ARTICLE IN PRESS

Clinical Nutrition supplements (2004) 1, 5 17

http:/ /intl.elsevierhealth.com/journals/clnu

Dietary carbohydrates and health: do we still need the bre concept?

John H. Cummings, Laurie M. Edmond, Elizabeth A. Magee
Department of Pathology and Neuroscience, Ninewells Hospital and Medical School, Dundee, DD1 9SY, UK

KEYWORDS
Non-starch polysaccharides; Resistant starch; Oligosaccharides; Carbohydrate; Historical

Summary Dietary bre, principally the non-starch polysaccharides of the plant cell wall, is an important component of our diet. After more than 30 years of research into the many and varied claims for its benets, it is now clear that bre has uniquely signicant physical effects in the gut and in addition through fermentation is a major determinant of large bowel function and bowel habit. Its physical properties in the small bowel effect lipid absorption and the glycaemic response. Fibre has some modest effects on appetite. These benets feed through into a protective role in large bowel cancer, diabetes and coronary heart disease. Equally important, bre has led us in our thinking about other dietary carbohydrates. Work on bre was instrumental in the discovery of both resistant starch and the prebiotic oligosaccharides. Resistant starch of itself is not so important as bre although does provide a substrate for fermentation and may be a good source for butyrate production. However, the discovery of resistant starch led to a revision of our understanding of the digestion of starch in the gut and the important contribution this makes to glycaemic response. The prebiotic oligosaccharides have been the newest and most exciting development in this story. Whilst again they are substrates for fermentation, their effects on bowel habit are negligible but their capacity to alter selectively the composition of the large bowel ora is a novel and highly signicant development in our understanding of the intestinal microora. The clinical benets of this have yet to be established but will almost inevitably stem from the improvements to colonisation resistance that these changes bring. The dietary carbohydrates, therefore, which include NSP, RS and the oligosaccharides, are a complex group of substances with diverse physiological and health benets. Instead of considering them to be a single macro nutrient component of the

Abbreviations: NSP, non-starch polysaccharides; RS, resistant starch; SCFA, short chain fatty acids; ANOVAR, analysis of variance; Transgalacto, transgalactooligosaccharides; Fructo, fructooligosaccharides; Galacto, galactooligosaccharides Corresponding author. Department of Molecular and Cellular Pathology, Ninewells Hospital and Medical School, Dundee, DD1 9SY, UK. Tel.: +44 1382 632425; fax: +44 1382 633952. E-mail address: j.h.cummings@dundee.ac.uk (J.H. Cummings). 1744-1161/$ - see front matter & 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.clnu.2004.09.003

ARTICLE IN PRESS
6 J.H. Cummings et al.
diet, we should, as we do for dietary fat, start to look at their individual contributions to physiology, epidemiology and clinical benet. & 2004 Elsevier Ltd. All rights reserved.

Introduction
The dietary bre hypothesis was one of the most compelling in nutrition and public health in the latter half of the twentieth century. It provided the stimulus to a great deal of research, both epidemiological, physiological, analytical and technical. It has been the catalyst for progress in our understanding of the cause of a number of common diseases, especially those of the large bowel and diabetes and has given governments and the food industry valuable targets for healthy eating. It is now 30 years since Burkitt noted that the close relationship between bowel cancer and other non-infective diseases y and y their close association with the rened diet characteristic of economic development suggests that the removal of dietary bre may be the causative factor. This hypothesis was subsequently expanded by Trowel, Walker and others to encompass many of the major diseases of Western culture including coronary heart disease, obesity, diabetes, gallstones and other bowel disorders. None of these early pioneers actually measured bre in the diet or intakes in any population, and whilst proposing an exciting concept, they left us with a confusion of denitions and methodologies. Since then, nutritional science has progressed, especially our understanding of dietary carbohydrates and with this the apparent central role of bre in many physiological processes and in disease prevention has weakened. To see bre in its true perspective today, we need to start with a look back at how ideas have developed.

Historical perspective
Although much had already been written about bre, the rst epidemiological observations of importance were probably those of Higginson and Oettle in 1960 who noted that constipation and bowel cancer were rare in Africa whilst being common in the West, and that in Africa a large amount of roughage is consumed y and stools are bulkier and more frequent y.1,2 In 1971, Burkitt3 suggested a specic hypothesis relating lack of dietary bre to the cause of bowel cancer and a mechanism whereby bre, through its capacity to regulate the speed of transit, bulk

and consistency of stool y and effect microbial metabolism was able to prevent bowel cancer. About this time, Trowell, inuenced by his experience in Uganda, which he had summarized in his book Non-infective diseases in Africa,4 and by reading Cleaves The Saccharine Disease,5 suggested that a number of other chronic Western diseases such as obesity, diabetes, coronary heart disease, gallstones, etc might also be due to consumption of bre-depleted foods. Concurrently, Painter, working in Oxford, proposed that diverticular disease was due to colonic stasis induced by lack of bre.6,7 Burkitt and Walker then added varicose veins and haemorrhoids to the list of bredeciency disorders.8 Parallel with these epidemiological observations was the development of knowledge of the physiological properties of bre in the gut. The polysaccharides of the plant cell wall were thought not to be digested in the upper intestine of man, but to be fermented in the large bowel by bacteria and thus increase stool output, shorten transit time, alter bile acid metabolism and produce volatile fatty acids.9 A number of animal studies had shown cholesterol lowering and atheroma preventing effects of puried bre extracts and in man the benecial effect of oats10 and pectin11 was already known. Advances were also occurring at this time in methods for determination of dietary bre in the laboratory,1214 most notable of which were those of Southgate,12 and in animal nutrition, of Van Soest.15 The convergence of these lines of evidence gave birth and credence to the dietary bre concept and stimulated Burkitt and Trowell to prepare to write a book.16 Burkitt wrote to Trowell asking him to dene dietary bre. Trowell thought this would be simple but could not nd bre dened in any textbooks of medicine or nutrition.17 Some food tables listed values for the crude bre content of foods. Crude bre, a gravimetric method used to analyse animal foods, had been in use for over 150 years. However, Van Soest and colleagues had already shown that this method failed to recover the majority of the plant cell wall material.15 In the United Kingdom, the Medical Research Councils Composition of Foods18 contained data on unavailable carbohydrate, a term used by McCance and Lawrence in 192919 and earlier by

