100 to 1200 h (P = 0.26). In the OB children, ghrelin declined gradually from 0800 to its nadir at 1200 h (P < 0.0001). Ghrelin AUC was significantly lower in the NW subjects (−61%) than in the OB group (−28%; P = 0.02). As with the high-carbohydrate meal, hunger ratings declined significantly from baseline to 0830 h (Fig. 2B; P < 0.0001 NW and OB) and then increased from 0830 to 1200 h (P < 0.0001 NW and OB) in both groups. However, unlike the high-carbohydrate meal, the two groups reported similar levels of hunger at all time points. Additionally, hunger AUC was not significantly different between NW and OB children.

Figure 2.

Mean ± se serum total ghrelin (A; mean absolute basal concentrations, 924 pg/ml NW, 817 pg/ml OB), hunger (B; mean basal values, 82 mm NW, 85 mm OB), total PYY (C; mean basal concentrations, 101 pg/ml NW, 88 pg/ml OB), and satiety (D; mean basal values, 24 mm NW, 19 mm OB) in NW (triangles and solid lines) and OB (squares and dashed lines) children after high-protein meal.

PYY concentrations increased from 0800 to a maximum at 1200 h in the NW children (Fig. 2C; P < 0.0001). In the OB group, PYY peaked at 1000 h (P < 0.0001 vs. 0800 h) without subsequent decline from 1000 to 1200 h (P = 0.28). Unlike ghrelin, there was no difference in PYY AUC between the groups. In both NW and OB subjects, satiety increased from 0800 to 0830 h (Fig. 2D; P < 0.0001 NW and OB) and then decreased from 0830 to 1200 h (P < 0.0001 NW and OB). Similar to hunger ratings, levels of satiety were not significantly different between the groups at any time point, nor was satiety AUC different between the groups.

High-fat meal

In both groups, ghrelin declined significantly from 0800 to a nadir at 0900 h (Fig. 3A; P < 0.0001 NW; P = 0.003 OB). In the NW group, ghrelin did not increase between 0900 and 1200 h, although it did rise significantly in the OB group (P < 0.0001) over that timeframe. Additionally, ghrelin AUC was significantly lower in the NW children (−42%) than the OB subjects (−8%; P = 0.02). In both groups, hunger declined from 0800 to 0830 h (Fig. 3B; P < 0.0001 NW and OB) and then increased from 0830 to 1200 h (P < 0.0001 NW and OB). As with the high-protein meal, the hunger ratings were not different between the groups at any time point, nor was there a difference in hunger AUC.

Figure 3.

Mean ± se serum total ghrelin (A; mean absolute basal concentrations, 876 pg/ml NW, 737 pg/ml OB), hunger (B; mean basal values, 79 pg/ml NW, 71 mm OB), total PYY (C; mean basal concentrations, 88 pg/ml NW, 133 pg/ml OB), and satiety (D; mean basal values, 21 mm NW, 30 mm OB) in NW (triangles and solid lines) and OB (squares and dashed lines) children after high-fat meal.

PYY increased significantly in the NW children from 0800 to 0900 h (Fig. 3C; P < 0.0001) and in the OB children from 0800 to 0830 h (P = 0.006). Concentrations of PYY then declined from 0900 to 1200 h in the NW group (P = 0.03) and from 0830 to 1200 h in the OB group (P = 0.002). OB subjects had significantly lower PYY AUC than NW children (53 vs. 163%; P = 0.03). Levels of satiety increased from 0800 to 0830 h (P < 0.0001 NW; P = 0.0001 OB) and then declined from 0830 to 1200 h (P < 0.0001 NW and OB) with no differences observed between the groups in satiety ratings at any time point or in satiety AUC.

Comparisons between meals

AUC analysis within groups across the different test meals revealed no significant effect of macronutrient type on ghrelin suppression in either NW or OB children (Table 2). However, in the OB group, there was a significant effect on hunger, with protein leading to a greater reduction compared with carbohydrate (P = 0.006) and fat (P = 0.05), and there was a trend toward the same findings in the NW group (protein vs. carbohydrate, P = 0.04; protein vs. fat, P = 0.06).

Table 2.

