| BackgroundThe glycemic index (GI) is a parameter for the evaluation of the blood glucose raising potential of carbohydrate foods. Many studies have shown a beneficial effect of low GI diets in chronic diseases such as type 2 diabetes, cardiovascular disease, obesity and cancer. Diet GI has become a useful tool in chronic diseases'prevention and control.However, the bioavailability of carbohydrates was different because of different food choices, cultural acceptability and individual likes. In fact, the GI of a carbohydrate-rich food can vary greatly depending on a number of factors including the variety, origin, processing, and preparation of the food and the other nutrients that are simultaneously consumed with the food. To evaluate the affecting factors would help us discover the mechanism of GI and promote the use of GI in heath maintenance and control of diseases.However, the determination of glycemic index of mixed diet is difficult and costful, so that, it is very useful to develop a model on prediction of diet GI. Three methods were found in those studies:determinations of carbohydrate digestion rate in vitro, sum of the weighted GI value and some equations on mixed meals. Unfortunately, theses models are still debatable. Chinese diets contain lots of kind food in one dish and are different from diets in west countries. We cannot use these methods to evaluate our diet. The development of GI model would have benefitial effects on prevention of chronic diseases.Objectives 1. To evaluate the factors affecting GI and discover the glycemic response of diet with different physical-chemical factors, food group and energy nutrients with the aim of providing basis for the prediction model.2. To develop prediction models of Chinese diet through human experiment and statistics methods.3. To test the validity of the prediction models for glycemic responses through different diets determined by volunteers in vivo, and to discover the mechanism of GI with digestion method in vitro, and with 13C analysis in breath, satiety and insulin responses.4. To explain the reliability of prediction model and to provide guides for wide application of GI.Materials and methods1. Development of diet glycemic index database and prediction modelsThe GI values were taken mainly from 2002 Chinese food composition data and 2002 international data of glycemic index. All data were expressed in terms of the same standard. Strict inclusion and exclusion criterion were made to screen food items. General liner model was used to develop the equation. R2, residual graph and residual Q-Q graph were used to test the validity of the equation. Then A simplified model was developed. Food items get predicted GI value back to the model, comparison of the difference between predicted and measured GI was made to test the predictive capability and accuracy.2. Test of energy nutrients on glycemic responseBy an orthogonal experimental design, totally 18 healthy young men were assigned to consume 9 testing diets composed of different levels of resistant starch, fat and Protion.. Each kind of testing food was taken by 2 subjects. The study was repeated once. Blood samples and breath samples were collected at different time points. GI and insulin index (Ⅱ) were calculated. Regression equation was established to compare the consistency between human test and GI prediction model. Metabolic rate of exogenous glucose could be measured by the analysis of 13CO2 in breath. The accumulated metabolic rate of 13C glucose was measured for 30 hours. The carbohydrate bioavailability of total 9 test diets was analyzed in vitro by the methods of Englyst. The reduced sugar release index (RRI2h), rapidly available glucose (RAG), slowly available glucose (SAG) was determined to evaluate the effect of protein and fat on the digestion in vitro.3. Test of physical-chemical factors on glycemic responseAccording to an orthogonal experimental design, a total of 18 healthy young men were assigned to consume 18 test diets composed of different levels of vinegar addition, water content and particle size. The study was repeated four times. Blood samples and satiety questionnaire were collected at different time points. GI,Ⅱand satiety index (SI) were calculated. Regression equation was established to compare the consistency between human test and GI prediction model. The RRI2h, RAG and SAG were determined by Englyst method.4. Test of Chinese typical dietHealthy volunteers consumed 10 test diets with different components of food and the response of blood glucose was tested. GI was calculated and Regression equation was established to compare the consistency between human test and GI prediction the GI value was calculated and evaluated based on national nutrition survey 2002.Results1. Development of diet glycemic index database and prediction modelsThe diet GI database contains 142 food items including 29 high GI foods,41 middle GI foods and 72 low GI foods. The regression equation of energy nutrients was:GI=48.84-1.26Prot.