ARTICLE IN PRESS
Carbohydrates and health Atwater. In trying to prepare accurate food tables for diabetic diets, McCance and Lawrence tried to distinguish between available carbohydrate consists of starch and soluble sugars y and the y unavailable, mainly y hemicellulose and bre (cellulose). Trowell, however, wanted a generic term for the undigested plant cell wall that he thought was important to health and so he decided to redene bre.17 The earliest formal description of dietary bre was, therefore, by Trowell who dened it as the proportion of food which is derived from the cellular walls of plants which is digested very poorly in human beings.20 This was largely an attempt to distinguish it from crude bre. It was a physiological-botanical denition which did not identify the precise nature of dietary bre nor dene exactly the limiting characteristics of substances that should be included or excluded in the description, apart from the reference to plant cell wall material. Moreover, there was no method specied. Subsequently, Trowell discovered that Hipsley21 had used the term dietary bre in an article on pregnancy toxaemia, stating it was cellulose, hemicellulose and lignin. However, his paper contained data on crude bre. After visiting the USA in 1974, Trowell22 enlarged his denition of dietary bre to include structural polysaccharides, lignins, unavailable lipids, nitrogen, trace elements and cell wall enzymes and mineral salts. This was really an acknowledgement that the concept of bre and disease could not be ascribed to any particular cell wall component at that time. In 1976, in conjunction with Southgate, Wolever, Leeds, Gassull and Jenkins, bre was redened as the plant polysaccharides and lignin which are resistant to hydrolysis by the digestive enzymes of man.23 This denition was specically chosen so that polysaccharides structurally related to plant cell wall polysaccharides would be included in the denition. These isolated polysaccharides were being used experimentally to investigate the physiological effects of dietary bre. More pragmatically, the isolated polysaccharides were also analytically indistinguishable from the corresponding plant cell wall components. Since 1976, both Trowell and many others have dened and redened dietary bre. Even today, 30 years later, the debate on denition continues.24,25 It is, however, a largely sterile debate because the world has moved on. The alleged benets of bre in the diet are hard for the consumer to see; antiquated methodology, by nutritional standards, still persists and has prevented international 7 agreement and dietary recommendations. For example, in a recent publication that included a table of the bre and lignin values of many foods, reference to more than 25 methods are given in a footnote to the table.26 Exciting new concepts of pro and prebiotics are providing more scope than bre for food manufacturers and in nutritional science. What has emerged from analytical, physiological and epidemiological studies is that the health benets of bre are conferred primarily by the polysaccharides of the plant cell wall, namely cellulose, hemicellulose and pectin as a result of their chemical and physical properties which in turn lead to specic physiological changes. No other plant cell wall component has been shown to be important in man in the context of preventing Western diseases, although other dietary constituents such as resistant starch, oligosaccharides and lignin may contribute to or modify these effects. The cell wall polysaccharides are a chemically distinct group of carbohydrates, which do not contain the alphaglucosidic bonds present in starch and they have been called non-starch polysaccharides (NSP).27,28 The amount of NSP in the diet is quantitatively related to bowel function.29 This latter property has formed the basis for dietary recommendations for bre intakes in both the UK and WHO recently.3032 So where does the dietary bre hypothesis stand today? The early papers of Burkitt and colleagues have stimulated a great deal of research. Prior to 1970, there are very few papers on dietary bre, less than ve per year. By 1980, around 300 papers were being published annually on this topic, and in 1990, the number peaked at around 500. Today a web search for dietary bre papers will turn up about 400 hits. Is there anything unique or of such fundamental importance to health about bre, that requires us to single it out from other dietary carbohydrates?

Physiological and health effects of bre and other carbohydrates

Were it not for the discovery of resistant starch and, subsequently, the emergence of the prebiotic carbohydrates, bre might have survived as a denable dietary constituent with clear effects on colonic function and the glycaemic response. However, the identication of bre as one of the effectors of healthy gut function inevitably led to the search for other food components with like properties. The result has been a transformation of our understanding of carbohydrates in the diet and

ARTICLE IN PRESS
8
Table 1 Property Substrate for fermentation Established physiological properties and health benets of dietary bre.* Mechanism Microbial growth stimulated Short chain fatty acids Changes in nitrogen, bile acid and xenobiotic metabolism Gel properties Secondary affects on insulin secretion and gut hormones Chewing of food Delay in gastric emptying Related conditions Bowel habit/constipation Diverticular disease Colo-rectal cancer Glycaemic responses/diabetes Lipid absorption/coronary heart disease Short term appetite reduced

J.H. Cummings et al.

Physical effects in small bowel

Satiety and gastric emptying

*

Principally cell wall NSP and related polysaccharides.

the discovery of new and exciting concepts in gut biology. From a long initial list of benets or dietary bre, only a few remain. Table 1 shows three key areas where dietary bre can be shown to affect physiological function, and that might feed through to disease prevention.

Table 2 Short term studies of bre and appetite (n=18)) (From Burley and Blundell,34 Levine et al.,82 Delargy et al.83). Source N Amount Hunger Reduced No change Methylcellulose Guar/Bran Guar Psyllium Bran Alginate Fruit Mixed sources 3 1 5 1 3 1 2 5 13 12 28 22 6 2+ 1416 739 2 1 3 0 2 1 2 4 1 0 2 1 1 0 0 1

Appetite
Many studies have been reported of the effect of various sources of NSP on both short term appetite and satiety and long term weight changes. NSP-rich foods are generally less energy dense, more bulky and it is not unreasonable to suppose they may affect appetite. Table 2 summarises the results from 18 reports of 21 short term studies on appetite and show that the majority, 15 out of 21, observed a reduction in hunger. It is, of course, easy to criticize these studies because of the difculty in blinding either subjects or investigators to the food being consumed. Moreover, to allow valid changes in food intake under experimental conditions to occur, when appetite is subject to so many physiological, psychological and social controls, is difcult. Nevertheless, there is a feeling amongst researchers in this eld that NSP containing foods or supplements may be satiating. Turning short term satiety into long term weight control is another matter, and the results of such studies are less clear.33,34 A major problem in long term studies is, of course, compliance, particularly in the obese. Moreover, physiological adaptation may occur to changes in energy intake. There are no good epidemiological studies of obesity and NSP intake, but some potential mechanisms exist. NSP, by virtue of its physical properties, may delay gastric emptying, produce increased bulk in the upper gut and regulate carbohydrate and lipid absorption. However, in

the large intestine, the fermentation of NSP produces short chain fatty acids (SCFA), which in turn provide energy. Evidence from animal studies do not provide a role for SCFA in controlling appetite.35 Overall, therefore, bre containing foods or supplements show a reduction in appetite in short term studies but not consistent change in energy balance long term. Some possible mechanisms have been described affecting the upper gut but NSP is a source of energy through fermentation. Many other factors contribute to the control of appetite and weight. NSP is probably not the principal one. The way bre affects lipid and carbohydrate metabolism in the small bowel is unclear, although physical properties are key, since bre is not digested in any way.

