The incidence of obesity has increased over recent years and has reached an epidemic extent(1); this is also associated with other health issues such as an increased prevalence of insulin resistance and type 2 diabetes. Various strategies have been investigated to determine how this problem could be tackled and to prevent a further increase in body weight and the associated conditions. These strategies include investigations into individuals' diets and specific dietary components that could be altered to subsequently impact on weight, insulin resistance and type 2 diabetes. One such component is dietary fibre. This diverse group of carbohydrates has been proposed to increase satiety(Reference Slavin and Green2) and therefore may have a beneficial role in weight management. Dietary fibre may increase satiety by several mechanisms and different fibres may exert their effects by different means although the specific processes are not known. Proposed mechanisms include the lower energy density of fibre foods, the impact of fibre on delaying gastric emptying, which subsequently could prolong the feeling of fullness, and fibre slowing glucose absorption, which may prolong satiety(Reference Pereira and Ludwig3). There is also evidence to suggest that dietary fibres may influence insulin sensitivity(Reference Weickert and Pfeiffer4) independently to any effects on appetite.
Resistant starch (RS), a fermentable carbohydrate, has been proposed to have properties similar to dietary fibre(Reference Champ5) and therefore could also affect satiety and exert a beneficial role in weight regulation. However, as RS is a non-viscous fibre and is not thought to affect gastric emptying or the absorption of nutrients(Reference Robertson, Bickerton and Dennis6), the effects on satiety, if indeed there are any, are unclear. Use of RS may be a more advantageous means of increasing fibre intake compared to traditional fibres, as it is easy to incorporate into everyday foods(Reference Brown7) without adverse alterations to either taste or texture(Reference Nugent8).
There are few studies that have investigated the effects of including RS into the diet and the consequent effects on appetite. Animal studies show consistent positive effects on appetite regulation and insulin sensitivity; however, data in human studies are mixed(Reference Higgins9). In particular, there are no current studies in human subjects that have explored the effects on appetite of including RS in mixed meals when compared to energy and carbohydrate-matched placebo. Of the four classifications of RS, the majority of studies looking at effects on appetite have used type 2, which are intrinsically resistant, and type 3, which are retrograded starches.
The postprandial effects of consumption of RS are also not known; however, as a non-viscous fibre, it would not be predicted to affect glucose absorption. Nevertheless, it has been shown in a recent study that short-term consumption of RS type 2 improves postprandial glucose metabolism in healthy individuals(Reference Robertson, Bickerton and Dennis6).
The present study therefore aimed to investigate the acute effects of including 48 g of RS type 2 on energy intake, subjective measures of appetite and changes to postprandial glucose and insulin, compared to a placebo supplement in a randomised, single-blind balanced crossover, in which the 48 g were divided equally between breakfast and lunch (providing 24 g at each meal) and consumed as part of these mixed meals.
Twenty young, healthy, adult males, aged 19–31 years, with a mean BMI 23·2 (sem 0·65) kg/m2 (Table 1) participated in the study. Subjects had no history of gastrointestinal disease or endocrine disorders and were weight stable for at least the preceding 3 months. Highly restrained eaters, identified by the Dutch Eating Behaviour Questionnaire(Reference van Strien, Frijters and Bergers10), were excluded from the study. The participants included in the study had a mean score of 2·1 (sem 0·2) on the restraint scale, 1·98 (sem 0·1) on the emotional scale and 3·1 (sem 0·1) on the external eating scale. The present study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the University of Surrey's Ethics Committee; written, informed consent was obtained from all subjects.
* Measured by bioimpedance (Tanita TBF-300, Tanita, UK).
† Mean of three readings taken with the subject in a sitting position, measured by an automatic blood pressure cuff (Omron MX3 Plus; Omron Healthcare Europe, Kruisweg, UK).
