Overview

The aim of the present review is not to consider glycaemic index (GI) in general but to debate the relationship between this method of classifying carbohydrate and appetite regulation. The term GI was originally introduced to classify different sources of carbohydrate (CHO) by their effect on post-meal glycaemia⁽Reference Jenkins, Wolever and Taylor¹⁾. The aim was to improve glycaemic control in individuals living with diabetes. The glycaemic response to foods and its role in glycaemic control in individuals living with diabetes has been reviewed a number of times⁽Reference Thomas and Elliott²⁾, but relatively little consideration has been given to its potential impact on satiety and long-term regulation of body weight. In particular, there is no clear evidence on whether low-GI foods or diets may be used as part of a strategy to reduce food and energy intake and provide a tool to manage obesity. Several intervention studies have shown that low-GI-CHO diets have practically no impact on body weight compared with isoenergetic high-GI-CHO diets⁽Reference Livesey³⁾. A review of several ad libitum studies has concluded that low-GI-CHO v. high-GI-CHO diets eaten ad libitum result in a lower body weight⁽Reference Livesey³⁾. This conclusion is also borne out by a systematic review, which has demonstrated an impact on body weight with the introduction of a low-GI diet⁽Reference Thomas, Elliott and Baur⁴⁾.

Glycaemic index of foods or diets

GI as originally defined is the indexing of the glycaemic response to 50 g available CHO from a test food to the same amount of available CHO from glucose⁽Reference Jenkins, Wolever and Taylor¹⁾. In practice, it corresponds to the incremental area under the blood response curve measured over 2 h of a 50 g CHO portion of a test food expressed as percentage of the response to the same amount of CHO from a standard food consumed by the same subject⁽Reference Mann, Cummings and Englyst⁵⁾. There are now a number of systematic reviews on the effect of GI in the management of diabetes and CHD⁽Reference Thomas and Elliott^2, Reference Mann, Cummings and Englyst^5, Reference Kelly, Frost and Whittaker⁶⁾.

Glycaemic index and satiety

Two recent systematic reviews have been conducted in this area⁽Reference Livesey^3, Reference Thomas, Elliott and Baur⁴⁾. They have examined the world literature in the case of effects on energy homeostasis in both ‘short-term’ studies (duration of 1 d) and ‘long-term’ studies (several days or weeks duration). Of the twenty-five short-term studies selected by the quality criteria of the systematic review most of the studies were randomised and satiety was assessed either by a subjective method based on visual scales or by an objective method using a preload test meal design. This methodology consists of either measuring the spontaneous energy intakes when a meal is served ad libitum at a given time after the consumption of the test foods or by measuring the time of spontaneous request for food and the corresponding energy intakes when snacks are provided ad libitum on request. The foods or meals tested in the studies were pure CHO, CHO foods or mixed meals. The findings are summarised as follows:

1. 1. human studies testing pure CHO: an early short-term satiating effect (within the first hour) has been observed with high-GI CHO. An inverse association has been reported between the blood glucose response and the subjective appetite and food intake in the 1 h after consumption of isovolumetric preloads of CHO⁽Reference Anderson and Woodend⁷⁾. At 1 h after their consumption high-GI CHO (glucose, polycose and sucrose) suppress food intake whereas low-GI CHO (amylose, amylopectin and a fructose–glucose mixture) do not. The early satiating effect of high-GI CHO has been confirmed by the results of several other studies showing that high-GI CHO such as glucose and sucrose suppress short-term food intake 60–90 min after consumption of the preload. A later short-term satiating effect has been reported with low-GI CHO such as fructose. In the studies in which both fructose and glucose have been tested fructose produces an increased satiety compared with glucose at 135 min after consumption. These findings are corroborated by the fact that most studies showing increased satiety after the consumption of low-GI foods but not high-GI foods have observed this effect at 2–6 h after the ingestion of the preload⁽Reference Anderson and Woodend⁷⁾;

2. 2. human studies testing CHO foods or mixed meals: the overall view is that whatever the method used to assess satiety (subjective or objective method) and despite the possible contribution of some confounding factors (e.g. fibres, palatability) more than half the reviewed studies support an increased short-term satiety with low-GI foods or meals compared with high-GI foods or meals⁽Reference Livesey³⁾. A short-term study in children has demonstrated that a low-GI breakfast reduces intake at lunch⁽Reference Warren, Henry and Simonite⁸⁾.