Lipids and coronary heart disease

It is now well established that certain forms of bre can lower blood cholesterol in man. This was shown as far back as 196111 and the many subsequent

ARTICLE IN PRESS
Carbohydrates and health studies have been reviewed extensively.36,37 In summary, they show that soluble forms of NSP, such as pectin and guar gum and the soluble bglucans in oats, lower blood cholesterol in healthy and hyperlipidaemic subjects. Insoluble NSP, such as that present in wheat bran or pure cellulose, do not usually have this effect. The possible mechanisms whereby NSP may lower blood cholesterol are several and not yet conclusively decided. They include an effect of the water-soluble gel-forming materials in inhibiting cholesterol absorption in the small intestine, or affecting bile acid absorption and metabolism. Also soluble NSP is readily fermented in the large intestine producing short chain fatty acids, including propionic acid. There are some in vitro and animal and human studies to show that propionic acid may be hypocholesterolaemic.3842 An association between coronary heart disease and dietary bre was suggested in the 1950s.43,44 A number of prospective studies have been done of diet and heart disease and these show a consistently protective effect of bre. These include the studies of Morris et al.45 in London, Yano et al.46 in Honolulu, the Zutphen study.47 and the Boston-Irish study.48 In all these, some measure of total bre intake has been made and the coronary heart disease cases have had uni-association dietary patterns which reect high fat, low carbohydrate and vice-versa. In Morris et al.s study,45 the protective relationship was with cereals not vegetables and fruit, whilst the protective effect of crude bre in the Boston-Irish study48 was not signicant when adjusted for other risk factors. Similarly, in the Zutphen study,47 the signicant inverse relationship between coronary heart disease mortality and bre disappeared with multivariate analysis. This study did show a protective effect for polysaccharide consumption and, in both the Honolulu heart study46 and a study in Puerto Rico,49 starch intakes were apparently protective. Anderson and colleagues50 have recently reported a meta-analysis of studies of bre intake and risk of coronary heart disease. Sixteen of seventeen reports of total bre intakes reported a protective effect of which 14 were signicant. A signicant benet was also shown for whole grain cereals and for cereal bre whilst the strength of association with fruit and vegetable bre was much lower. More importantly, results were reported after adjustment for other risk factors such as smoking, hypertension, BMI etc. The relative risk for total bre was 0.73 (95% CI 0.650.83) and for whole grain 0.71 (0.480.94). Cereal bre alone was not signicant. 9 Coronary heart disease is a condition in which there are a number of cellular and pathological events that lead ultimately to coronary occlusion and often to death. The major risk factors are a raised blood pressure, raised serum cholesterol and smoking. A number of other processes also relate to risk in this condition including amount and type of fat intake, physical activity, stress and genetic susceptibility. It can, therefore, readily be seen that diet itself is only one of a number of causes contributing to risk of coronary heart disease and, furthermore, that bre is only one of the many dietary components which effect risk. Overall, therefore, with regard to bre and coronary heart disease, we have a multifactorial problem for which there is good evidence for a contribution from bre. Soluble forms of NSP usually lower blood cholesterol and there may be effects on the clotting system as well. Insoluble forms of NSP are not, however, generally regarded as hypocholesterolaemic. Prospective epidemiological studies show a protective effect for diets characterised by high bre intakes and a number of physiological mechanisms exist which relate these events. Fibre, or more probably a high bre diet, has a signicant contributory role in the aetiology of coronary heart disease.

Glucose, insulin and diabetes

To Trowell must go the credit for rst identifying a link between bre and diabetes,51,52 and to Jenkins and colleagues for publishing the rst experimental evidence in man that bre moderated blood glucose and insulin responses.53 Since then, recommendations for the diabetic diet have changed from a low carbohydrate high fat high protein diet to one moderately low in fat and high in starch and NSP.54,55 The subject has been extensively researched and evidence is summarised in recent reviews.56,57 Guar gum has probably been most studied and, as Wolever and Jenkins point out,57 an average reduction of 44% in blood glucose was seen in 15 studies of 24 groups of people with guar. In a similar meta-analysis, pectin reduced blood glucose by 29% and psyllium 29% also. Jenkins pointed out at an early stage that is was primarily soluble sources of NSP which formed gels, which were most effective in this context.58 However, events have overtaken the soluble dietary bre story in diabetes. Long term studies of soluble bre use highlighted a number of problems of acceptability, side effects, dose, etc. Equally good effects in diabetes could be achieved

