Introduction

Both obesity and its associated metabolic disorders have become a global health burden in recent years [1]. Excessive calorie intake and macronutrient composition of consumed food are major factors in the development of this obesity pandemic [2–5]. Therefore, investigation into the quantity and food composition an individual consumes and why may assist in developing a more comprehensive understanding of how metabolic disorders develop.

The amount of food we consume is regulated by a balance of orexigenic and anorexigenic influences on areas of the central nervous system [6]. This balance is determined by a large number of interneuronal and endocrine signalling substances [7–9]. How these factors are computed in the resulting hunger, satiety and overall energy intake and which factors are the most dominant in this equation is not well understood. Hormones involved in the control of glucose metabolism are likely to be part of these crosslinks, and insulin is understood to be one of the factors that act as an anorexigenic signal [10].How big the influence of single aspects of these regulatory processes like hormones of glucose metabolism is remains uncertain.

Previous studies in rodents found differences in food selection after short fasting periods [11], however it has not yet been investigated whether the same would occur in humans after the acute energy shortage has been partly compensated. Short-term energy status of the body is reflected by hormones like insulin and C-peptide while corticosteroids play a fundamental role in control of glucose metabolism. To which degree the physiological release of insulin as a response to raising glucose levels in the blood after ingestion of carbohydrates can ameliorate feelings of hunger is unknown.

While the quantity of food consumed appears to be the central factor in development of obesity [5], the quality of its components seems to wield some influence as well. Studies suggest that the composition of macronutrients an individual chooses before initiating a therapeutic diet has implications for the diet’s success [12] and similarly that the composition of the therapeutic diet influences diet outcomes and physiological parameters [13,14]. Furthermore, it has been shown that the variation in macronutrient composition of consumed food results in different hormonal responses [15] which might regulate satiety and hunger signals, as well as the choice of macronutrients one prefers. One of these hormones is adiponectin whose serum levels vary after diets as a function of fat and carbohydrate content of the diet [16,17]. Other hormones associated with glucose metabolism, such as insulin [18] and cortisol [19,20], have also been found to be influenced by different macronutrient compositions.

Nevertheless, the interplay between metabolic conditions, hormonal responses and food intake is not well understood and especially little is known about the interplay of macronutrient intake, endocrine factors and absolute energy intake. Most studies examine the effect of diets and food composition on endocrine parameters, food consumption and weight loss. Essential parts of the puzzle, such as hormones regulating the size of the meal and the specific composition of macronutrients, might therefore be missing [21]. Also, the regulation of macronutrient intake in different metabolic situations and long-term deficits or excesses is largely unknown. It has been shown that there are efficient compensatory mechanisms that are unlikely to be explained by fluctuation of hormones with fast kinetics [22]. Adipokines with slower kinetic characteristics may play a major role in this context [23].

To address the questions of how energy deficits control our eating behaviour even if energy deficits have been partially removed by glucose ingestion was the aim of this study. Additionally, it was attempted to find out if adiponectin, one of the emerging major players in the understanding of the metabolic syndrome, plays a role in the regulation of non-acute food intake and its composition under the metabolic conditions that apply in our experiment.

To fill in some of the research gaps, we measured the amount and composition of food consumed after a period of fasting or non-fasting (control); and after acute metabolic energy deficits were at least partly compensated for by oral ingestion of a large load of carbohydrates. It was hypothesised that fasting might still have implications on total food intake after the partial compensation of energy shortage by ingestion of glucose and the resulting alterations of hormones concerned with glucose metabolism towards an anabolic status. To induce insulin and C-peptide levels as high as possible after the glucose administration in the fasting condition to evaluate their role on appetite control as clearly as possible, studies that showed fasting induced glucose intolerance were used to determine the length of the fasting period [24,25]. These studies usually used fasting periods around 70 hours. After consultation with the ethics committee of the University of Luebeck we decided that a fasting period of 42 hours would be ideal in terms of combining the aims of likelihood of producing high insulin and C-peptide levels, safety and practicability. Furthermore, it was hypothesised that an association of the proportions of consumed food with adiponectin levels might exist in this context. This study focussed on a limited array of hormones because they were predominantly important to assess the impact of the oral glucose administration, due to reasons of practicability and the available means.

