Observational studies indicate that sugar-sweetened soft drinks (SSSD) may promote obesity, among other factors, owing to low-satiating effects. The effect of energy in drinks on appetite is still unclear. We examined the effect of two isocaloric, but macronutrient, different beverages (SSSD versus semi-skimmed milk) and two non-energy-containing beverages (aspartame-sweetened soft drink (ASSD) and water) on appetite, appetite-regulating hormones and energy intake (EI).
In all, 24 obese individuals were included in a crossover trial. Each subject was served either 500 ml of SSSD (regular cola: 900 kJ), semi-skimmed milk (950 kJ), ASSD (diet cola: 7.5 kJ), or water. Subjective appetite scores, ghrelin, GLP-1, and GIP concentrations were measured at baseline and continuously 4-h post intake. Ad libitum EI was measured 4 h after intake of the test drinks.
Milk induced greater subjective fullness and less hunger than regular cola (P<0.05). Also, milk led to 31% higher GLP-1 (95% CI: 20, 44; P<0.01) and 45% higher GIP (95% CI: 23, 72; P<0.01) concentrations compared with SSSD. Ghrelin was equally 20% lower after milk and SSSD compared with water. The total EI (ad libitum EI+EI from the drink) was higher after the energy-containing drinks compared with diet cola and water (P<0.01).
Milk increased appetite scores and GLP-1 and GIP responses compared with SSSD. The energy containing beverages were not compensated by decreased EI at the following meal, emphasizing the risk of generating a positive energy balance by consuming energy containing beverages. Furthermore, there were no indications of ASSD increased appetite or EI compared with water.
Semi-skimmed milk and sucrose-sweetened soft drinks are isocaloric beverages (volume to volume) but because of their different macro-nutritional composition they may affect satiety in different ways (Westerterp-Plantenga et al., 1999; Halton and Hu, 2004; Bertenshaw et al., 2008). Investigation of the effects of milk compared with sugar-sweetened soft drinks (SSSD) on appetite and energy intake (EI) seems warranted, as milk is considered to be part of a healthy diet, whereas SSSD has been associated with obesity development, which is suggested to be owing to lack of proper satiation induced by this beverage (DiMeglio and Mattes, 2000; Ludwig et al., 2001; Dubois et al., 2007; Leidy et al., 2009). Recently, Dove et al. (2009) showed that skim milk compared with a sugar-sweetened fruit drink decreased hunger and EI 4 h after consumption of the test drink. Sugary beverages may be less satiating than milk because of the content of protein or fat in milk (Fischer et al., 2004; Bertenshaw et al., 2008; Paddon-Jones et al., 2008; Dove et al., 2009) that might affect insulin and appetite-regulating hormones to a larger extent than carbohydrates and promote satiation to a larger degree (Foster-Schubert et al., 2008; Wolnerhanssen and Beglinger, 2010). However, other studies have failed to show decreased EI after milk compared with SSSD (Almiron-Roig and Drewnowski, 2003; DellaValle et al., 2005; Harper et al., 2007; Soenen and Westerterp-Plantenga, 2007).
As a non-caloric alternative to SSSD aspartame-sweetened soft drinks (ASSD) have been introduced to the market in order to reduce the energy content from soft drinks at the population level and thereby eventually to have preventive effects on the prevalence of obesity. Therefore, the effect of ASSD on appetite and energy intake has been discussed intensively (Rolls, 1991; Rogers and Blundell, 1993). Even though some results showed body weight reducing effects of ASSD compared with SSSD (Raben et al., 2002), increased consumption of ASSD has not had the expected moderating impact on the obesity epidemic (Fowler et al., 2008). Previous studies have suggested that ASSD may lead to increased hunger sensation due to the sweet taste unaccompanied by calories (Mattes and Popkin, 2009). However, others show variable effects of ASSD (Melanson et al., 1999) or were unable to associate ASSD consumption with increases in hunger or EI intake (Raben et al., 2002; Mattes and Popkin, 2009). Thus, the impacts of ASSD on appetite are still unknown.
Our aim was to investigate the acute effects of two energy-containing drinks (sucrose-sweetened regular cola and isocaloric semi-skimmed milk) and two non-energy-containing drinks (aspartame-sweetened diet cola and water) on appetite scores, appetite regulating hormones and EI. Our hypotheses were that semi-skimmed milk would induce more satiety signals than isocaloric regular cola and that diet cola would increase EI at the following meal as compared with water.