ARTICLE IN PRESS
10 by diets high in starchy foods59 and the nature of these foods inuenced blood glucose.60,61 The concept of the Glycaemic index was born62 and a whole new area of nutritional science was developed, which has concerned itself with glycaemic responses to carbohydrate and other foods. The nature of the starch is the key and the subject of much current intensive research.63,64 Despite this interest in starch, there may still be a unique role for NSP in the control of carbohydrate metabolism. The early studies of viscosity and glucose responses pointed to the physical properties of NSP as being important. The classic study of Haber et al.65 with apples shows clearly that where carbohydrate, in this case sugar, is entrapped intracellularly in plant foods, its release in the gut is slowed and blood glucose and insulin responses lowered. A unique property of NSP, therefore, is that it forms the plant cell wall and thus a physical structure to foods. Numerous studies have now shown that the physical structure of starchy foods determines glycaemic index. NSP may, therefore, have a role to play in the regulation of carbohydrate metabolism through the physical structure it imparts to foods. The use of viscous soluble NSP isolates in this context will probably be seen as a stepping stone to understanding bre but not the ultimate goal. In trying to understand the dietary control of blood glucose and insulin, however, it is essential to start with a study of all carbohydrate, of which NSP is a part. J.H. Cummings et al. provides a means whereby energy can be salvaged from carbohydrate not digested in the small bowel, through absorption of SCFA. The physiological consequences of fermentation are diverse and several have important implications for human diseases. The best documented effect is that on bowel habit. Many hundreds of papers have been published on this subject and the key ones are summarised in.67 Table 3 is a compilation of faecal weight data from around 120 papers published between 1932 and 1992 detailing 150 separate studies. It shows the average increase in stool output expressed as grams of stool (wet weight) per gram of bre fed using weighted means from published data. Studies were included only if they contained quantitative data on changes in stool weight and had at least one control and one test period. Wheat as a source comes out as the most effective with raw bran at 7.2g/g more effective than cooked bran, 4.4 (Po0.05), but has the disadvantage of containing 3% phytatea known inhibitor of the absorption of divalent cations (calcium, magnesium, zinc and iron). Fruit and vegetables are remarkably effective, 6.0 g/g, and with wheat are well ahead of the rest. Whilst many people believe fruit and vegetables to contain mainly soluble NSP, in fact this is not the case. They contain signicant amounts of insoluble NSP and, moreover, most of these laxative studies were done using whole foods containing intact plant cell walls, which probably contributes to their effect. After fruit and vegetables comes psyllium at 4.0 g/g then the league table progresses through oats 3.4, other gums and mucilages 3.1, corn 2.9, legumes (mainly soya) 1.5 and lastly pectin 1.3.

Bowel habit
It is now well established that NSP reaches the large intestine and is fermented with the production of SCFA, H2, CO2 and biomass.66 This fermentative process dominates human large bowel function and

Table 3 Source

Effect of NSP (dietary bre) on bowel habit.* Ny 82 175 338 119 53 66 32 98 95 Increase in stool weight (g/g bre fed) 7.2 6.0 4.4 4.0 3.4 3.1 2.9 1.5 1.3 Median 6.5 3.7 4.9 4.3 4.8 1.9 2.9 1.5 1.0 Range 314.4 1.419.6 212.3 0.96.6 15.5 0.310.2 2.83.0 0.33.1 03.6

Raw bran Fruit and vegetables Cooked bran Psyllium/ispaghula Oats Other gums and mucilages Corn Soya and other legumes Pectin
* y

Modied and recalculated from Cummings.67 Number of individual feeding periods.

ARTICLE IN PRESS
Carbohydrates and health
Table 4 Source Potato RS2 Banana RS2 Wheat RS3 Maize RS3 Hylon VII RS2 Hylon VII RS3 Hylon VII Mixed potato RS2 and Hylon VII Mixed food sources Kelloggs cornakes Hylon VII RS2 Effect of RS on bowel habit. N 9 8 9 8 23 23 12 8 11 8 14 Amount fed g/d* 26.8 30.0 17.4 19.0 32 32 55 40 39 10 28 Increase in stool weight g/g RS fed 1.6 1.7 2.5 2.7 1.4 2.2 0.8 0.9 1.8 Zero 1.0 Reference Cummings et al.84

11

Heijnen et al.85 Hylla et al.86 Silvester et al.87 Phillips et al.88 Tomlin and Read89 Van Munster et al.90

Overall weighted mean (SEM) 1.5 g/g (0.24). N=147 diet periods in 11 studies. Other relevant papers: Grubben et al.,91 Noakes et al.,92 Shetty and Kurpad.93 * In some studies the basal diet contained RS and this amount has been subtracted from the stool weight calculation.

How reliable are these rankings? Within each group, there is a great variability, although despite this, the overall differences are statistically signicant (by ANOVAR F 4.78 Po0.001). The variability is due in part to the inherent difference in individual responses and to varying experimental designs, some of which are uncontrolled diets. A further problem is lack of consistency in methodology for measurement of dietary bre. Around 20 different methods were used in these studies, some of which give only a rough approximation of the true NSP content of the food or source. These laxative properties of NSP are used in the prevention and treatment of constipation.68 Along with the clearly dened changes in bowel habit that bre sources produce come many other changes in bowel function. These include some reduction transit time, changes in bile acid metabolism and fermentation by the commensal anaerobic ora with the production of short chain fatty acids, gas and biomass. These issues are described in detail elsewhere in these proceedings. Are any of the changes brought about by plant cell wall NSP unique? Mostly, not, although the physical properties of NSP in the gut, especially gel function in the small bowel and surface effects in the large intestine come closest to dening bre. What has put these other physiological and health effects into perspective has been the arrival on the nutritional scene in the last 20 years of resistant starch and prebiotic carbohydrates.

Resistant starch
If starch or its hydrolysis products escape digestion they pass into the large intestine where they may be fermented. This is known as resistant starch (RS).69,70 RS shows some of the attributes of NSP in that it provides substrate for fermentation with the production of SCFA and there are other consequences.71 A good biomarker of the effect of a carbohydrate on large bowel function is its effect on bowel habit, and as can be seen from Table 4, the effect of RS is modest. Seven studies have reported accurate RS measurements in diet and adequate faecal collections, from which it can be seen that the average increase in stool weight is 1.5g/g RS fed. No study has been able to distinguish amongst the different types of RS in this context. This puts RS very much towards the bottom of the league table of carbohydrates that affect bowel habit (Table 3) and very much a minor contributor compared with NSP from whole grain cereals, fruit, vegetables, psyllium, oats and corn. Nevertheless, the contribution if RS to butyrate in the large bowel is important (see Table 5). Much more important is the concept that the digestion of starch can be modied in the small bowel. One consequence of this is RS, but another is its effect on glycaemic control. All starch can ultimately be degraded by human a-amylase. However, the rate and extent to which it is digested in the small intestine determines its physiological properties. This in turn is dependent

ARTICLE IN PRESS
12 on three principal factors: (a) its physical state, namely, whole grains vs powdered or dispersed forms known as RS1, (b) the nature of the crystalline form in the starch granule and the amylose/ amylopectin ratio, for example cereal starches which are readily digestible but banana and potato are not unless cooked; legume starches lie somewhere in between RS2, (c) any food processing which has led to retrogradation, for example cooked and cooled potato RS3.60,62,70,7274 Slowing starch digestion, or modifying other factors such as lipid and protein content of the meal and thus slowing gastric emptying, reduces the glycaemic index and insulin responses. This is clearly important for diabetes and has led to major changes in dietary recommendations for diabetics. The importance of maintaining lower blood glucose proles in healthy humans has been questioned, but recent studies suggest that this is a signicant determinant of protein glycosylation.7577 Because of the physical nature of starch it has been assumed that it is hydrolysed and absorbed more slowly than sugars. This is not always the case. Some starches are rapidly digested and give rise to blood glucose responses (glycaemic index) similar to or even greater than sugars.78 The rate at which glucose in J.H. Cummings et al. a food, either as free glucose, or from sucrose or starch, becomes available for absorption in an in vitro system correlates well with the glycaemic response to that food.72 In many ways, therefore, the bre-like properties of RS are very limited, but the control of starch digestion, one consequence of which is RS, is fundamentally important for glycaemic control (Table 4).