Subjects attended the investigation unit after an overnight fast, on two occasions, at least 1 week apart, when they consumed either the RS supplement or the placebo supplement within the test breakfast and lunch meals, which provided 24 g RS at breakfast and 24 g RS at lunch on the RS leg. Blood samples and appetite ratings, including desire for different foods, (measured using 100 mm visual analogue scales, as described by Flint et al. (Reference Flint, Raben and Blundell11)), were taken every 30 min for the 7-h intervention period. The participants remained in the investigation unit for the duration of the study and were required to minimise activity.
On the day before each study day, each participant consumed an identical evening meal and was required to avoid alcohol, caffeine and strenuous exercise for at least 24 h before the study.
On the two separate days, subjects received either 80 g Hi-Maize® 260 product (60 % RS type 2 and 40 % rapidly digestible starch (RDS)) which therefore provided 48 g RS and 32 g RDS (as measured by The Association of Official Analytical Chemists for total dietary fiber method 991·43) or 32 g of the placebo Amioca® (100 % RDS) providing 32 g RDS. Both the supplements were supplied by the National Starch Company, LLC (Bridgewater, NJ, USA) and were incorporated into a flavoured mousse. Available carbohydrate from the RDS portion of both supplements was exactly matched on both occasions, giving meals with comparable glycaemic load. The mousses were of a similar taste and texture and were consumed as part of the test breakfast and lunch meals. Forty grams of the Hi-Maize supplement were the highest quantity that could be added to one mousse portion without adverse effects on taste or texture and a similar high level of RS has been given in other studies without adverse gastrointestinal effects. Subjects were able to choose their preferred flavour of mousse from three choices to enhance compliance, but consumed the same flavour on both visits.
Subjects were cannulated; two fasting blood samples were taken, one 15 min before breakfast and the other just before breakfast, which was served at time zero. The breakfast on both days was of a standardised portion size and consisted of Rice Krispies® (Kellogg's, Manchester, UK) with semi-skimmed milk and one portion of the mousse which contained the test carbohydrate; the energy and macronutrient composition of this meal is shown in Table 2.
* Resistant starch meal only.
Lunch, containing the second half of the test carbohydrate, was served at 180 min, to allow the participants to receive their meals at times similar to their habitual pattern. The lunch meal was either ham or cheese sandwiches (the same filling was consumed on each study day by the same subject), crisps, an orange flavoured drink and one portion of the mousse with the supplement. All food was weighed before being given to the subjects and on the first visit subjects were able to regulate their intake from the offered food (except for the mousse which they were required to fully consume on all visits). Whatever was not consumed was weighed, and the subjects were then required to consume an identical amount on the subsequent visit to ensure the energy and macronutrient intake was identical with only the presence of the RS differing. A similar study design has been used successfully in previous studies(Reference Weststrate and van Amelsvoort12). The mean values for the amounts consumed at lunch are shown in Table 2.
At the end of the intervention (420 min) participants were placed in individual areas and provided with a large pre-weighed ad libitum homogenous test meal, in excess of normal portion sizes (the whole dish provided: 9765 kJ energy, 81·5 g protein; 339·1 g carbohydrate; 70·0 g fat; 15·9 g fibre). Subjects were instructed to consume freely until comfortably full and then the leftovers were weighed. The subjects were informed that they could take home any leftover test meal in order to prevent over-consumption. The ad libitum test meal was a pasta-based meal which was made to a standard recipe with standard cooking times; however, while the energy content was identical on both visits, the energy density per gram would have varied depending on the amount of water absorbed during cooking. Therefore, this was taken into consideration when the energy intake was calculated for each visit.
Overall 24-h intake on the study days was assessed from weighed intakes from the breakfast, lunch and ad libitum dinner provided and from diet diaries completed during the late evening by subjects after they had left the unit. All dietary analysis was performed using WinDiets Professional Version program (Robert Gordon University, Aberdeen, UK). Bowel habit diaries were completed on the day of the study and the following day for assessment of gastrointestinal tolerance.
