Ultra-processing of carbohydrate-containing foods has been temporally associated with rising rates of obesity, type 2 diabetes mellitus and CHD(Reference Augustin, Kendall and Jenkins1). The ability of carbohydrates to influence metabolic disease risk via an effect on postprandial glycaemia has been debated through the glycaemic index (GI) literature since 1981(Reference Augustin, Kendall and Jenkins1). Although evidence is conflicting, the Glycaemic Index Foundation claims that low GI not only results in less glycaemic response but also can keep you feeling fuller for longer(2). The suggestion that glycaemic responses elicited by carbohydrate-containing foods may be causal to the rising rates of metabolic diseases(Reference Augustin, Kendall and Jenkins1) demands attention be paid to the impact of postprandial glycaemia on satiety.
The glucostatic theory of food intake regulation has been a long-standing proposition(Reference Mayer3). The suggestion is that increased blood glucose concentrations promote satiety, whilst low blood glucose concentrations lead to increased feelings of hunger(Reference Jenkins, Wolever and Taylor4). Following a review of the literature, it has been proposed that foods inducing a rapid rise and fall in glycaemia are satiating within an hour of eating and that foods producing a lower, but more prolonged, glycaemic response are satiating over a longer time course(Reference Anderson and Woodend5). However, evidence of a low glycaemic response correlating with higher satiation is conflicting(Reference Alfenas and Paiva6), and there are multifactorial determinants of appetite and satiety(Reference Gregersen, Moller and Raben7). The inconsistencies among postprandial glycaemic studies may be explained by factors other than the glycaemic effect such as cooking and processing methods, the presence of fibre, the presence of other macronutrients, food form, palatability and energy density(Reference Brouns, Bjorck and Frayn8). The differences could also be attributed to variable study designs and methods used to measure satiety(Reference Mayer9).
To examine whether GI has an effect on satiety, it is important to isolate the glycaemic response from other confounding factors. One such strategy is to use sucrose- and isomaltulose-sweetened beverages, both disaccharides comprising fructose and glucose moieties. Although a fully digestible carbohydrate, isomaltulose is digested at a slower rate than sucrose due to its more stable α-1,6 glycosidic bond connecting the fructose and glucose moieties, compared with the sucrose α-1,2 linkage(Reference Holub, Gostner and Theis10). The objective of the study is to assess the effect of a difference in glycaemia per se on primary outcomes of satiety by controlling for bias by double-blinding and controlling for confounding factors. The primary outcomes will be satiety subjectively assessed using visual analogue scales (VAS) and subsequent energy intake objectively assessed using weighed diet record data.
Methods
Participants
This study was a double-blinded randomised crossover trial, occurring during March 2018 at the University of Otago, Dunedin, New Zealand. Participants (n 77) were students of Human Nutrition Department from the University of Otago aged between 18 and 60 years old. Participants were not eligible if they had diabetes or an intolerance to the sweeteners being used. Randomisation of the order in which participants received the test beverages was computer-generated. Fig. 1 shows the flow of participants through the study. Ethical approval was granted by the Human Ethics Committee of the University of Otago in October 2017 (ethics committee number 17/011). The trial was registered with the Australian New Zealand Clinical Trial Registry (ACTRN12618000901202). All subjects gave written informed consent before entering the study.
Fig. 1. Study design and flow of participants through the study. * One individual was missing from both sessions.