Oligosaccharides
The discovery of the prebiotic oligosaccharide has, more than anything since the bre story began, focused attention on the importance of dietary carbohydrate and its constituent parts.7981 Whilst the prebiotic carbohydrates have been considered by some to be bre, they bear not even a remote resemblance to the classical description of bre. Prebiotic carbohydrates must, by denition, selectively stimulate the growth of either bidobacteria or lactobacilli in the human colon. They are mostly sugar-like compounds comprising between two and ten carbohydrate units that are soluble in 80% ethanol and largely resist digestion by pancreatic and brush board enzymes. They are
Table 7 Physiological effects of prebiotic carbohydrates.

Table 5

Physiological properties of RS.

 Contributes to control of glycaemic response  Substrate for fermentation

SCFA production, especially butyrate Effect on bowel habit Microbial metabolism Gas Bile acid effects

 Selective stimulation of bidobacteria and lactobacilli

 Substrate for fermentation

SCFA Gas  Enhanced calcium absorption

Table 6 Source

Effect of prebiotic oligosaccharides on bowel habit. N 13 14 8 8 4 24 24 12 12 12 Amount fed g/d 7.5 15.0 10.0 15.0 15.0 5.0 15.0 2.5 5.0 10.0 g/g Increase in stool weight 0 0 0 1.0 2.1 0 0 0 0 1.1 Reference Alles et al.94 Bouhnik et al.95 Gibson et al.96 Alles et al.97 Ito et al.98

Transgalacto. Transgalacto. Fructo. Inulin Fructo. Galacto.

Mean 0.2 g/g in 131 diet periods (N*). See also: Swanson et al.,99 Chen et al.,100 Bouhnik et al.,101 Kleesen et al.102

ARTICLE IN PRESS
Carbohydrates and health
Table 8 Contrasting properties of some dietary carbohydrates. NSP (bre) Physical effects Bowel habit Cholesterol lowering Fermentation Prebiosis Calcium absorption Glycaemic response Appetite + + + + + 7 Resistant starch 7 + + ? Prebiotic oligosaccharides + + + ?

13

substrates for fermentation but their effects on bowel habit are negligible as Table 6 shows. The key to their importance to human health relies on their remarkable effects on the large intestinal ora (see Table 7). The health consequence of prebiotic carbohydrates are by no means fully documented but if we can learn lessons from the role of probiotic bacteria, the growth of which is stimulated in the large bowel by prebiotic carbohydrates, then we might expect these highly water soluble substances to enhance resistance to pathogen invasion, be a source of SCFAs and in conjunction with probiotic species reduce the risk of neoplastic change in the gut epithelium. There is also evidence emerging that they may have effects in the small intestine, particularly in enhancing calcium absorption.

that these also be called bre, the distinct properties and health benets of these carbohydrates (Table 8) make it important to consider each fraction separately, physiologically, epidemiologically and in clinical studies.

Appendix A. Consensus discussion

Barbara Schneeman: This concept of physical effects in the gut is clearly an important aspect of dietary bre. With the data we have gathered, we can show changes, but where should we start? What should we dene as normal in normal healthy people, for instance in viscosity, and what should we recommend in the diet to ensure that people fall within that? John Cummings: There are enormous methodological problems in understanding what the physical state of these materials is in the upper intestine because it depends on the rest of the diet, pH and so on. For the most part, we are taking these materials in as part of the diet and not as isolated materials, which makes it difcult. So we dont know where to start and where to nish. But I would say it is not just about viscosity. I think our understanding of the role of surfaces in the gut has been totally neglected, yet surfaces can catalyse reactions in the gut. This is an important dimension that will be easier to sort out. Barbara Schneeman: But we have to start somewhere. Perhaps we should dene a certain bre intake with certain mixed sources of bre and characterise these properties within the gut, and say normal healthy people have this, and then start from there. John Cummings: It would be nice to have a physical properties index for foods, which we know would be contributed to largely by the non-starch polysaccharide (NSP), for instance. The physical chemists would have to do this, but it would have

Conclusion
Biological science has progressed in many ways in recent years, and this is reected in current bre research. Fibre, principally the non-starch polysaccharide of the plant cell wall, through its fermentation, is known to affect the growth and differentiation of epithelial cells in the gut, hepatic metabolism of carbohydrate and lipid, and genotoxic events within the large bowel. The physical properties of bre have proved difcult to study, but are essential to its role in moderating glucose and insulin responses in the blood. The breakdown of bre by the colonic bacteria has given new insights into their importance for health. Fibre has thus facilitated a broadening of our understanding of gut physiology and metabolism and has identied the colon as a major digestive organ. Equally important, bre research has led to a re-appraisal of the digestive physiology of other carbohydrates, particularly starches and oligosaccharides. Whilst there have been some suggestions