Subjects and methods
Twenty-four overweight to obese but otherwise healthy subjects were recruited by advertisement in local newspapers and at a national recruitment website. The study was performed as a two center study undertaken simultaneously at Aarhus University Hospital and Faculty of Life Science, Copenhagen University. Inclusion criteria were a body mass index of 28–36 kg/m2; aged 20–50 years; no current illnesses, diabetes or pregnancy. All participants provided written informed consent. The research project was approved by the Ethics Committee of Middle Jutland, Denmark, and was done in accordance with the Declaration of Helsinki.
A crossover design was used with each subject participating for four separate test days, with a minimum of 2 weeks washout period between the test days. The serving order of the test drinks was Latin Square randomized to avoid the order-of-treatment effect. The female subjects were each time tested at the same period of their menstrual cycle.
Participants were asked to abstain from alcohol, medicine and vigorous exercise for 36 h before each test day. To avoid a second meal effect, a standardized evening meal (male 4.2 MJ and female 3.2 MJ) was consumed the night before the test day. The subjects were requested to maintain their usual life style for the duration of the study.
Following an overnight fast, subjects arrived at the laboratory at 0800 hours. Anthropometric measures, appetite scores by visual analog scales (VAS), and blood samples were collected before consumption of the test drink. VAS ratings were used to assess the level of subjective appetite every half hour throughout each testing day. The subjects were placed in a quiet room. They had access to newspapers and toilet facilities. After 4 h the last blood sample and VAS were collected and the ad libitum meal was served and the EI from this meal was calculated.
Body weight was determined with a digital scale (Tanita Corporation, Tokyo, Japan) to the nearest 0.1 kg in light clothes without shoes. Height was measured to the nearest 0.5 cm with a wall-mounted stadiometer (SECA, Hamburg, Germany). Body mass index (kg/m2) was calculated.
The test drinks were 500 ml of (1) sucrose-sweetened regular cola (Coca Cola, Copenhagen, Denmark), (2) semi-skimmed milk (Arla Foods, Aarhus, Denmark), (3) aspartame-sweetened diet cola (Coca Cola), and (4) bottled still water (Aqua d’or mineral water, Brande, Denmark). The energy content and macronutrient composition of the beverages are given in Table 1. The test beverages in the study were not blinded. All test drinks were chilled to 4 °C and were consumed within 10 min.
The ad libitum meal was served 4 h after the test drink was consumed and consisted of pizza slices with ham and cheese (see Table 1 for energy and macronutrient content). The subjects were instructed to eat at a constant pace and to stop when they felt comfortably satiated. Food was weighed before serving and plate waste was collected and weighed in order to calculate EI. The meal was served together with 250 ml still water.
VAS were used to assess subjective appetite (Flint et al., 2000). The feeling of satiety, hunger, fullness, prospective food intake (assessment of how large a meal the subject expects to consume at that moment to feel comfortably full) and thirst was measured with anchored 100 mm VAS. The scale ranged from ‘not at all’ to ‘extremely’. All VAS were provided in individual booklets for each time point and were collected immediately after completion.
Commercially available assays were used to analyze glucose (Gluco-quant, Roche Diagnostics, Rotkreuz, Switzerland), insulin (Human Insulin Elisa kit, Dako, Glostrup, Denmark). Acyl-Ghrelin was collected into EDTA plasma tubes and immediately added Pefabloc. After which the sample was centrifuged at 4 °C for 15 min at 2000 × g. Subsequently, 50 μg/ml HCL was added to the sample before it was frozen and kept at −80 °C until later analysis with a commercially available kit (Human acyl-Ghrelin Elisa kit, Mitsubishi, Tokyo, Japan). Intact GIP and total GLP-1 concentrations in plasma were measured after extraction of plasma with 70% ethanol (vol/vol, final concentration). Intact, biologically active GIP was measured using antibody 98171 as previously described (Deacon et al., 2000). The plasma concentrations of GLP-1 were measured as described elsewhere (Orskov et al., 1994).
Statistical analyses were performed with the Statistical software package SAS version 9.2 (SAS Institute Inc., Cary, NC, USA). Descriptive statistics are presented as means±s.d. It required 24 subjects to show a 10-mm difference in VAS satiety score with a study power of 0.9 (Flint et al., 2000). Results are presented as effects and 95% CIs in the text or as adjusted means±s.e.m.s in the figures. In the figure each subject's repeated measurements were adjusted for the average area under the curve (AUC) for the subject during all four test days for the given variable. This was done to avoid introducing bias into the data by subtracting the subject's baseline value from all other time point values.
Repeated measures were summed to a trapezoid AUC. All variables were analyzed in a mixed model with subject as the random factor and corrections for gender and test site. Model selection also included considerations of effects from trial day and body mass index as well as modification from gender on the treatment effect. The appetite-regulating hormones, glucose and insulin were all log-transformed to meet the assumptions of normal distribution. Peak and nadir values of hormones and VAS were also analyzed.