Study design
A total of seventy-seven participants were asked to attend two test sessions 2 weeks apart. As this was a crossover design, participants were asked to keep their eating and activity patterns consistent on the mornings of the two test days and to report to the testing premises at 24.00 hours. On arrival, participants were provided with a standard lunch consisting of eight pieces of sushi (1290 kJ) and a cup of water. The purpose of providing lunch was to standardise appetite across the two test days. All participants consumed all of the sushi. After eating, participants were free to leave the premises with instructions not to eat or drink (except for water) or to undertake strenuous exercise and to return at 13.45 hours. Anthropometric measurements and a demographic questionnaire were completed on arrival at the first session. After removing shoes and jackets, participants were weighed using calibrated electronic scales (model 770; Seca GmbH & Co.). Height was measured using a free-standing stadiometer (Holtain Limited). The measurements were used to calculate BMI by dividing weight in kg by height in m2. After a baseline measure of satiety, participants were given 10 min to drink the entire test beverage followed by assessments of satiety and subsequent food intake. The order in which participants received the test beverages was randomised using random length blocks in Stata 15.1 (StataCorp). A university staff member otherwise uninvolved in the study coded the beverages such that the principal investigators, participants and the statistician were blinded to treatment with the code revealed after statistical analysis of the data.
Test beverages
Quantities of 50 g of sucrose (caster sugar; Smart Choice) or isomaltulose (unflavoured Palatinose®; Myprotein) were measured on calibrated electronic scales (model 1702; Sartorius) and added to 500 ml bottles of Pure New Zealand sparkling water. The sweetness of the two beverages was matched by adding 0·035 g of sucralose (98 % sucralose powder, J66736, lot: T21D050; Alfa Aesar) to the isomaltulose beverage. To ensure double-blinding, the drinks had the same volume and appearance and 0·05 ml of lemon flavouring (Lemon 59223, lot: 1002802470; Invita NZ Ltd). The beverages were matched for sweetness and immediate taste using a triangle sensory testing protocol prior to the laboratories, involving six people independent of the present study. Glycaemic and insulinaemic responses to the beverages were tested in twelve people to confirm between-drink differences in these factors; the results of these tests have been reported(Reference Keesing, Mills and Rapsey11). The mean published GI value of sucrose is 65 (sem 4)(Reference Atkinson, Foster-Powell and Brand-Miller12) and of isomaltulose 32(13).
Satiety and subsequent food intake
Satiety was assessed according to a published methodology(Reference Flint, Raben and Blundell14) using 100 mm VAS administered on paper at baseline (14.00 hours) and at 30, 60, 90, 120, 150 and 180 min following baseline. The questions were ‘How hungry do you feel?’ (not at all/never been more hungry); ‘How satisfied do you feel?’ (completely empty/cannot eat another bite); ‘How full do you feel?’ (not at all/totally full) and ‘How much do you think you can eat?’ (nothing at all/a lot). To assess subsequent energy intake over the remainder of the day, participants were asked to keep a weighed diet record of all food and beverages ingested from 17.00 hours to midnight using electronic scales reading to 0·1 g (Salter). Participants were trained in dietary recording and in the use of the scales and given a diet record sheet to fill in and return. The investigators used proprietary University of Otago software (Kaiculator) to input and process the dietary data using the New Zealand Food Composition Database as the source of nutrient information(15).
Statistical analysis
The sample size required to provide 80 % power to the α = 0·05 level to detect a 5 mm difference in VAS scores was 70; and to detect a 400 kJ difference in subsequent energy intake, the sample size required was 60. Mean differences, 95 % CI and P values of outcomes between treatments were calculated using mixed-effects regression analysis, with the participant as a random effect and adjusting for randomised order and baseline measures. Only participants with complete data were included in the analysis. The statistical analysis was undertaken using Stata/1C version 13.1 (StataCorp. 2013).
Results
The flow of participants is given in Fig. 1. Data from sixty-nine participants who drank both beverages were included in the analysis. Participant demographics are given in Table 1. The median (10th and 90th percentiles) age was 20·9 (95 % CI 19·9, 24·8) years and BMI ranged from 17·4 to 34·1 kg/m2.
Table 1. Participant characteristics (n 69)
(Mean values and standard deviations; numbers and percentages)
Plasma glucose and insulin concentrations following ingestion of the test beverages are plotted in Fig. 2.
Fig. 2. Plasma glucose and insulin responses to the ingestion of sucrose and isomaltulose + sucralose beverages. Values are medians, and vertical bars represent the range of the 25th to 75th percentiles. 
, Sucrose; 
, isomaltulose.