Effect of the consumption of low-glycaemic-index diets on food intake and body-weight regulation

Only a few studies have attempted to assess the effect of the lowering of glycaemic response on satiety and effects on body weight in the long term. A systematic review has shown inconsistent results, suggesting that the improved satiety observed in certain short-term studies with low-GI foods or meals does not lead to a long-term reduction in spontaneous energy intake and/or regulation of body weight⁽Reference Bornet, Jardy-Gennetier and Jacquet⁹⁾. This notion has been recently underlined by a study that has found no effect on appetite, energy intake and weight loss in volunteers randomised to a low-GI diet⁽Reference Aston, Stokes and Jebb¹⁰⁾. Furthermore, a long-term intervention study has failed to show the appetite regulation found in short-term studies of children exposed to a low-GI diet⁽Reference Henry, Lightowler and Strik¹¹⁾. However, an overall systematic review does suggest an overall positive effect of low GI on weight loss⁽Reference Thomas, Elliott and Baur⁴⁾.

Factors confounding the results

There are many factors that may be affecting the results of these long-term studies. The most obvious one is the heterogeneity in foods classified at different GI.

Carbohydrates: the complex group

From simple sugars such as glucose to complex polysaccharides such as amylopectin and cellulose the term ‘dietary CHO’ embraces a vast array of molecules. Moreover, CHO in unprocessed food have a physical structure such as the cell wall, whereas processed foods may present CHO for which this structure is lost. It is this diversity of not only chemical structure but also physical form that proves problematic in categorising CHO, even before physiological effects are discussed⁽Reference Mann, Cummings and Englyst^5, Reference Brouns, Bjorck and Frayn¹²⁾. Even within two low-GI diets the interaction between the GI tract and the different CHO making up the diet can be very different despite the similar glucose profile. One major component is fermentable CHO (FCHO). The chemical make-up and the food structure in which a CHO is delivered to the small bowel determine the rate of absorption and the section of the GI tract at which products of CHO digestion are absorbed. Most simple CHO and starch are absorbed rapidly in the small bowel⁽Reference Kien, Heitlinger and Li¹³⁾. However, molecules with a 1–4-β link or retrograde starch cannot be digested by amylase so survive the small bowel; they are processed by the microbiota in the colon and the products of this fermentation are absorbed and have a metabolic effect⁽Reference Kien, Heitlinger and Li¹³⁾. For the purpose of the present review the focus will be on CHO that are malabsorbed in the small intestine and are fermented by the colonic microbiota. For ease, they will be termed FCHO.

Body composition and fermentable carbohydrates

The first observation that a diet high in FCHO may have an influence on body composition was made in a study of rats fed a high-resistant-starch diet⁽Reference Lerer-Metzger, Rizkalla and Luo¹⁴⁾. By week 5 of this diet total body weight was found to be the same as that of the control group, but with smaller epididymal fat pads and adipocyte size. This observation was taken further by the demonstration that in rats on a diet high in resistant starch there is no reduction in whole-body weight but there is a reduction in adipose tissue, which is associated with a decrease in insulin and glucose response to a standard glucose load⁽Reference Pawlak, Kushner and Ludwig¹⁵⁾. Studies in mice have confirmed these observations and have demonstrated that a diet high in resistant starch leads to decreased hepatic cellular lipid content along with an increase in adipose tissue insulin-stimulated glucose uptake⁽Reference So, Yu and Kuo¹⁶⁾. This finding suggests that FCHO have a central role to play in adipocyte metabolism and body composition in animals.