ARTICLE IN PRESS
14 to be based on more than just viscosity. Nutrition has to move along these lines, rather than macro and micro nutrientswe have to look at the global food in the diet. Remy Meier: Do you think some bres will be more important than others? John Cummings: Not at all. The non-starch polysaccharides bre cell wall story to me feeds through to a pattern of diet, which has fruit, vegetables and whole grain cereals in it. I would not want to pick out one type of bre that looked exciting. However, the oligosaccharides are a totally different group of carbohydrates, in which potentially there will be some individual molecules with hugely important properties. Wolfgang Scheppach: Can you explain the different denitions of prebiotics? Glenn Gibson: The original denition was of a material that would induce a benecial shift in colonic ora. Currently, we consider they have an effect elsewhere in the human or animal body apart from the large intestine. Perhaps the next generation may induce a shift downwards, and certainly the types of products of fermentation also need to be taken into account. But we would look for a perceived positive health shift in the gut ora composition. Stig Bengmark: You did not mention pectin, yet this is a very strong anti-oxidant, for example, a good transport vehicle for probiotic bacteria to the gut and it is a good substrate for fermentation. Also, there was an in vitro study that looked at some 700 bacteria and their ability to ferment inulin. Only a few could do that, the strongest being a Lactobacillus species that only 25% of westerners are colonised with. Could it be that we provide inulin and even fructo-oligosaccharides (FOS), yet there are no bacteria to ferment them? John Cummings: The anti-oxidant properties of pectin should certainly be discussed. If it is real then it is very important. Glenn Gibson: Much of the evidence on bacterial species comes from work where the test system is a non-competitive one. The real test is in the mixed environment of the large intestine, where there are several hundred different species competing for substrates. In that context, the bidobacteria seem to do very well in fermenting inulin. Paul van Leeuwen: The culturing and numbers of bacteria in the stool is always confusing as there are two orathe luminal ora, which is easily washed out by lavage, and the gut wall-associated ora, which consists of anaerobes and aerobes. It is hard to determine whether there is an increase in a certain type of bacteria. What is the clinical signicance if there is a shift from one to another J.H. Cummings et al. and if there is an increase in lactobacillus, is this good or bad? John Cummings: It is possible with modern bacteriological technology to identify what is in the gut. Whether it is healthy or not is the question everyone is asking. Certainly in terms of the prebiotic carbohydrate substrates, there are no good clinical studies that show benet. But if you look at the role of probiotics, the clinical benets are highly signicant, for example in treating childhood infectious diarrhoeas, and preventing antibiotic associated diarrhoea, and preventing childhood allergies. One would hope that a different way of enhancing these bacteria in the gut would prove equally benecial but it is probably too early to say. Henrik Andersson: Regarding increased absorption of calcium from oligosaccharides, do you think it is possible that this could come from the colon? John Cummings: I would be happy to accept that this is a colonic event. Your ileostomy studies support this, and it is possible to conceive of a mechanism for it. It would certainly make the whole story more interesting. There is also the lactase story with calcium, which may be a colonic event, but is presumably only operative in those who are lactase decient.

References
1. Higginson J, Oettle AG. Cancer incidence in the Bantu and Cape Colored races of South Africa: report of a cancer survey in the Transvaal (195355). J Natl Cancer Inst 1960;24:589671. 2. Oettle AG. Cancer in Africa, especially in regions south of the Sahara. J Natl Cancer Inst 1964;33:383439. 3. Burkitt DP. Epidemiology of cancer of the colon and rectum. Cancer 1971;28:313. 4. Trowell HC. Non-infective diseases in Africa. London: Edward Arnold; 1960. 5. Cleave TL. The Saccharine Disease. Bristol: John Wright; 1974. 6. Painter NS. Diverticular disease of the colon: a disease of this century. Lancet 1969;2:5868. 7. Painter NS, Burkitt DP. Diverticular disease of the colon: a deciency disease of Western civilisation. Br Med J 1971;2:4504. 8. Burkitt DP, Walker ARP, Painter NS. Effect of dietary bre on stools and transit times, and its role in the causation of disease. Lancet 1972;ii:140812. 9. Cummings JH. Progress report: dietary bre. Gut 1973;14:6981. 10. Groot APd, Luyken R, Pikaar NA. Cholesterol lowering effect of rolled oats (letter). Lancet 1963;2:3034. 11. Keys A, Grande F, Anderson JT. Fibre and pectin in the diet and serum cholesterol concentration in man. Proc Soc Exp Biol (NY) 1961;106:5558. 12. Southgate DAT. Determination of carbohydrates in foods. II. Unavailable carbohydrates. J Sci Food Agri 1969;20:3315.

ARTICLE IN PRESS
Carbohydrates and health
13. Thomas M. Die Nahr und Ballastoffe der Getreidemehle in ihrer Bedeutung fur die Britnahrung. Stuttgart: Wissenschaftliche Verlagsgesellschaft; 1964. 14. Hellendoorn EW, Noordhoff MG, Slagman J. Enzymatic determination of the indigestible residue (dietary bre) content of human food. J Sci Food Agri 1975;26:14618. 15. Van Soest PJ, McQueen RW. The chemistry and estimation of bre. Proc Nutrl Soc USA 1973;32:12330. 16. Burkitt DP, Trowell HS. Rened Carbohydrate Foods and Disease: Some Implications of Dietary Fibre. London: Academic Press; 1975. 17. Trowell H. Dietary bre: a paradigm. In: Trowell H, Burkitt D, Heaton KW editors. Dietary Fibre, Fibre-Depleted Foods and Diseases. London: Academic Press; 1985. p. 120. 18. McCance RA, Widdowson EM. The chemical composition of foods. Med Res Council 1940. 19. McCance RA, Lawrence RD. The carbohydrate content of foods. Special Report series of the Medical Research Council No 135. London: HMSO; 1929. 20. Trowell H. Dietary bre and coronary heart disease. Eur J Clin Biol Res 1972;17:3459. 21. Hipsley EH. Dietary bre and pregnancy toxaemia. Br Med J 1953;ii:4202. 22. Trowell H. Denitions of bre (Letter). Lancet 1974;i:503. 23. Trowell H, et al. Dietary bre redened (Letter). Lancet 1976;i:967. 24. Donnelly B. National Academies of Science denitions relating to food bre only add confusion. Br J Nutr 2003;90:4813. 25. De Vries JW. On dening dietary bre. Proc Nutr Soc 2003;62:3743. 26. Mongeau R, Brooks SPJ. Chemistry and analysis of lignin. In: Choo SS, Dreher ML editors. Handbook of dietary bre. New York: Marcel Dekker, Inc; 2001. p. 32173. 27. Cummings JH. Some aspects of dietary bre metabolism in the human gut. In: Birch GA editor. Food and HealthScience and Technology. London: Applied Science Publishers Ltd; 1980. p. 44158. 28. Englyst HN, et al. editors. Dietary bre and resistant starch (Editorial). Am J Clin Nutr 1987;46:8734. 29. Cummings JH, et al. Fecal weight, colon cancer risk and dietary intake of non-starch polysaccharides (dietary bre). Gastroenterology 1992;103:17839. 30. Department of Health, ed. Dietary reference values for food energy and nutrients for the United Kingdom. Report on Health and Social Subjects, vol. 41. London: HMSO; 1991. 31. WHO. Diet, Nutrition and the Prevention of Chronic Diseases. Geneva: WHO; 1990. 32. FAO/WHO, Carbohydrates in human nutrition. FAO, 1997. p. 1140. 33. Cummings JH. Dietary bre intakes in Europe: overview and summary of European research activities, conducted by members of the Management Committee of COST 92. In: Carnovale E, Tomassi G, Cummings JH, editors. Topics in Dietary Fibre Research, European Journal of Clinical Nutrition 1995. p. S5S9. 34. Burley VJ, Blundell JE. Action of dietary bre on the satiety cascade. In: Kritchevsky D, Boneld C, Anderson JW editors. Dietary bre: chemistry, physiology and health effects. New York: Plenum; 1990. p. 22746. 35. de Jong A. Short chain fatty acids, pancreatic hormones and appetite control. In: Cummings JH, Rombeau JL, Sakata T editors. Physiological and Clinical Aspects of Short Chain Fatty Acids. Cambridge: Cambridge University Press; 1995. p. 25774.