Results from VAS questionnaires assessing hunger, satisfaction, fullness and prospective food intake are presented as mean values and standard deviations AUC in Table 2 with no between-beverage differences for any of the four questions.
Table 2. AUC visual analogue scale (VAS) questions between the sucrose- and isomaltulose-sweetened beverages (n 69)
(Mean values and standard deviations; mean differences and 95 % confidence intervals)
* Mixed-effects regression with participantʼs ID as a random effect and adjusting for randomised order and baseline.
† Hunger – ‘How hungry do you feel?’; Satisfaction – ‘How satisfied do you feel?’; Fullness – ‘How full do you feel?’; Food intake – ‘How much do you think you can eat?’ For ‘Hunger’ and ‘Prospective food intake’, a larger AUC corresponds to greater hunger and a desire for more food. For ‘Satisfaction’ and ‘Fullness’, a larger AUC corresponds to greater satisfaction and fullness.
A visual illustration of appetite over time as measured by VAS is given in Fig. 3. The mean response scores after each beverage were close at all time points resulting in virtually superimposed plots for all four questions.
Fig. 3. Satiety scores over time to the questions (a) ‘How hungry do you feel?’ (b) ‘How satisfied do you feel?’ (c) ‘How full do you feel?’ (d) ‘How much do you think you can eat?’ (n 69). Values are medians, and vertical bars represent the range of the 25th to 75th percentiles. , Sucrose; , isomaltulose.
Eight people did not complete diet records following both the sucrose- and isomaltulose-testing days, and hence, sixty-one paired records were available for analysis. The mean energy and macronutrient intake of the weighed diet records representing food intake from 17.00 to 24.00 hours are presented in Table 3. There was no between-treatment difference in macronutrient or dietary energy intake –291 (95 % CI –845, 267) kJ over the remainder of the day. A sensitivity analysis that included the amount of energy participants consumed before or during the test had no appreciable impact on the effect sizes and significance of these results.
Table 3. Energy and macronutrient intake from weighed diet records (n 61)
(Mean values and standard deviations; mean differences and 95 % confidence intervals)
* Mixed-effects regression with participantʼs ID as a random effect and adjusting for randomised order.
Discussion
There were no significant differences following ingestion of a sucrose- and an isomaltulose-sweetened beverage in either subjective satiety assessed using VAS or by an objective measure of subsequent food intake. A direct comparison between sucrose and isomaltulose on satiety has not previously been reported in humans. However, our findings are contrary to those of an animal study in which rats provided with these sugars consumed less food during a period of isomaltulose administration compared with sucrose(Reference Sasagawa, Mineo and Hirayama16). It is unclear why this occurred but one possibility is that the sucrose solution was more palatable to the rats.
Over the years, comparable studies to ours have been carried out in humans using glucose- and fructose-sweetened beverages with variable findings. When a total sample of twenty-four normal and overweight people ingested 50 g glucose- or fructose-sweetened beverages, less energy was consumed at a subsequent buffet lunch following the fructose preload compared with the glucose preload(Reference Rodin17). In contrast, when four women ingested 10 % solution of glucose- or fructose-sweetened preloads, there was no difference in satiety VAS ratings or subsequent energy intake of a lunch meal(Reference Guss, Kissileff and Pi-Sunyer18). Similarly, giving 75 g solution of glucose or fructose to eight men resulted in no difference in satiety VAS ratings or in subsequent energy intake at a buffet lunch(Reference Kong, Chapman and Goble19). After ten people with type 2 diabetes and ten people with normal glucose tolerance ingested beverages sweetened with 75 g fructose or glucose, satiety assessed with VAS and subsequent energy intake of a buffet meal were favourable compared with a control beverage (flavoured water), but there were no differences in these outcomes between sugary beverages(Reference Vozzo, Baker and Wittert20). When twenty-eight obese men ingested beverages sweetened with 50 g glucose or fructose, there were no between-treatment differences found for VAS responses or subsequent energy intake at a buffet lunch(Reference Bowen, Noakes and Clifton21). The lack of effect on satiety in these studies despite measured differences in postprandial glycaemia is not in accord with the glucostatic mechanism of food intake proposed by Mayer(Reference Mayer9). The glucostatic theory states that temporary increases in blood glucose concentration correspond to a decrease in food intake and viceversa(Reference Mayer9). When comparing short-term (1-h) satiety after consumption of high- and low-GI isoenergetic beverages, it was found that the high-GI beverage kept participants feeling more full at 60 min, in accordance with the glucostatic theory(Reference Anderson, Catherine and Woodend22). It was thus proposed that high GI keeps you fuller short-term and that low GI may sustain satiety long-term(Reference Anderson, Catherine and Woodend22).