-0.1FAT+0.43CHO-0.89DF, R2=0.4672, the effects of these variables on GI are as follows in descending order:CHO> Prot.> DF> FAT. The regression equation of physical-chemical factors was:GI=67.02-19.02×vegetable and fruit-35.27×milk and beans-16.97×beverage+12.38×sugar-8.11×boiling, R2=0.4986. The effects of these variables in descending order are:milk and beans> vegetable and fruit> beverage> boiling. The regression equation of all infect factors was:GI= (8.01-0.04Prot.+0.01CHO-0.06DF-1.14×vegetable and fruit-1.83 X milk and beans-1.10×beverage-0.43×boiling-0.57×mixed meal-0.03FAT) 2, R2=0.5951, the effects of these variables in descending order are:milk and beans> vegetable and fruit> beverage> mixed meal> boiling>CHO>Prot.>DF>FAT。The simplified GI model was developed and the average relative error was 8%, the standard error 13.97, and the accuracy rate of discrimination 78.2%.2. Test of energy nutrients on glycemic responseIt was found that resistant starch was the most important factor affecting diet GI,Ⅱ, RRI2h the accumulated metabolic rate of 13C glucose 2-8 hours after food consumption (p<0.05). High level of RS could decrease these markers. Fat was the secondary factors affecting diet GI. High content of fat can lower GI value (P<0.05) and decrease the RAG (P<0.05). P did not influence postprandial glucose, insulin response, the RRI2h and the metabolic rate of 13C glucose (P>0.05). The glucose responses were related to the RS or fat content of the test meal (P<0.05). RS, fat content and Prot. accounted for 31.44% of the variation in the glycemic index. We found that the relationship between GI and energy nutrients was the same and their contribution to GI was consistent with GI prediction model, but the effects of fat and P on GI were different.3. Test of physical-chemical factors on glycemic responseThe foremost factor affecting GI,Ⅱand RRI2h was particle size, and vinegar addition affected SI most. The GI,Ⅱand SI value of meals were significantly different (P<0.05), the GI of the smallest size maize meals was higher than that of middle size and the largest size. TheⅡvalue of water content on 20:1 was higher than that of others, and the SI of 16ml addition vinegar was higher than others. The GI was not related to the vinegar addition, water content and particle size by regression analysis. The results confirmed the GI prediction model.4.Test of Chinese typical dietGI of 10 mixed meals showed significant differences (P<0.05). Protein and dietary fiber can reduce the blood glucose response and can be significantly related to GI. Fat can also inhibit the increment of blood glucose, but is not significantly related to GI. When Prot. was co-ingested with carbohydrate, the glycemic response reduced.The predicted GI by full factors GI model and measured GI were closely consistent. The average relative error was7% and the standard error was 8.85 between measured GI and the predicted GI by simplified GI. The predicted GI of Chinese typical diet was 52.1.Conclusions1. GI database were developed with strict inclusion criteria based on GI value. evaluation was made to discover the relationship between variables and GI.The GI prediction model was developed based on the GI database:GI= (8.01-0.04×Prot. +0.01CHO-0.06DF-1.14 X vegetable and fruit-1.83×milk and beans-1.10×beverange-0.43×boiling-0.57×mixed meal-0.03FAT) 2.2. The energy nutrients accounted for 47% of the variation in the GI. The effects of variables in energy nutrients was CHO> Prot.> DF> FAT in descending order. The physical-chemical factors accounted for 50%of the variation in the GI. The effects of variables in physical-chemical factors were:milk and beans> vegetable and fruit> beverage> boiling. All factors accounted for 60% of the variation in the GI. The effects of variables in all factors affecting GI was milk and beans> vegetable and fruit> beverage> mixed meal> boiling>CHO>Prot.>DF>FAT in descending order.3. The simplified GI model was developed and the average relative error was 8%, the standard error was 13.97, the accuracy rate of discrimination was 81.2%.4. The validity of GI prediction model was tested through experiments with multifactor in vivo. It showed good consistency between measured GI and the predicted GI, the same consistency is found with the effects of variables. However, the contribution of different variables is not identical.As a result, the prediction model was valid and reliable.5. The mechanism of experimental factors is as follows:The RS could reduce the glycemic response by decrease the digestion and absorbtion of glucose in intestine; the fat could decrease glycemic response by slowing down the rate of Gastric emptying; big partical size of food could decrease the glycemic response by parceling the starch and reducing the interface area of digest enzyme.Based on the above findings, we concluded that the GI prediction model and simplified model were effective in predicting the diet glycemic index. But more studies are needed to understand the effects and mechanisms of energy nutrients and other factors affecting glycemic response.
|
PDF Full Text Request