Ghrelin, hunger, PYY, and satiety AUC comparisons within NW and OB groups to the three test meals

	Carbohydrate	Protein	Fat	P value
NW
Ghrelin (%)	−36 ± 13	−61 ± 10	−42 ± 8	F (2, 27) = 1.5; P = 0.24
Hunger (mm)	−63 ± 17a	−137 ± 27	−71 ± 27b	F (2, 27) = 2.8; P = 0.08
PYY (%)	96 ± 30	130 ± 35	163 ± 36	F (2, 27) = 1.0; P = 0.39
Satiety (mm)	73 ± 23	122 ± 26	97 ± 31	F (2, 27) = 0.8; P = 0.44
OB
Ghrelin (%)	−8 ± 10	−28 ± 7	−8 ± 10	F (2, 26) = 1.6; P = 0.22
Hunger (mm)	−22 ± 22c	−158 ± 42	−63 ± 30d	F (2, 26) = 4.7; P = 0.02
PYY (%)	−20 ± 27e	181 ± 34	53 ± 31f	F (2, 26) = 11.3; P = 0.0003
Satiety (mm)	−24 ± 41g	170 ± 44	65 ± 36h	F (2, 26) = 5.8; P = 0.008

Data are expressed as mean ± se from baseline. 

^aP = 0.04 vs. protein; 

^bP = 0.06 vs. protein; 

^cP = 0.006 vs. protein; 

^dP = 0.05 vs. protein; 

^eP < 0.0001 vs. protein; 

^fP = 0.007 vs. protein; 

^gP = 0.002 vs. protein; 

^hP = 0.08 vs. protein. 

In the OB (but not NW) children, macronutrient type had a significant effect on PYY, with more elevation after protein than fat (P = 0.007) or carbohydrate (P < 0.0001). Similarly for satiety, there was a significant effect in the OB group (but not the NW group) for protein compared with carbohydrate (P = 0.002) and a trend toward protein compared with fat (P = 0.08).

Discussion

There is considerable evidence that protein is the most satiating macronutrient (28,29,30). Additionally, in both adults and children, high-protein/low-carbohydrate diets have been shown to be more effective at causing weight loss, at least over the short term, than calorie-restricted or low-fat diets (31,32,33,34). The mechanism of this effect has not been fully elucidated, but differences in the pattern of secretion of appetite hormones, such as ghrelin (13,14,16,18,19,20) and cholecystokinin (16), as well as increased thermogenic effect of protein compared with carbohydrate and fat (29), may be involved. Several observations from our data suggest that the differential effects the three macronutrients have on gut hormone responses in adults may also occur in prepubertal children. First, the high-protein meal resulted in gradual decline in ghrelin concentrations in both groups over the course of the morning without rebound, whereas the high-carbohydrate meal resulted in more rapid suppression in both NW and OB children to a nadir at 60 min, followed by increases in ghrelin levels. Similarly, after the high-protein meal, PYY concentrations rose steadily over the study period in NW and OB subjects without decline, whereas PYY levels peaked at 30 min after the high-carbohydrate meal in both groups with significant decline thereafter. Finally, AUC analyses within the OB group showed several “advantages” of protein compared with carbohydrate and fat, including lower integrated hunger and greater PYY and satiety responses.

It is unclear why the high-protein meal was associated with these distinct effects on ghrelin and PYY secretion. One possibility may be related to rates of gastric emptying for the three macronutrients. Meal-related ghrelin suppression has been shown to require postgastric stimulation (35). Because carbohydrate exits the stomach faster than protein and fat, this could explain the more rapid suppression and then rebound of ghrelin with the high-carbohydrate meal compared with the gradual decline seen with the high-protein meal. Another factor could be variations in insulin secretion because insulin is known to suppress ghrelin (36,37). In both NW and OB subjects in our study, the high-carbohydrate meal induced the greatest insulin response, followed by high protein, then high fat (data not shown). However, for all three phases in both groups, the insulin peak occurred at 30–60 min, followed by gradual decline. Although this corresponds to PYY and (inversely) to ghrelin in the high-carbohydrate phase, it would not be an explanation for the patterns observed with high protein and fat. Similarly, a correlation between insulin and ghrelin secretion was not observed in a study of adults given different test meals high in carbohydrate, protein, and fat (13), although a more complex relationship between ghrelin and insulin-mediated glucose metabolism could be involved. Finally, a possible mechanism related to PYY may involve coregulation of PYY and the urea cycle enzyme N-acetylglutamate synthase (NAGS). NAGS manufactures the regulatory cofactor N-acetylglutamate for carbamoyl phosphate synthetase, the rate-limiting step of the urea cycle, and the NAGS gene is located on chromosome 17 in a head-to-head arrangement with the PYY gene (Fig. 4). Dietary protein is the primary source of nitrogen in humans and acts to up-regulate the enzymes of the urea cycle, including NAGS (38,39). A common phenomenon in head-to-head gene pairs is the coregulation of both genes from a shared promoter or regulatory element (40,41), and several lines of evidence suggest that the PYY and NAGS genes may be regulated in a coordinate manner. First, the gene arrangement and promoter region are conserved in both mouse and rat. Additionally, in both mice (42) and humans (43,44,45), the only tissue with significant levels of NAGS expression outside of the liver is small intestine, where PYY is expressed. Finally, we have observed elevated PYY concentrations in serum in patients with primary urea cycle defects in whom elevated ammonia leads to up-regulation of other genes in the cycle including NAGS (46). Thus, coregulation of PYY and NAGS may provide a mechanism to link the processing of excess nitrogen in the urea cycle with suppression of further nitrogen intake, resulting in the above effects on appetite. However, additional studies examining the molecular aspects of this arrangement are needed to confirm this hypothesis.