The proposition that GI affects satiety is not specific regarding the timing of ingestion, although most studies are undertaken after an overnight fast. In our study, participants abstained from food and beverage for 2-h following a mid-day meal and the glycaemic and insulinaemic responses in response to isomaltulose ingestion remained below the pre-ingestion values. This pattern of response differs to that found after an overnight fast in which rises in blood glucose and insulin concentrations to an isomaltulose beverage have been found(Reference Tan, Tan and Henry23,Reference van Can, van Loon and Brouns24) . A possible explanation is that our participants were still in a postprandial state at the time of test beverage ingestion as suggested by the mean pre-ingestion glucose concentration of 7·0 mmol/l. This is relatively high for a young group of people given the overnight fasting blood glucose concentration in a Caucasian group with a mean age of 27·5 years was 4·8 mmol/l(Reference Tan, Tan and Henry23). Indeed, blood glucose concentrations are affected by the duration of fast, continuing to decline in young people past a 2-h postprandial period(Reference Moebus, Gores and Losch25). Nevertheless, in our study, there was a glycaemic and insulinaemic differential at 30 min between the two test beverages with the magnitude of separation comparable with that found when these sugars are tested following an overnight fast(Reference Holub, Gostner and Theis10,Reference van Can, van Loon and Brouns24) . Although our data do not support an effect on satiety of differing glycaemic responses induced by these beverages, our subjective findings are limited to a non-overnight fasting condition monitored over 3 h.
However, over time, a relationship between the glycaemic response and satiety to solid food has been found. Twelve obese teenage boys ate 81 % more energy over 5-h following consumption of high-GI meals compared with low-GI meals(Reference Ludwig, Majsoub and Al-Zahrani26). This finding is quite different to that of the sweetened beverage trials, in which generally no differences in subsequent energy intake were found. The explanation may lie with the composition of the meals. The low-GI meal comprised fruit (grapefruit and apple) and a vegetable omelette with an initial weight of solid food of 670 g. In comparison, the high-GI meal consisted of 61 g instant oatmeal and 19 g dextrose, with the remaining ingredients being liquid (160 ml milk, 15 ml cream and 397 ml water). There is substantial evidence that solid foods are more satiating than liquids(Reference Brindal, Baird and Slater27–Reference Kuzma, Cromer and Hagman29) with mechanisms being discussed in a review paper(Reference Chambers30). One such mechanism is the action of chewing which would have differed between the low-GI omelette and fruit meal, and the high-GI meal of instant oatmeal, with chewing eliciting a higher degree of satiety due to alterations in gut hormone responses(Reference Miquel-Kergoat, Azais-Braesco and Burton-Freeman31). Controlling for food texture and macronutrient content, mean hunger and perceived fullness in fifty participants were favourable after eating a low-GI carob cookie compared with a high-GI chocolate cookie 15 min after eating with no between-treatment difference thereafter; and subsequent energy intake at lunch following the preload was some 7 % higher for the high-GI chocolate cookie(Reference Papakonstantinou, Orfanakos and Farajian32). This might suggest a role for GI in satiety, although the carob cookie contained considerably more fibre than the chocolate cookie, and the palatability of the cookies was not reported, factors that have been variably associated with satiety(Reference Clark and Slavin33,Reference Sorensen, Moller and Flint34) . In a trial involving twenty-eight men consuming fourteen different breakfast meals having a range of GI, subjective feelings of satiety were unrelated to glycaemic response, whilst energy intake at a subsequent lunch was positively associated with the glycaemic response(Reference Flint, Moller and Raben35). Again this might suggest that the glycaemic response and satiety are related, but the breakfast meals were not matched on energy content such that the total energy intake of the breakfast and subsequent lunch was lower for the highest-GI breakfast (reference bread) and highest for the lowest-GI breakfasts (reference bread with butter and cheese; Finnish bread with butter and cheese and German bread with butter and cheese)(Reference Flint, Moller and Raben35). In a study in which the energy content of foods was standardised, the volume of ingested food was found to be the main determinant of satiety independent of the glycaemic response(Reference Holt, Miller and Petocz36).
