EDITORIAL (SEE LIESE ET AL., P. 2832)

Do Glycemic Index, Glycemic Load, and Fiber Play a Role in Sensitivity, Disposition Index, and ?

n this issue of Diabetes Care, Liese et al. fiber to have a positive association with and the other of the Health Professionals’ (1) report on the relation of dietary gly- insulin sensitivity and an inverse associa- Study (7). However, reading these reports I cemic index, glycemic load, carbohy- tion with adiposity and disposition index. carefully, neither is significant for a glyce- drate, and fiber intake to measures of Insulin sensitivity is a very important mic load effect and only one for glycemic insulin sensitivity, insulin secretion, and component of carbohydrate homeostasis. index effect at P Ͻ 0.04. Other epidemi- adiposity in the Ath- Individuals with insulin resistance are ological, observational, longitudinal stud- erosclerosis Study (IRAS). The evaluation more likely to eventually develop type 2 ies have shown no significant effect. These was confined to those with normal or im- diabetes. The potential effect of diet on include the Iowa Women’s study (8), the paired glucose tolerance and did not in- this physiological state is thus important San Luis Valley Study (9), and the Athero- clude diabetic individuals. As the authors in gauging risk. In short-term interven- sclerosis Risk in Communities study (10). state, there has been no previous large ep- tional metabolic studies in small numbers The present IRAS report bolsters the neg- idemiologic study relating glycemic index of people, the best trial to date, e.g., the ative data. A final study, the Nurses Study and glycemic load to direct measures of longest and most comprehensive, has II, showed a significant effect of glycemic insulin sensitivity and insulin secretion, shown an improvement in insulin sensi- index, but both glycemic load and total whose dysfunction are the two hallmarks tivity with a high– as opposed to a low– carbohydrates were inversely associated leading to type 2 diabetes. glycemic index diet (3). Other less with diabetes risk (11). These studies There are some caveats to this study rigorous studies have shown mixed tried to relate glycemic index and glyce- that must be pointed out. First, it is a results. mic load to risk of diabetes, they did not cross-sectional study looking at one point Let me deal with fiber first. The ben- measure insulin sensitivity. Insulin sensi- in time. Longitudinal studies are certainly efit of fiber in the diet on a number of tivity generally has been measured in met- preferable. Second, it is an observational chronic diseases has been documented abolic ward studies with interventional study and interventional studies are more repeatedly in epidemiological studies. An trials of short duration, such as the Kiens valuable, though much more difficult and increased consumption has been associ- and Richter (3) study mentioned earlier. expensive to carry out. Third, the food ated with lower incidence of diabetes, The Liese et al. study is the first to try frequency questionnaire used in this coronary artery disease, and obesity in ob- to document the impact of a higher– study was not specifically designed to test servational studies. With regard to diabe- versus a lower–glycemic index and/or for glycemic index and glycemic load. tes, most studies have singled out cereal glycemic load diet on insulin sensitivity in While it has been validated as an overall fiber as the important component, with instrument, it has not been validated for other types of fiber giving much lower or a large epidemiological study. They were reproducibility and reliability as an ap- no association. However, as stated in a unable to document a relationship be- propriate glycemic index instrument, and report of the National Academy of Sci- tween either glycemic index or glycemic this needs to be done by the IRAS group. ences “there is no conclusive evidence load and insulin sensitivity. A larger ob- This field has been dogged by the inaccu- that it is dietary fiber rather than the other servational study in Denmark (12) also racy of dietary records and the difficulty components of vegetables, fruits, and ce- could not document an association of gly- in calculating dietary glycemic index and real products that reduces the risk of cemic index with insulin resistance using glycemic load levels of individuals from those diseases” (4). The present study a homeostasis model assessment of insu- their reported intake of foods. Validation breaks no new ground here, it just bol- lin resistance and found an inverse asso- of experimental instruments is crucial. sters previous data impressively, suggest- ciation between glycemic load and insulin Fourth, the minimal model was instituted ing that the effect on lowering risk of type resistance. using 30 plasma samples to calculate in- 2 diabetes may work through enhancing The IRAS investigators were also un- sulin sensitivity (2). This study used 12. insulin sensitivity. It supports the recom- able to find an association of glycemic in- While this seems reasonable in a large ep- mendation in the 2005 dietary guidelines dex and glycemic load with disposition idemiological study, the reliability is for Americans (5) to increase their fiber index. The disposition index measures likely to not be as high as using the orig- intake. The present U.S. fiber intake is the ability of the to respond to an inally described method. very low and an increase undoubtedly increase in insulin resistance with an in- Having said the above, the study re- would improve health. creased secretion of insulin, thereby ports very interesting results. IRAS The question of glycemic index and maintaining normal blood glucose. An showed a lack of association of glycemic glycemic load is more contentious. As the abnormal disposition index suggests index, glycemic load, and carbohydrate authors state, high–glycemic index diets ␤-cell strain and can lead to eventual intake with measures of insulin sensitiv- have been linked to an elevated risk of ␤-cell failure. Thus, the inability to ob- ity, insulin secretion, and adiposity, ad- developing diabetes. There are two re- serve an inappropriately low pancreatic justing for energy intake. It also showed ports, one of the Nurses’ Health Study (6) response associated with higher glycemic