In human subjects there are very few data. It has been demonstrated that low-GI diets that are high in FCHO have an influence on adipocyte metabolism. In individuals with CHD and in women with and without risk of coronary disease 3–4 weeks exposure to a low-GI diet increases whole-body insulin sensitivity and insulin-stimulated glucose uptake in the adipocyte⁽Reference Frost, Keogh and Smith^17, Reference Frost, Leeds and Trew¹⁸⁾. This finding suggests that in human subjects low-GI diets with a high FCHO content could have a central metabolic role in body composition and adipocyte metabolism. There is one study that suggests that there is adipose tissue remodelling in human subjects following 5 weeks on a low-GI diet⁽Reference Bouche, Rizkalla and Luo¹⁹⁾.

Obesity and fermentable carbohydrates

Obesity rates have greatly increased in the last 50 years. This period has also seen marked changes in diet. In particular, the quantity and quality of dietary CHO has changed considerably. While the total amount of CHO in a typical Western-style diet has moderately increased, the amount of FCHO has fallen dramatically. FCHO in the current Western diet are derived predominantly from cereal grains, and in many products are refined to increase their digestibility. Thus, contemporary diets contain much lower levels of FCHO compared with intake 50 years ago, which contained a much more diverse range of unrefined FCHO⁽Reference Eaton, Eaton and Konner²⁰⁾. Recent epidemiological and experimental studies have demonstrated an inverse correlation between dietary fibre intake and body weight and adiposity⁽Reference Appleby, Thorogood and Mann^21–Reference Tucker and Thomas²⁸⁾. As outlined earlier, animal experiments have consistently demonstrated that FCHO influence body fat composition, reducing intra-abdominal adipose tissue and hepatic lipid load and increasing insulin sensitivity. Recent studies have shown that large amounts of orally-administered FCHO in food-deprived animals result in a central nervous system neuronal activation pattern similar to that observed in fully-satiated animals compared with animals fed a low-FCHO diet. This neuronal activation pattern appears similar to that observed following infusion of the anorexigenic gut hormones peptide YY (PYY) and glucagon-like peptide-1. Animals fed high doses of FCHO have higher plasma concentrations of glucagon-like peptide-1 than those fed low-FCHO diets⁽Reference Cani, Dewever and Delzenne²⁹⁾, suggesting that the changes in neural activation observed may be a result of changes in gut hormone release. The levels of FCHO used in these experiments are often between 7% and 30% of the total weight of the diet. High intake of FCHO has also been shown to improve insulin sensitivity in human subjects. A 1-year study has shown that children consuming a diet supplemented with the FCHO inulin (oligofructose) have greater weight loss than unsupplemented controls⁽Reference Abrams, Griffin and Hawthorne³⁰⁾.

Fermentation and appetite regulation

Over the last few years the role of the gut microbiota in adipose tissue development and energy homeostasis has been championed⁽Reference Backhed, Ding and Wang^31–Reference Ley, Backhed and Turnbaugh³⁴⁾. In the main, the observations have been made using a germ-free mouse model. It has been demonstrated that colonisation of the large bowel of germ-free animals by microbiota from conventionally-raised animals results in a 60% increase in body fat content, suppression of fasting-induced adipocyte factor leading to an increased activation of lipoprotein lipase and increased storage of lipid in adipocytes⁽Reference Backhed, Ding and Wang³¹⁾. Through a series of eloquent experiments the same investigators have suggested that obese human subjects⁽Reference Ley, Turnbaugh and Klein³²⁾and animals⁽Reference Turnbaugh, Ley and Mahowald^33, Reference Ley, Backhed and Turnbaugh³⁴⁾ display an increase in firmicutes in the microbiota that increase fermentation and so enhance energy uptake in the form of SCFA.