15
36. Anderson JW, Deakins DA, Bridges SR. Soluble bre hypocholesterolemic effects and proposed mechanisms. In: Kritchevsky D, Boneld C, Anderson JW editors. Dietary brechemistry, physiology and health effects. New York: Plenum Press; 1990. p. 33963. 37. Jenkins DJA, et al. Fibre in the treatment of hyperlipidemia. In: Spiller GA editor. CRC Handbook of Dietary Fibre in Human Nutrition. Florida: CRC Press; 1993. p. 41938. 38. Thacker PA, et al. Inuence of propionic acid on the cholesterol metabolism of pigs fed hypercholesterolemic diets. Can J Animal Sci 1981;61:96975. 39. Chen W-JL, Anderson JW, Jennings D. Propionate may mediate the hypocholesterolemic effects of certain soluble plant bres in cholesterol-fed rats. Proc Soc Exp Biol Med 1984;175:215. 40. Venter CS, Vorster HH, Cummings JH. Effects of dietary propionate on carbohydrate and lipid metabolism in man. Am J Gastroenterol 1989;85:54953. 41. Todesco T, et al. Propionate lowers blood glucose and alters lipid metabolism in healthy subjects. Am J Clin Nutr 1991;54:8605. 42. Illman RJ, et al. Hypocholesterolaemic effects of dietary propionate studies in whole animals and perfused rat liver. Ann Nutr Metabol 1988;32:97107. 43. Walker ARP, Arvidsson UB. J Clin Invest 1954;33:1358. 44. Bersohn I, Walker ARP, Higginson J. S Afr Med J 1956;30:411. 45. Morris JN, Marr JW, Clayton DG. Diet and heart: a postscript. Br Med J 1977;2:130714. 46. Yano K, et al. Dietary intake and the risk of coronary heart disease in Japanese men living in Hawaii. Am J Clin Nutr 1978;31:12709. 47. Kromhout D, Bosschieter EB, Coulander CL. Dietary bre and 10-year mortality from coronary heart disease, cancer and all causes. The Zutphen Study. Lancet 1982;ii:50821. 48. Kushi LH, et al. Diet and 20-year mortality from coronary heart disease. The Ireland-Boston Diet-Heart study. New Engl J Med 1985;312:8118. 49. Garcia-Palmieri MR, et al. Relationships of dietary intake to subsequent coronary heart disease incidence: the Puerto Rico Heart Health Program. Am J Clin Nutr 1980;33:181827. 50. Anderson JW. Whole grains protect against atherosclerotic cardiovascular disease. Proc Nutr Soc 2003;62:13542. 51. Trowell H. Ischemic heart disease and dietary bre. Am J Clin Nutr 1972;25:92632. 52. Trowell H. Dietary bre, ischaemic heart disease and diabetes mellitus. Proc Nutr Soc 1973;32:1517. 53. Jenkins DJA, et al. Unabsorbable carbohydrates and diabetes: decreased post-prandial hyperglycaemia. Lancet 1976;ii:1724. 54. Association BD. Dietary recommendations for diabetics for the 1980s. Human Nutr: Appl Nutr 1982;36A: 37894. 55. Diabetes and Nutrition Study Group of the European Association for the Study of Diabetes, Nutritional recommendations for individuals with diabetes mellitus. Diabetes Nutr Metabol 1988;1:1459. 56. Berger M, Venhaus A. Dietary bre in the prevention and treatment of diabetes mellitus. In: Schweizer TF, Edwards CA editors. Dietary FibreA Component of Food. London: Springer; 1992. p. 27993. 57. Wolever TMS, Jenkins DJA. Effect of dietary bre and foods on carbohydrate metabolism. In: Spiller GA editor. CRC handbook of dietary bre in human nutrition. Florida: CRC Press; 1993. p. 11152.