Whole blood for glucose analysis was collected into sodium oxalate tubes, for insulin analysis into potassium EDTA tubes and for C-peptide analysis into potassium EDTA tubes with 200 kallikrein inhibiting units (KIU) aprotinin per ml of whole blood (Trasylol; Bayer, Newbury, UK). All samples were centrifuged for 10 min at 3000 rpm and then plasma aliquots were stored at − 20°C until batch analysis to reduce inter-assay variation. Plasma glucose was measured enzymatically using a commercially available kit (Instrumentation Laboratory, Warrington, UK) for the ILab650 (Instrumentation Laboratory) and the inter-assay variation was < 2 %. Concentrations of plasma insulin and C-peptide were measured by RIA with commercially available kits (Millipore; Watford, UK). The sensitivity of the insulin assay was 12 pmol/l (2 μU/ml) and for the C-peptide assay was 0·1 ng/ml, with an inter- and intra-assay variation of < 10 %.
Calculations and statistical analysis
Fasted insulin sensitivity and β-cell function (homeostasis model assessment (HOMA) % S and HOMA % B, respectively) were assessed by HOMA(Reference Matthews, Hosker and Rudenski13) at the beginning of each study morning; while postprandial insulin sensitivity was assessed using the minimal model method described by Caumo et al. (Reference Caumo, Bergman and Cobelli14).
Area under the curve was calculated for glucose, insulin and C-peptide, using the trapezoid method. The ratio of C-peptide to insulin was calculated using the area under the curve for 2 h after each meal for both measures and used as a marker of hepatic insulin clearance. Time course data were analysed by repeated measures ANOVA. The data were normally distributed and paired t tests were used to compare between the groups. Statistical analyses were carried out using SPSS version 12.0.1 for Windows (SPSS, Inc., Chicago, IL, USA), with significance assumed as P < 0·05. All the results are means with their standard errors.
Both the supplements were well tolerated by the subjects with no adverse gastrointestinal effects reported on the day of the study or the following day.
The fasting insulin sensitivity and β-cell function (assessed by the homeostatic model assessment(Reference Matthews, Hosker and Rudenski13)) were not significantly different at the start of each of the study days (Table 3), which confirms that the subjects were in a similar metabolic state at the start of each study day.
HOMA % S, fasted oral insulin sensitivity, assessed by homeostasis model assessment; HOMA % B, β-cell function, assessed by homeostasis model assessment; AUC, area under the curve; S I, oral insulin sensitivity to each meal, calculated with a minimal model.
* Comparisons were made with a paired samples t test.
Energy and macronutrient intake
Supplementation with 48 g RS over two meals resulted in a significantly lower energy intake, 5241 (sem 313 kJ) at the offered ad libitum test meal compared to the energy intake seen with the placebo, 5606 (sem 345 kJ; P = 0·033).
Over the whole 24-h period there was also a significantly lower energy intake following the 48 g RS supplement compared to the placebo supplement, from 12 603 to 13 949 kJ (P = 0·044; Table 4). With the RS leg the mean energy intake was 104 % of calculated habitual energy requirements (calculated with the Schofield equation(Reference Schofield, Schofield and James15) and a moderate activity level of 1·6 for all subjects) compared to 116 % for the placebo leg.
* Comparisons were made with a paired samples t test.
The lower 24-h energy intake appeared to be mainly due to a significantly lower fat intake, 13·3 g lower, following the RS supplement compared to the placebo (Table 4). The dietary fibre intake during this 24-h period was significantly different due to the supplementation with 48 g RS, and there did not appear to be a difference in fibre intake by the subjects once they left the investigation unit at the end of the postprandial study period.
Subjective appetite measures
There was no difference in the subjective appetite scores, measured by the visual analogue scales, for hunger (Fig. 1), fullness, prospective food consumption, thirst or desire for different foods (sweet, salty, savoury or fatty foods) between the two supplements.
The postprandial glucose concentrations were not significantly different between the RS supplement and the placebo supplement (Fig. 2(a)).