From such work, it may be concluded that there is no clear relationship between the glycaemic impact of food and satiety. Discrepancies in findings may be due to study designs in which properties of the foods other than the glycaemic characteristics have not been controlled. Confounding factors are particularly prominent in long-term studies regarding associations between postprandial glycaemia and satiety. There appears to be little relationship between the GI of foods and weight gain over periods of months and years(Reference Aston37–Reference Sloth, Krog-Mikkelsen and Flint39). This is probably attributable to the GI of foods being unrelated to the energy density of foods, for example, cake and apples have comparable GI but very different energy densities(Reference Holt, Miller and Petocz36). Whilst controlling for energy density, our findings support a role for ingested volume as a causal factor in feelings of satiety as the beverages were isovolumetric. By using a beverage, we were able to standardise energy density whilst controlling for macronutrient content and palatability. The major strength of this study was the control for confounding and bias that allowed for glycaemic differences, whilst participants and investigators were blinded to treatment. A potential limitation is that we did not assess mood. Mood has been found to affect food intake(Reference Yeomans and Coughlan40) and if people had different moods on the two test days, this could have confounded feelings of satiety. In some studies, subsequent energy intake from a provided meal has been assessed by observing and measuring food intake under investigator-controlled conditions(Reference Furchner-Evanson, Petrisko and Howarth41,Reference Anderson, Cho and Akhavan42) . In contrast, our participants were asked to record food and beverage intake under free-living conditions, a technique that has been found to result in under-reporting(Reference Martin, Su and Jones43). This is likely a limitation to our estimate of absolute subsequent energy intake although being a crossover study, energy intake estimates on each testing day have been controlled for potential under-reporting by making within-person comparisons. Another limitation is that subsequent energy intake was assessed over a time frame of 7 h (17.00 hours to midnight). Although subsequent energy intakes have been assessed over shorter periods of 1–3 h following interventions designed to elicit differences in postprandial glycaemia, these study designs do not account for possible longer term carryover effects(Reference Sun, Li and Zhang44). The findings may be limited to a young, primarily female, university-educated demographic with normal glucose tolerance, and as such it would be of interest to test these beverages in males, other age and socio-demographic groups, and in people of different metabolic states to assess whether glycaemic response, per se, affects feelings of satiety.
In conclusion, the present study is the first to assess the satiating properties of isomaltulose compared with sucrose beverages in humans. In a healthy, young cohort of participants, feelings of satiety and subsequent energy intake were unaffected. For the magnitude of glycaemic differences attained in this study, our data do not support a relationship between postprandial blood glucose concentration and feelings of satiety.
Acknowledgements
We gratefully acknowledge Dr Mei Peng for supervising the sensory testing. We would like to acknowledge the technical support of Ivy Salih and Ken Thomson in the running of the laboratories and executive chef Charlie Mancey for the provision of the sushi lunches.
This research was supported by The University of Otago.
B. J. V. conceived the study. All authors designed the study. B. M. M., C. T. K. and B. J. V. undertook the practical work. B. J. V. supervised the study. J. J. H. analysed the data. B. M. M. drafted the manuscript. All authors edited the manuscript.
None of the authors has any conflicts of interest to declare.
