Nutrition Discussion Forum

A recent issue of the British Journal of Nutrition reflects a marked change in the importance we now attach to the role of dietary carbohydrates in both managing and retarding the development of non-communicable diseases, with four commentaries: Choosing your carbohydrates to prevent diabetes (Mathers, 2002); Low carbohydrate does not mean low glycaemic index! (Wolever, 2002); Diet, satiety and obesity treatment (Tremblay et al. 2002); Handbook of dietary fibre (Cummings, 2002). Among these there are important messages on public health and expressions of hope for the future. However, the commentaries on starchy staples and glycaemic index (GI) have implications beyond diabetes and may be more easily understood in a broader context that aims for low postprandial glycaemia, eventually by more than one approach based on diet together with weight restriction and/or reduction. Certainly one can welcome a possible role for low-GIstarch foods in the prevention of type 2 diabetes mellitus (DM), as suggested by Mathers (2002). The weight of published evidence to date, nevertheless, is in diabetic subjects (not pre-diabetic or glucose-intolerant subjects), is based largely on intervention trials with diets characterised analytically as low-GI-carbohydrate (not necessarily lowerGI-starch as sometimes intended) and measurement of biochemical risk factors (not clinical outcomes, either microvascular or macrovascular). Advice to consume low-GI-starch foods can also elevate intake of sugars (G Livesey, unpublished results), many of which have GI values less than common starch-rich staple foods. Interestingly, there is anecdotal evidence in favour of using low-GI-starch foods in diabetic subjects in a non-western setting, where the likelihood of a spontaneous intake of more sugars would seem less: physicians of ancient India treated DM with barley and Bengal gram (Kapur & Kapur, 2001), which we now know to be amongst the lowest-GI-starch foods available. Efficacy of such treatment on clinical outcome was unreported and now possibly unknown without further study. There is a scarcity of published dietary intervention studies with low-GI foods that examine prevention of clinical disease in subjects at risk, such as those who are glucose intolerant. However, there appears to be improved pancreatic b-cell function with low-GI-carbohydrate diets among some subjects who are glucose intolerant (Wolever & Mehling, 2002). Moreover, in the STOP-NIDDM randomised trial the glycaemic response to diet was reduced by slowing digestion with an a-glucosidase inhibitor (acarbose), and there were both significantly fewer conversions of glucose-intolerant to type 2 DM patients and more reversions to normal (Chiasson et al. 2002). Mathers’ (2002) comments are presumably a call to further arm us with similar knowledge as may be got from more research on diets. Hence, however good the outlook, we do not know yet the level of disease reduction that will come from intervention with low-GI diets (let alone low-GIstarch foods). Nevertheless, the future burden of disease from diabetes (obesity, stroke and heart disease too, which may be greater problems) is expected to overburden government (and private) health budgets. Waiting for conclusive proof on the magnitude of efficacy of low-GI-carbohydrate foods on clinical end points may therefore be unwise, given the suggested absence of risk from reduced postprandial glycaemia (American Diabetes Association, 2001), a positive outlook (c.f. Canadian Diabetes Association, 2000; Buykens et al. 2000, 2001) and prospective evidence from epidemiological studies that low-GI diets (again not uniquely low-GI-starch diets, though inclusive of wholegrain cereals) appear to lower the advent of type 2 DM, CHD and possibly stroke (all disease outcomes rather than biochemical markers of disease) (Salmerón et al. 1997a,b; Liu et al. 2000a,b, 2002). Further, we have to consider whether reduced glycaemia might benefit a majority of the population (Khaw et al. 2001) and not just those believed to be at greatest risk. The last paper (Khaw et al. 2001) indicates glycosylated haemoglobin (HbA1c) is an important prospective marker of macrovascular disease. This, in addition to microvascular diseases seen in diabetics, and the need to control glycaemia more broadly than for diabetes prevention, supports earlier epidemiological work on glycaemic index (or glycaemic load) and CHD. Interpretation of the epidemiological evidence is difficult. It is unclear whether the advent of type 2 DM and CHD is linked to the GI of the carbohydrate ingested or to glycaemic load (i.e. GI £ carbohydrate intake). The latter seems to hold the stronger association with both advent of diseases (Salmerón et al. 1997a,b; Liu et al. 2000a) and C-reactive protein, a marker for CHD risk (Liu et al. 2002). Yet, this might not have been expected: as with low-carbohydrate intake, low glycaemic load could imply higher fat intake, causing damage that obscures the true strength of benefit from low glycaemia. Such obscuring might explain why sometimes (e.g. Meyer et al. 2000) the association between disease and glycaemia does not always manifest itself strongly. Wolever (2002) rightly points out that low GI (as applied to carbohydrate) does not mean the same as low carbohydrate. GI describes carbohydrate quality not quantity, and so low GI cannot imply low-fat intake. Indeed, low GI can imply any level of fat intake, which makes it important to specify that healthy diets are those that are of high intake of low-GI carbohydrates. However, overfocusing on GI limits our scope of vision. We may: (1) British Journal of Nutrition (2002), 88, 741–744 DOI: 10.1079/BJN2002730 q The Author 2002