All-over changes between the four groups were determined by analysis of variance and post-hoc pair-wise analysis was Bonferroni corrected for multiple comparisons. Concerning calculation of energy compensation at the ad libitum meal it was only relevant to compared the two energy-containing beverages (SSSD vs milk), which was done with a paired t-test. Two-tailed P-values below 0.05 were regarded as significant.
Subjects and baseline characteristics
A total of 24 subjects (12 females, aged 33.5±9.2 years, body mass index: 31.4±3.11 kg/m2, body weight: 94.6±11.8 kg) completed all four test days with no differences in their basal characteristics between the test days (data not shown).
Effects on appetite regulating hormones
The effects of the test drinks on circulating levels of ghrelin, GLP-1 and GIP are shown in Figure 1. The ghrelin response was significantly different between the four groups (P<0.05, analysis of variance). With water as the control drink the AUC of ghrelin was significantly reduced after regular cola (P<0.05) and was non-significantly reduced after milk (P=0.07), but no differences were observed compared with diet cola (Figure 1a). The GLP-1 response was also significantly different between the four groups (P<0.001, analysis of variance) and with water as reference both regular cola (P=0.04) and milk (P<0.001) induced a significantly higher GLP-1 response than water. There were no differences between diet cola and water. Moreover, the GLP-1 response after milk was significantly higher by 31% (95%CI: 20, 44; P<0.01) compared with regular cola (Figure 1b). Finally, the GIP response was very similar to the GLP-1 response with significant differences between the four groups (P<0.01). The GIP AUC was higher after regular cola and milk as compared with both diet cola and water (P<0.01 for all comparisons). The GIP response after milk was significantly higher by 45% (95% CI: 23, 72; P<0.01; Figure 1c) compared with regular cola. No differences were observed between the effects of diet cola and water on GLP-1 and GIP concentrations.
Analyzing the differences in peak/nadir concentrations of these hormones gave very similar results as were obtained with the AUC values (data not shown).
Effects on insulin and glucose
As expected the peak glucose concentration was significantly different between the four groups (P<0.01), with higher levels after regular cola compared with all three other test drinks (P<0.01, data not shown). The peak glucose concentrations were increased after milk compared with diet cola and water (P<0.05).
The AUC of insulin was significantly different between the four groups (P<0.001). Insulin (AUC) was increased by regular cola and milk compared with diet cola and water (P<0.01). There was no significant difference in insulin (AUC or peak value) between regular cola and milk or between diet cola and water (data not shown).
Effects on appetite sensations (VAS)
Satiety, hunger, fullness and prospective consumption (AUC) for all four drinks are shown in Figure 2. Satiety was overall different between the four groups (P<0.01). In post hoc analysis satiety was higher after milk compared with water (P<0.01) and tended to be higher after milk compared with regular cola (6.0 mm × min; 95% CI: 0.1–11.9; P=0.08). Hunger was different between the four groups (P<0.001) and in the post-hoc analysis hunger was lower after milk compared with regular cola (−7.5 mm × min; 95% CI: −12.7 to −2.3; P<0.05) and compared with water (P<0.01). Also fullness sensation was different between the four groups (P<0.05). Milk led to higher fullness compared with regular cola (7.7 mm2; 95% CI: 1.0–14.4; P<0.05). The subjective sensation of prospective intake (all-over differences, P<0.01) tended to be lower after milk compared with regular cola (P=0.07). No significant difference in AUC between diet cola and water was found in these subjective appetite sensations.
Analyzing the peak and nadir values from VAS it was found that peak values of fullness were lower in the water group compared with each of the other groups also compared with diet cola (P<0.05 for all). Otherwise, analysis of peak/nadir values gave no further information to the data obtained with the AUC values (data not shown).
Effects on ad libitum meal EI
No significant differences between the four groups were found on the EI from the ad libitum meal eaten 4 h after the test drinks (Table 2). However, total EI (ad libitum+test drink) was higher after regular cola and semi-skimmed milk as compared with diet cola and water (P<0.01; Figure 3).
With water as the reference drink there was virtually no compensation for the EI from the energy containing beverages at the ad libitum meal. The EI from milk (950 kJ) was only compensated by 14.7% at the following meal and the EI from regular cola (900 kJ) was not compensated at all (÷3.6%) (Table 2). This energy compensation was not different between the milk and regular cola groups (paired t-test, P=0.6). The mean 216 kJ higher EI after ASSD as compared with water (Table 2) was non-significant.
We investigated two energy-containing beverages (sucrose-sweetened regular cola and isocaloric semi-skimmed milk) and two non-caloric beverages (aspartame-sweetened diet cola and water) on appetite sensations (measured by VAS), on appetite hormones and on ad libitum EI.