Figure 4.

Relationship of human PYY and NAGS genes on chromosome 17q21. Arrows indicate the direction of transcription and the currently reported transcriptional start sites.

Our data also suggest that there may be some differences between NW and OB children in both their gut hormone and appetite responses to different macronutrients. First and most striking was the difference between the two groups in reported hunger and satiety after the high-carbohydrate meal. At both 1000 and 1100 h, the OB subjects had significantly higher hunger and lower satiety than the NW subjects, which did not seem to be explained by the ghrelin and PYY profiles. Additionally, the OB group had lower satiety AUC than the NW group. This is in contrast to the high-protein and high-fat meals in the present study, as well as previous reports of children (6,26) and adolescents (47) given a mixed meal, where there were no differences in hunger ratings between NW and OB groups. Second, OB children had a higher ghrelin AUC response than NW subjects with the high-protein and -fat meals, a trend toward higher AUC response to the high-carbohydrate meal, and an increase in ghrelin after its nadir at 0900 with the high-fat meal that was not observed in the NW group. Finally, the OB children had lower PYY AUC than NW children to both the high-carbohydrate and high-fat meals but not the high-protein meal. Again, these discrepancies were not completely reflected in the hunger and satiety ratings. It is of interest that most of these “unfavorable” differences in the OB group occurred with the high-carbohydrate and high-fat meals as opposed to the high-protein meal. However, given that the gut hormone profiles did not correspond directly with appetite ratings, it is unclear what role these differences in ghrelin and PYY secretion between NW and OB subjects actually play in the pathogenesis of obesity. Additionally, it is likely that differences in many other physiological signals (including other gut hormones like pancreatic polypeptide, glucagon-like peptide 1, and cholecystokinin) contribute to variations in energy intake between NW and OB children.

We found that fasting metabolic parameters showed the expected differences between NW and OB children: specifically, the NW group had lower leptin, insulin, and homeostasis model of assessment of insulin resistance than the OB group (Table 1). Fasting ghrelin was significantly higher in the NW subjects than the OB subjects, which is consistent with other literature in children (6,47,48) and adults (11). Fasting PYY was not significantly different between the groups, but this may be due to the small number of subjects because we previously showed a positive correlation between fasting PYY and several markers of adiposity (6).

Limitations of our study include the relatively small sample size. Additionally, the high-protein meal had only 44% of calories from protein compared with 88% and 81% of calories, respectively, from carbohydrate and fat for those phases of the study. We were unable to create a higher protein meal that was palatable for the young children in this study. However, this percentage of protein is similar (14) or higher (18,19,20) than that used in other reports of high-protein test meals. Furthermore, our study did not address ghrelin and PYY receptor sensitivity. Little is known about differences in the sensitivity of the receptors for ghrelin and PYY between NW and OB subjects, which may be another explanation for observed differences in their secretion and the discrepancy with appetite ratings. Finally, we measured total ghrelin and total PYY, which may not reflect active hormone concentrations. However, in other reports where both total and active ghrelin (13,49,50) and PYY (51) were measured, similar changes after meal consumption were observed, suggesting that total ghrelin and PYY are good markers for the active hormone.

In conclusion, our data show that the macronutrient content of foods has differential effects on ghrelin and PYY secretion and appetite ratings in prepubertal children and are consistent with the observed effectiveness of high-protein/low-carbohydrate diets. Given the small number of subjects studied, our data certainly need to be confirmed. Further study to assess whether OB prepubertal children respond better to lifestyle changes that include higher protein and lower carbohydrate intake as a means of improving weight and BMI, as well as evaluating further the relationship between PYY and NAGS regulation, is warranted.