ARTICLE IN PRESS
16
58. Jenkins DJA, et al. Dietary bres, bre analogues and glucose tolerance: importance of viscosity. Br Med J 1978;1:13924. 59. Kiehm TG, Anderson JW, Ward K. Benecial effects of a high carbohydrate, high bre diet on hyperglycemic diabetic men. Am J Clin Nutr 1976;29:8959. 60. Crapo PA, Reaven G, Olefsky J. Plasma glucose and insulin responses to orally administered simple and complex carbohydrates. Diabetes 1976;25:7417. 61. Crapo PA, Reaven G, Olefsky J. Postprandial plasma-glucose and insulin responses to different complex crbohydrates. Diabetes 1977;26:117883. 62. Jenkins DJA, et al. Glycemic index of foods: a physiological basis for carbohydrate exchange. Am J Clin Nutr 1981;34:3626. 63. Asp NG. Resistant Starch. Physiological implications of the consumption of resistant starch in man. Proceedings from the Second plenary meeting of EURESTA: European FLAIR Concerted Action No. 11 (COST 911). Eur J Clin Nutr 1992;46(Suppl. 2):1148. 64. Stephen AM. Starch in human nutrition. Can J Physiol Pharmacol 1991;69(Symposium Proceedings): 53136. 65. Haber GB, et al. Depletion and disruption of dietary bre: effects on satiety, plasma-glucose and serum insulin. Lancet 1977;ii:67982. 66. Cummings JH. Fermentation in the human large intestine: evidence and implications for health. Lancet 1983;1:12069. 67. Cummings JH. The effect of dietary bre on fecal weight and composition. In: Spiller GA editor. CRC handbook of dietary bre in human nutrition. Tampa, Florida: CRC Press LLC; 2001. p. 183252. 68. Cummings JH. Non-starch polysaccharides (dietary bre) including bulk laxatives in constipation. In: Kamm MA, Lennard-Jones JE editors. Constipation. Peterseld, UK: Wrightson Biomedical Publishing Ltd.; 1994. p. 30714. 69. Englyst H, Wiggins HS, Cummings JH. Determination of the non-starch polysaccharides in plant foods by gas liquid chromatography of constituent sugars as alditol acetates. Analyst 1982;107:30718. 70. Englyst HN, Kingman SM, Cummings JH. Classication and measurement of nutritionally important starch fractions. Eur J Clin Nutr 1992;46:S3350. 71. Bird AR, Toppings DL. Resistant starches, fermentation, and large bowel health. In: Cho SS, Dreher ML editors. Handbook of dietary bre. New York: Marcell Dekker, Inc.; 2001. p. 14758. 72. Englyst EN, Hudson GJ. The classication and measurement of dietary carbohydrates. Food Chem 1996;57(1): 1521. 73. Hermansen K, et al. Differential glycaemic effects of potato, rice and spaghetti in type 1 (insulin dependent) diabetic patients at constant insulinaemia. Diabetalogia 1986;29:35861. 74. Hiele M, et al. 13CO2 breath test to measure the hydrolysis of various starch formulations in healthy subjects. Gut 1990;31:1758. 75. Brownlee M, Cerami A, Vlassara H. Advanced glycosylation endproducts in tissue and the biochemical basis of diabetic complications. New Engl J Med 1988;318:131521. 76. Dills WL. Protein fructosylation: fructose and the Maillard reaction. Am J Clin Nutr 1993;58(5(S)):779S87S. 77. McDonald RB. Inuence of dietary sucrose on biological ageing. Am J Clin Nutr 1995;62(1(S)):284S93S. 78. Wolever TMS, Miller JB. Sugars and blood glucose control. Am J Clin Nutr 1995;62(1(S)):212S27S.

J.H. Cummings et al.

79. Cummings JH, Macfarlane GT. Gastrointestinal effects of prebiotics. Br J Nutr 2002;87:S145151. 80. Roberfroid M, Gibson G. Nutritional health benets of inulin and oligofructose. Br J Nutr 2002;87(S139-S311). 81. Schrezenmeir J, de Vrese M, Heller K. Probiotics and Prebiotics. Am J Clin Nutr 2001;73(2(S)):361S498S. 82. Levine AS, et al. Effect of breakfast cereals on short-term food intake. Am J Clin Nutr 1989;50:13037. 83. Delargy HJ, et al. Effects of amount and type of dietary bre (soluble and insoluble) on short-term control of appetite. Int J Food Sci Nutr 1997;48:6777. 84. Cummings JH, et al. Digestion and physiological properties of resistant starch in the human large bowel. Br J Nutr 1996;75:73347. 85. Heijnen MLA, et al. Limited effect of consumption of uncooked (RS2) or retrograded (RS3) resistant starch on putative risk factors for colon cancer in healthy men. Am J Clin Nutr 1998;67:32231. 86. Hylla S, et al. Effects of resistant starch on the colon in healthy volunteers: Possible implications for cancer prevention. Am J Clin Nutr 1998;67:13642. 87. Silvester KR, et al. Effect of meat and resistant starch on fecal excretion of apparent N-nitroso compounds and ammonia from the human large bowel. Nutr CancerInt J 1997;29:1323. 88. Phillips J, et al. Effect of resistant starch on fecal bulk and fermentation-dependent events in humans. Am J Clin Nutr 1995;62:12130. 89. Tomlin J, Read NW. The effect of resistant starch on colon function in humans. Br J Nutr 1990;64:58995. 90. Vanmunster IP, Tangerman A, Nagengast FM. Effect of resistant starch on colonic fermentation, acid bile metabolism and mucosal proliferation. Dig Dis Sci 1994;39:83442. 91. Grubben MJAL, et al. Effect of resistant starch on potential biomarkers for colonic cancer risk in patients with colonic adenomasA controlled trial. Dig Dis Sci 2001;46:7506. 92. Noakes M, et al. Effect of high-amylose starch and oat bran on metabolic variables and bowel functions in subjects with hypertriglyceridemia. Am J Clin Nutr 1996;64:94451. 93. Shetty PS, Kurpad AV. Increasing starch intake in the human diet increases fecal bulking. Am J Clin Nutr 1986;43:2102. 94. Alles MS, et al. Effect of transgalactooligosaccharides on the composition of the human intestinal microora and on putative risk markers for colon cancer. Am J Clin Nutr 1999;69:98091. 95. Bouhnik Y, et al. Administration of transgalacto-oligosaccharides increases fecal bidobacteria and modies colonic fermentation metabolism in healthy humans. J Nutr 1997;127:4448. 96. Gibson GR, et al. Selective stimulation of Bidobacteria in the human colon by oligofructose and inulin. Gastroenterology 1995;108:97582. 97. Alles MS, et al. Fate of fructo-oligosaccarides in the human intestine. Br J Nutr 1996;76:21121. 98. Ito M, et al. Effects of administration of galactooligosaccharides on the human faecal microora, stool weight and abdominal sensation. Microb Ecol Health Dis 1990;3:28592. 99. Swanson KS, et al. Fructooligosaccharides and Lactobacillus acidophilus modify bowel function and protein catabolites excreted by healthy humans. J Nutr 2002;132:304250. 100. Chen HL, et al. Effects of isomalto-oligosaccharides on bowel functions and indicators of nutritional status in constipated elderly men. J Am Coll Nutr 2001;20:449.

ARTICLE IN PRESS
Carbohydrates and health
101. Bouhnik Y, et al. Short-chain fructo-oligosaccharide administration dose-dependently increases fecal bifodobacteria in healthy humans. J Nutr 1999;129:1136.

17
102. Kleessen B, et al. Effects of inulin and lactose on fecal microora, microbial activity, and bowel habit in elderley constipated persons. Am J Clin Nutr 1997;65:1397492.