Semi-skimmed milk induced more satiation sensations and had more pronounced impact on satiety hormones than isocaloric regular cola. Compared with the energy-containing beverages, the total EI after water and diet cola was significantly reduced. This indicates that energy from intake of energy-containing beverages is not compensated for at the following meal.
Milk increased satiety and fullness and decreased hunger and prospective intake compared with regular cola and had a more pronounced effect on GLP-1 compared with the other test drinks. Over-all diet cola had rather similar effects on appetite hormones and VAS compared with water.
It is generally suggested that milk has more satiating power than carbohydrate-containing beverages. Dove et al. (2009) found that 600 ml of skim milk with 25 g protein reduced the EI at the following meal by 8.6% corresponding to a difference of 226 kJ (P<0.05) compared with isocaloric fruit juice. However, in agreement with our results DellaValle et al. (2005) found no changes in EI between regular cola, fruit juice or milk (1% fat) after the volunteers consumed 360 g of the various beverages together with a meal. Also, Soenen and Westerterp-Plantenga (2007) found similar effects of milk and SSSD on appetite sensations in a 140-min intervention. The different results may be because of the fact that the effect of milk on satiation sensations and subsequently on EI grows over time and is most pronounced after 2 h or longer after the milk intake (Anderson et al., 2004; Dove et al., 2009).
Dove et al. (2009) performed a study rather like ours with a test drink followed by an ad libitum meal 4 h later. Our results are rather similar except that their difference of EI between milk and fruit juice at 226 kJ reached statistical significance, whereas the difference of 173 kJ between regular cola and milk in our study did not. This may be due to the fact that Dove et al. (2009) served 600 ml of the test drink along with a standardized breakfast meal, whereas we only served 500 ml test drink after an overnight fast. The subjects of our study may, therefore, have been hungrier, when the ad libitum meal was served. Thus, the effect of the test drink may not have worked through. This is supported by the fact that EI of our subjects was about twice as high as the EI consumed by the subjects in the Dove study (Dove et al., 2009). Thus, if the ad libitum meal was served earlier, other results may eventually have been found. Finally, it may be due to less statistical power due to only 24 subjects in our study compared with 35 subjects in the study of Dove et al. (2009). The more pronounced effect of milk on satiation and appetite hormones as compared with regular cola may be due to factors all involved in appetite regulation, such as high protein and fat containment, higher viscosity (Mattes and Rothacker, 2001; Zijlstra et al., 2008), and differences in carbohydrate content (lactose vs sucrose) (Bowen et al., 2006).
We found that energy in the test drinks (regular cola and milk) was not compensated by subsequently decreased EI. This is in agreement with other studies (Flood et al., 2006; Harper et al., 2007; Soenen and Westerterp-Plantenga, 2007). Moreover, Flood et al. (2006) found that a larger portion size of the drink was associated with a larger total EI. Thus, these findings are suggesting that intake of energy-containing beverages in general aggravates the risk of a positive energy balance.
Our hypothesis that ASSD would enhance the ad libitum EI was not supported by the present results. We found no indications that intake of diet cola leads to less satiation than water. The ad libitum EI was similar after the two beverages and no differences in the effects on the appetite hormones were found. These observations are in agreement with intervention studies where ASSD is not associated with positive energy balance (Tordoff and Alleva, 1990; Raben et al., 2002). Thus, the association between ASSD and obesity observed in several epidemiological studies (Colditz et al., 1990; Fowler et al., 2008) may be due to other factors than a direct effect of ASSD. However, various other artificial sweeteners (sucralose, acesulfame-K, aspartame and so on) may affect the appetite differently (Brown et al., 2010).
In conclusion, our study supports that semi-skimmed milk seems to increase the subjective appetite sensations and the release of the anorexic GLP-1 to a higher degree than isocaloric regular cola. However, these energy-containing beverages were not compensated by decreased EI at the following meal, emphasizing the risk of generating a positive energy balance by consuming energy containing beverages. Furthermore, we did not find any indications that ASSD (aspartame-sweetened diet cola) increases appetite or EI compared with water. Long-term intervention studies are needed to clarify whether the different effects of the presented beverages on appetite sensations, appetite hormones and EI have consequences for body weight and obesity development.
The study was supported by The Danish Council for Strategic Research, The Food Study Group/ Danish Ministry of Food, Agriculture and Fisheries, Novo Nordic Foundation, and Clinical Institute at Aarhus University, Denmark.
The funding entities had no role in study design and implementation or in analyzing and interpretation of the data.
The trial was registered at http://www.clinicaltrials.gov. ID no. NCT00776971.