2978 DIABETES CARE, VOLUME 28, NUMBER 12, DECEMBER 2005 Editorial index and glycemic load suggests normal with the development of type 2 diabetes. 5. USDA dietary guidelines [article online], pancreatic functioning on such diets. This then, greatly diminishes the impor- 2005. Available at http://www.healthierus. In the last few years, there has been a tance of high glycemic load as an impor- gov/dietaryguidelines. Accessed 3 Sep- very strong push by some investigators to tant risk. tember 2005 declare a high–glycemic index and a My suggestion then, looking at the 6. Salmero´n J, Manson JE, Stampfer MJ, high–glycemic load diet detrimental to present study and others, is that until fur- Colditz GA, Wing AL, Willett WC: Di- health, particularly in relation to the de- ther evidence is available, we should con- etary fiber, glycemic load, and risk of non- insulin-dependent diabetes mellitus in velopment of obesity and type 2 diabetes. centrate on educating the public to opt for women. JAMA 277:472–477, 1997 They have pressured public health au- higher-fiber foods (especially cereal fiber) 7. Salmero´n J, Ascherio A, Rimm EB, Colditz thorities to recommend that such diets and downplay the glycemic index and GA, Spiegelman D, Jenkins DJ, Stampfer should be restricted for the population at glycemic load. There is excellent evidence MJ, Wing AL, Willett WC: Dietary fiber, large. But it must be remembered that the that the higher-fiber foods, made up of glucemic load, and risk of NIDDM in concept of the glycemic index was first whole grains, fruits, and vegetables, will men. Diabetes Care 20:545–550, 1997 proposed as a tool to try to improve glu- do people good. 8. Meyer KA, Kushi LH, Jacobs DR, Slavin J, cose control in diabetic patients, where it Sellers TA, Folsom AR: Carbohydrates, may be of some help to patients with poor XAVIER PI-SUNYER, MD dietary fiber, and incident type 2 diabetes postprandial glucose control. But there is in older women. Am J Clin Nutr 71:921– a great deal of confusion in interpreting From the Division of Endocrinology, Diabetes and 930, 2000 Nutrition, St. Luke’s-Roosevelt Hospital Center, Co- 9. Marshall J, Bessessen D, Hamman R: High the database available on glycemic index lumbia University, New York, New York. because data from studies in diabetic pa- Address correspondence to Dr. Xavier Pi-Sunyer, saturated fat and low starch and fiber are tients are often quoted to bolster policy Division of Endocrinology, Diabetes, and Nutrition, associated with hyperinsulinemia in a suggestions for normal populations. This St. Luke’s-Roosevelt Hospital Center, 1111 Amster- non-diabetic population: the San Luis dam Ave., New York, NY. E-mail: fxp1@columbia. confusion is not conducive to a clear anal- Valley Diabetes Study. Diabetologia 40: edu. 430–438, 1997 ysis of the issue. A pro and con discussion © 2005 by the American Diabetes Association. 10. Stevens J, Ahn K, Juhaeri, Houston D, Ste- of the use of the glycemic index in normal ●●●●●●●●●●●●●●●●●●●●●●● ffan L, Couper D: Dietary fiber intake and population groups that is relevant in this glycemic index and incidence of diabetes context was published recently (13,14). References in African-American and white adults: the It is important to note that the win- 1. Liese AD, Schulz M, Fang F, Wolever ARIC study. Diabetes Care 25:1715–1721, dow of glycemic index variability in a TMS, D’Agostino RB Jr, Sparks KC, 2002 free-living population is quite narrow. Mayer-Davis EJ: Dietary glycemic index 11. Schulze MB, Liu S, Rimm EB, Manson JE, This can be seen by the small SD of 4.0 and glycemic load, carbohydrate and fiber Willett WS, Hu FG: Glycemic index, gly- intake, and measures of insulin sensitiv- around a mean of 58 in the Liese et al. cemic load, and dietary fiber intake and ity, secretion, and adiposity in the Insulin incidence of type 2 diabetes in younger study. It can also be seen that the glycemic Resistance Atherosclerosis Study. Diabetes and middle-aged women. Am J Clin Nutr index is actually already quite low. With Care 28:2832–2838, 2005 such a low level of glycemic index and 2. Steil GM, Volund A, Kahn SE, Bergman 80:348–356, 2004 such a small variation around it, what de- RN: Reduced sample number for calcula- 12. Lau C, Faerch K, Glumer C, Tetens I, Ped- termines the level of the glycemic load is tion of insulin sensitivity and glucose ef- ersen O, Carstensen B, Jorgensen T, the total amount of carbohydrate that an fectiveness from the minimal model: Borch-Johnsen K: Dietary glycemic index, individual eats. Thus, with a small glyce- suitability for use in population studies. glycemic load, fiber, simple sugars, and insulin resistance: the Inter99 Study. Di- mic index window, the glycemic load pri- Diabetes 42:250–256, 1993 3. Kiens B, Richter EA: Types of carbohy- abetes Care 28:1397–1403, 2005 marily reflects the carbohydrate intake. 13. Pi-Sunyer FX: Glycemic index and disease And we know from repeated studies (in drate in an ordinary diet affect insulin ac- tion and muscle substrates in humans. (Review). Am J Clin Nutr 76 (Suppl.): fact, all the epidemiological studies men- Am J Clin Nutr 63:47–53, 1996 290S–298S, 2002 tioned above plus the present IRAS study, 4. National Research Council: Diet and 14. Willett W, Manson J, Liu S: Glycemic in- plus many others) that no one to date has Health: Implications for Lowering Chronic dex, glycemic load, and risk of type 2 di- found that the amount of carbohydrate Disease Risk. Washington, DC, National abetes. Am J Clin Nutr 76 (Suppl.):274S– eaten per day is significantly associated Academy Press, 1989 280S

DIABETES CARE, VOLUME 28, NUMBER 12, DECEMBER 2005 2979