In many ways these observation are contradictory to those reported with increasing intake of FCHO in animals and human subjects. The consensus of these studies suggests that supplementing animal feed with FCHO leads to an increase in gut microbiota species that increase fermentation and so increase SCFA production in the large bowel⁽Reference Kleessen, Hartmann and Blaut^35–Reference Cresci, Orpianesi and Silvi³⁷⁾. Indeed, the animals seem to have a lower body fat content and when fed obesity-inducing diet they are protected against weight gain⁽Reference Cani, Dewever and Delzenne^29, Reference Cani, Daubioul and Reusens³⁸⁾. The adipocytes of these animals tend to be smaller and more insulin sensitive. Recent data from a human study show that long-term feeding of FCHO results in weight loss and an enhanced release of the anorectic gut hormone glucagon-like peptide-1⁽Reference Freeland, Wilson and Wolever³⁹⁾. Thus, it could be suggested that another factor comes into play.

SCFA, GPR43, L-cells and appetite regulation

GPR43 is a SCFA receptor. In 2003 it was demonstrated that SCFA act as ligands for GPR43⁽Reference Le Poul, Loison and Struyf⁴⁰⁾ (formerly orphaned G-protein coupled receptor). The GPR43 receptors are present on the luminal side of the L-cell in the rodent and human colon⁽Reference Karaki, Mitsui and Hayashi^41, Reference Karaki, Tazoe and Hayashi⁴²⁾. GPR43 has a particularly high affinity for the SCFA propionate⁽Reference Brown, Goldsworthy and Barnes⁴³⁾, which is produced in the colon by the fermentation of CHO. SCFA produced during fermentation in the colon may be responsible for the changes in body composition and insulin sensitivity observed following FCHO administration in animals. In particular, colonic production of the SCFA propionate appears to play a critical role. Propionate is believed to have direct effects on L-cells in the colon and systemic effects at the level of the adipocyte⁽Reference Brown, Goldsworthy and Barnes^43, Reference Ge, Li and Weiszmann⁴⁴⁾.

The L-cell, the most abundant endocrine cell in the intestine, synthesises and releases the anorexigenic gut hormones glucagon-like peptide-1, PYY and oxyntomodulin. These three hormones are released in the physiological response to food and have been shown to inhibit food intake in animals and human subjects⁽Reference Druce and Bloom⁴⁵⁾. Bariatric surgery is thought to cause sustained weight loss by increasing endogenous anorectic gut hormone secretion⁽Reference Olivan, Teixeira and Bose⁴⁶⁾. Propionate induces activation of GPR43 on enteroendocrine L-cells and stimulates PYY⁽Reference Karaki, Mitsui and Hayashi⁴¹⁾ release in vitro. In animals FCHO ingestion stimulates L-cells to release glucagon-like peptide-1 and PYY, resulting in a reduction in food intake⁽Reference Cani, Hoste and Guiot⁴⁷⁾. This effect is replicated by propionate in several mammalian species, suggesting that propionate production is at least in part responsible for mediating the effects of FCHO on appetite and gut hormone release. GPR43 receptors have recently been shown to be present on PYY3–36-producing enteroendocrine L-cells in the human colon⁽Reference Tazoe, Otomo and Kaji⁴⁸⁾. Few human studies have demonstrated the effect of FCHO on appetite⁽Reference Willis, Eldridge and Beiseigel^49, Reference Cani, Joly and Horsmans⁵⁰⁾, which could be because the high concentration of SCFA is needed to trigger a gut hormone response. The normal SCFA concentration in the colon is 100 mm and the release of PYY occurs at 300 mm⁽Reference Cherbut, Ferrier and Roze⁵¹⁾, which may explain the differences between the consistent animal observations that feed containing >7% (w/w) of a FCHO produces high concentrations of PYY, reduction in body weight, reduction in adipose tissue and improved insulin sensitivity. In human studies levels are often ⩽15 g/d (<1% total weight of food consumed). Levels of 300 mm are only likely to be reached following high FCHO intake similar to those of man's ancestors of 100 g/d⁽Reference Cherbut, Ferrier and Roze⁵¹⁾. This factor has been highlighted in a recent study in which weight loss has been reported following a high intake of fructans (21 g/d), which is associated with an increase in the gut hormone PYY⁽Reference Parnell and Reimer⁵²⁾.