Materials and methods

Results

Buffet

The total number of kilocalories and the overall amount of each macronutrient consumed differed significantly between the conditions (Fig 1). Absolute food consumption was higher in the fasting group for all measured variables including the total amount of food (EAT: 1442.7±496.7 kcal; FAST: 1841.5±618.7 kcal; t(23) = 5.2, p<0.001), protein (EAT: 235.1±85.2 kcal; FAST: 300.9±103.6 kcal; t(23) = 5.4, p<0.001), fat (EAT: 686.7±262.5 kcal; FAST: 833.5±270.6 kcal; t(23) = 3.8, p = 0.001) and carbohydrates(EAT: 521±191.5 kcal; FAST: 707.2±302.5 kcal; t(23) = 3.7, p<0.001). However, the proportion of macronutrients in kilocalories showed no significant statistical difference between both conditions. Both conditions consumed the same percentage of proteins (16%, t(23) = 0.06, n.s.) and relatively similar percentages of fat (48% in the non-fasting condition and 45% in the fasting condition, t(23) = 1.3, n.s.) and carbohydrates (36% in the non-fasting condition and 38% in the fasting condition, t(23) = 1.2, n.s.).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Food consumption during the buffet after 42 hours of fasting and the control condition.

Box Plots of the total food consumption in the buffet scenario for both fasting and normal eating condition. The total amount of kilocalories and the overall amount of each macronutrient consumed differed significantly between the conditions.

https://doi.org/10.1371/journal.pone.0211514.g001

Hormones and blood glucose

Repeated measures ANOVAs including all 20 measurement timepoints revealed a main effect of conditions for insulin (F(1,23) = 27.7, p<0.001), C-peptide (F(1,23) = 62.2, p<0.001) and ACTH (F(1,23) = 4.7, p = 0.042) but not for cortisol (F(1,23) = 2.8, n.s.), glucose (F(1,23) = 1.2, n.s.) and adiponectin (F(1,23) = 3.3, n.s.). When the timepoints were separated into pre-oral glucose (tp 1–13) and post-oral glucose (tp 14–20), a significant difference was found for insulin (pre: F(1,23) = 371.9, p<0.001; post: F(1,23) = 12.6, p = 0.002), C-peptide (pre: F(1,23) = 331.8, p<0.001; post: F(1,23) = 19.2, p<0.001) and glucose (pre: F(1,23) = 30.5, p<0.001; post: F(1,23) = 42.8, p<0.001) in both sectors. Cortisol only showed significant differences in the post-oral glucose sector (pre: F(1,23) = 0.024, n.s.; post: F(1,23) = 7.6, p = 0.01). Both adiponectin (pre: F(1,23) = 2.8, p = 0.11; post: F(1,23) = 2.9 p = 0.1) and ACTH (pre: F(1,23) = 2.4 p = 0.14; post: F(1,23) = 2.9, p = 0.1) did not show any significant differences in either sector. Figs 2 and 3 show the time courses of all measured serum parameters.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

Levels of cortisol (A), ACTH (B) and adiponectin (C). The Figure shows the mean ± standard error of the mean for both fasting (blue line, crosses) and normal eating (red spotted line, circles) condition of the serum parameters during the experiment. Boxes on the bottom of the graph indicate the timepoints of meals (b = breakfast, L = lunch, D = Dinner, G = oral glucose administration, B = Buffet).

https://doi.org/10.1371/journal.pone.0211514.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

Levels of insulin (A), C-peptide (B), blood glucose (C). The Figure shows the mean ± standard error of the mean for both fasting (blue line, crosses) and normal eating (red spotted line, circles) condition of the serum parameters during the experiment. Boxes on the bottom of the graph indicate the timepoints of meals (b = breakfast, L = lunch, D = Dinner, G = oral glucose administration, B = Buffet).

https://doi.org/10.1371/journal.pone.0211514.g003

Discussion

This study compared food intake after either fasting or non-fasting periods when an acute caloric deficit was partially removed by oral glucose towards the end of the experiment. Whereas the total quantity of food ingested differed between conditions, the relative proportion of each macronutrient consumed did not. The relative amount of protein intake in the non-fasting condition correlated positively with the AUC-value of adiponectin throughout the experiment.

The observed increase in energy intake following a period of energy restriction has been described before [22,27,28] and appears to be intuitively the expected result. These studies suggest that there is a compensatory increase in overall energy intake after a period of caloric restriction as opposed to when individuals are not fasted. No previous study determined whether similar results would be obtained if the acute caloric deficit was compensated by the means of administering a large quantity of carbohydrates. Also the question whether the relative proportion of macronutrients consumed changes between fasting and non-fasting conditions has not been tackled previously.