There was a significantly lower postprandial insulin response following the RS supplementation compared to the placebo supplement over the whole acute study period (P = 0·029; Fig. 2(b)). However, the corresponding C-peptide concentrations were not significantly different between the two supplements. Consequently, there was an increase in molar ratio of C-peptide to insulin as a surrogate marker for hepatic insulin clearance (Table 3); however, there was no significant difference in postprandial oral insulin sensitivity between the two supplements at either meal (Table 3).
The present study found that after consumption of 48 g RS (split equally over the test breakfast and lunch meals), there was a lower energy intake at both the ad libitum test meal and over the whole 24-h period, without an associated effect on subjective appetite ratings; the study also found a significant effect of the RS supplement on lowering the postprandial insulin response. To our knowledge this is the first study where RS has been provided to participants as part of a mixed meal and compared to a placebo, where available carbohydrate and energy load have been matched.
Previous studies investigating the effects of RS on appetite have replaced proportions of carbohydrate with RS and therefore the amount of glycaemic carbohydrate provided has varied between the supplements which would confound the interpretation of the results. Indeed, in one study consumption of RS appeared to cause a reduction in subjective feelings of satiety(Reference Raben, Tagliabue and Christensen16); however, a limitation of the study was that the RS and the corresponding 100 % digestible starch were mixed into fruit syrup drinks as a means of delivering the starches, which resulted in the two supplement drinks having a different texture, one liquid and the other semi-solid, with liquids being known to be less satiating than solid foods(Reference Mourao, Bressan and Campbell17); this study also matched the supplements by weight of starch and therefore the supplements differed in energy and available carbohydrate content. In another study by de Roos et al. (Reference De Roos, Heijnen and de Graaf18) two types of RS (RS type 2 and RS type 3) were compared to glucose and, while the supplements were matched for total carbohydrate content, the proportion of available carbohydrate was lower in both of the RS supplements. The study also found that after consumption of each product for 1 week, there was little effect of the RS on appetite. Other studies have investigated the effects on appetite of varying the amylose to amylopectin ratios. One of these studies found that immediately after the meal the high level of amylase was more satisfying than the low amylase but that the high amylose was also the least palatable(Reference van Amelsvoort and Weststrate19). Another study that varied amylose and amylopectin ratios, found no significant effect between their treatments on visual analogue scale ratings(Reference Weststrate and van Amelsvoort12).
In the present study, on both study days, the participants consumed more than their estimated requirements (104 % on the RS leg and 116 % on the placebo leg) and therefore there was likely to be an element of over-consumption as the participants were given food and did not have to prepare it for themselves; however, the individuals still appeared to over-consume less on the RS leg than on the placebo leg and this can be directly attributed to the RS as the only difference between the test meals. However, this over-consumption may explain why no effects were found in any of the subjective appetite ratings. The overall 24-h intake was derived from the weighed food given to participants at the three meals within the unit and from the reported participant intake later in the evening, so only a small contribution to total intake was obtained indirectly, from participant self-reporting. When using the Goldberg cut-off(Reference Goldberg, Black and Jebb20) to identify under-reporting none of the participants would have been classified as under-reporters for either leg of this intervention and so the results are likely to be representative of true intake. All meals were based on standard portion sizes rather than set to BMR, as one of the main outcomes of the study was an effect on the metabolic response, which would have been masked if intakes were set to BMR. The only difference between the two legs, apart from the 48 g of RS, could have been a small difference in the energy density of the meals due to the presence of the RS; however, this was minor (for breakfast and lunch combined the density per gram was 7·2 kJ/g for the RS leg and 7·7 kJ/g for the placebo leg) and the glycaemic load of the two legs was identical.
The lower energy intake seen in the present study over 24 h appeared to be mainly explained by a significantly lower fat intake following the RS supplement. However, the reason for this lower fat intake is unclear, as the participants did not report a difference in desire for fatty foods on the subjective appetite ratings during the postprandial study. This difference in macronutrient choice following RS consumption would require further investigation to determine whether it was an incidental finding or a true effect. There also did not appear to be an influence of the high RS intake on fibre intake later in the day; however, there may have been an effect on the following day's fibre intake that was not monitored in the present study.