Studies in obese subjects suggested that a compensatory increase in energy intake after a short energy restricted period does not occur [27,29], whereas data obtained in normal weight individuals tend to find an overall compensatory increase in energy intake after energy restriction thus corroborating the current results. This current experiment extended previous results by showing that food ingestion was increased after fasting in spite of application of oral glucose. The overall energy deficit of the body appeared to be driving overall energy consumption, even though insulin, C-peptide and serum glucose had higher levels in the fasting conditions at this time than in the control-condition. These higher levels reflect the well described phenomenon of “post-fasting glucose intolerance” [25,24]. In our case, this phenomenon shows that these parameters are unlikely to be the only factors providing the brain with feedback about the energy status of the body.

The relative macronutrient proportions of the ad libitum consumed food were not significantly different in between the conditions. We stress the fact that the results of the assessment of the macronutrient composition were only assessed in the context of the pre-meal glucose drink. The administration of the glucose was necessary to test for the main outcome of overall energy intake after a 42 caloric restriction after partial compensation with carbohydrates. Since there was no additional non-replenishment control group the influence of the glucose on the composition of food can’t be determined by this experiment. Further research is needed to confirm this finding and to determine whether the administered glucose might have affected this measurement as it has been shown that administering carbohydrates before a meal can have an influence on food choice. It has been shown before that carbohydrates can increase subsequent food consumption [30]. Thus, further evaluation of the influence a carbohydrate enriched diet has on food choice is needed. Furthermore, there are other possibilities why the food choice was driven into a certain direction. Some of the food items might have been more appetible than others due to their size or looks. Also, parts of the food were offered in a heated state, which might have directed attention towards these food items. Nevertheless, the stability of the proportions of food consumed under the different metabolic conditions remains noteworthy.

Adiponectin concentration did not differ between the conditions in this study. Due to the short but total caloric restriction, this result is in line with previous work showing that short total caloric restriction for 72 hours does not alter adiponectin levels over time [31]. It has also been shown that low caloric diets do not interfere with adiponectin levels for short [32] and intermediate [33] time periods providing that weight loss is minimal. Usually only long term fasting that results in meaningful weight loss results in rising adiponectin levels [34,35]; and recent work in rodents shows that caloric restriction for four weeks increases adiponectin expression on a cellular level [36]. Of note, declining adiponectin levels during fasting interventions have also been reported [37]. The duration of our study was too short to show alterations in levels of a hormone with slow kinetics such as adiponectin. It therefore supports current literature suggesting a very low diurnal and post-prandial biovariability of adiponectin [38], even when sweetened beverages were offered with normal meals [39].

The proportion of consumed proteins correlated with adiponectin levels only in the non-fasting condition. While existing evidence usually deals with the influence of dietary interventions on adiponectin levels [17,40,41], our finding may hint at a role of adiponectin level in food choice, i.e. macronutrient composition. Indeed, there is evidence adiponectin may take part in appetite regulation in the central nervous system [42,43]. However, whether adiponectin has an appetite stimulating [40] or attenuating role [41] remains controversial. The fact that our experiment did not show a difference in adiponectin levels during the experiment makes it less likely that the detected statistical association is derived from a physiological causal mechanism and overall harder to interpret the finding. No change in both adiponectin levels and macronutrient intake in between conditions while the associations only exists in one of the conditions does not leave any room for causal conclusions. However, different metabolic conditions might also alter the influence of endocrine signals on ingestive behaviour and adiponectin might exercise control differently over central nervous function in a fasting metabolic state than in a normally nourished one. Further research will be needed to evaluate the role of adiponectin in this context. Additionally, with regard to the relationship between adiponectin and protein intake, it has to be stressed that the direction of causality cannot fully determined by the current experiment. Although the adiponectin concentrations were measured before the buffet which might indicate that adiponectin was the cause and the protein intake the effect, as pointed out before, adiponectin has rather slow kinetic characteristics making reverse causality a possible option for the interpretation of the data.

Further limitations of this study include that neither women nor people outside an average weight range were included. Thus, these results might not extend to groups other than young, healthy normal weight men. Moreover, the oral glucose administration prior to the administration of the buffet test might have masked effects of fasting on food choice. Future studies might also explore to administer a more tailored quantity of glucose to compensate the energy deficit more precisely, which was not attempted in our study. This study measured only a limited number of hormones directly involved in glucose metabolism and one adipokine that has been shown before to be influenced by the composition of food. Further research into other endocrine signal parameters such as CKK, glucagon, GLP-1, ghrelin and leptin in this context will be needed to gain a more comprehensive understanding of quantity and quality of food intake.

Supporting information

Acknowledgments

Supported by the Deutsche Forschungsgemeinschaft (Research Training Group 1957; Adipocyte Brain Crosstalk [KJC, TFM] and T-CRC 134 project C1, Ingestive Behavior [TFM]).