Although in the present study a reduced energy intake was found following consumption of the RS supplement, it is not possible from the design of the present study to determine exactly what the mechanisms are for the observed effect on food intake. However, as the main effect on food intake appeared to occur at the ad libitum dinner and then later in the day once the participants had left the unit, a possible mechanism could be the fermentation in the colon by colonic microflora and the subsequent effects of SCFA, which has been hypothesised as a mechanism for the effects of fibre on appetite(Reference Pereira and Ludwig3).
While the metabolism of SCFA may result in potential additional energy being provided to the body, this was not accounted for in the present study and no value was assigned for the fermentation of the RS. The actual amount delivered to each participant is likely to vary, dependant on the extent of fermentation and metabolism within the colon(Reference Wolever, Schrade and Vogt21) and variations in individuals' gut transit time(Reference Macfarlane and Macfarlane22), as well as a loss to the faeces of a proportion of the ingested RS and the SCFA produced. Published works have stated an energy value of 8 kJ/g for Hi-Maize® 260(Reference Behall and Howe23) accounting for SCFA metabolism compared to the 16 kJ/g for the placebo. In the present study, we matched the supplements by glycaemic load as the largest confounder to the glycaemic response. When taking into consideration the amounts of each supplement given this would have resulted in maximal energy doses of 640 kJ for the RS and 512 kJ for the placebo, with a difference of only 128 kJ. This therefore does not account for the full difference in energy intake observed in the study over the 24-h period which was 1346 kJ.
It is possible that the increase in production of SCFA may consequently increase production of anorexigenic hormones from the colon, such as peptide YY(Reference Cuche, Cuber and Malbert24, Reference Cherbut25). However, so far the only evidence of an effect of RS on these hormones, particularly peptide YY and glucagon-like peptide, has been shown in rodent studies(Reference Keenan, Zhou and McCutcheon26–Reference Zhou, Martin and Tulley28), which are, in terms of the gastrointestinal tract, anatomically different(Reference Topping and Clifton29) to human subjects. Human studies which have measured gastrointestinal peptides have found no effect of fibre feeding(Reference Robertson, Bickerton and Dennis6, Reference Raben, Tagliabue and Christensen16).
As insulin and C-peptide are co-secreted, the lower postprandial insulin concentration detected was likely to be due to increased hepatic insulin clearance, as there was a trend towards significance between the two supplements for the molar ratio of C-peptide to insulin. An increase in hepatic insulin clearance has previously been reported following RS intake over 24 h(Reference Robertson, Currie and Morgan30). It has been proposed that the increase in production of SCFA and their exposure to the liver may ultimately be responsible for the increase in insulin clearance.
The lower energy intake seen in the present study following RS consumption as part of a mixed meal could have beneficial implications in weight management and, potentially, weight loss; however, further studies are required to confirm whether a similar finding is seen in other population groups such as the overweight or obese, and to determine the actual mechanisms for the effect. Although the dose in the present study was well tolerated over the one day of the study and a lower energy intake was observed, further investigations are needed to establish whether the dose has similar effects when ingested chronically. A lower postprandial insulin response was also observed which could be explained by an increase in hepatic insulin clearance. Increased intakes of RS in the diet may therefore have beneficial implications in weight management.
We thank all of the participants for their time. C. L. B. was involved in designing the study protocol, recruitment of volunteers, conducting the study days, analysis of data and writing the paper. G. S. F. was involved in obtaining the funding, the protocol and refining the paper. M. D. R. was involved in designing the study protocol, assistance on several of the study days and analysis of data, and refining the paper. There were no conflicts of interest. C. L. B. was supported by an educational fellowship from Premier Foods and supplements were supplied by National Starch Company, LLC.