Journal of the American College of Nutrition

ISSN: 0731-5724 (Print) 1541-1087 (Online) Journal homepage: http://www.tandfonline.com/loi/uacn20

The Association Between Dietary Glycemic Index and Glycemic Load and a and Fat Distribution Among Apparently Healthy Iranian Adults

Abdul-Mutala Fuseini, Mohammad Hossein Rahimi, Mehdi Mollahosseini, Mir Saeed Yekaninejad, Zhila Maghbooli & Khadijeh Mirzaei

To cite this article: Abdul-Mutala Fuseini, Mohammad Hossein Rahimi, Mehdi Mollahosseini, Mir Saeed Yekaninejad, Zhila Maghbooli & Khadijeh Mirzaei (2018): The Association Between Dietary Glycemic Index and Glycemic Load and a Body Shape and Fat Distribution Among Apparently Healthy Iranian Adults, Journal of the American College of Nutrition, DOI: 10.1080/07315724.2017.1416312 To link to this article: https://doi.org/10.1080/07315724.2017.1416312

Published online: 13 Mar 2018.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=uacn20 JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION https://doi.org/10.1080/07315724.2017.1416312

The Association Between Dietary Glycemic Index and Glycemic Load and a Body Shape and Fat Distribution Among Apparently Healthy Iranian Adults

Abdul-Mutala Fuseinia, Mohammad Hossein Rahimib, Mehdi Mollahosseinic, Mir Saeed Yekaninejad d, Zhila Maghboolie, and Khadijeh Mirzaeic aDepartment of Community Nutrition, Tehran University of Medical Sciences–International Campus, Tehran, Iran; bDepartment of Community Nutrition, Tehran University of Medical Sciences, Tehran, Iran; cDepartment of Community Nutrition, School of Nutritional Sciences and Dietetics, Tehran University of Medical Sciences, Tehran, Iran; dDepartment of Epidemiology and Biostatistics, School of Public Health, Tehran University of Medical Sciences, Tehran, Iran; eEndocrinology and Metabolism Clinical Sciences Institute, Tehran University of Medical Sciences, Tehran, Iran

ABSTRACT ARTICLE HISTORY Objective: The role of dietary glycemic index (GI) and glycemic load (GL) in the development of has Received 28 September 2017 been debated globally. The relationship with body shape and fat distribution was examined in this cross- Accepted 7 December 2017 sectional association study among apparently healthy Iranian adults. KEYWORDS Methods and materials: A study population of 265 (126 males and 139 females) aged 18–55 years Glycemic index; glycemic participated in this cross-sectional study from the communities of Tehran based on cluster sampling. GI load; fat mass; fat free mass; and GL were assessed by the 147-item food frequency questionnaire (FFQ) completed by a trained a body shape index; body dietitian. Weight, height, waist circumference (WC), and hip circumference of the participants were mass index; obesity; fat measured, and (BMI), waist-to-hip ratio (WHR), and A Body Shape Index (ABSI) were distribution; waist further calculated. Fat mass and fat-free mass were also measured using a body composition analyzer, and circumference fat mass index (FMI) and fat-free mass index (FFMI) were then calculated. Multivariate regression models were fitted to assess the association between GI/GL and fat distribution measures such as FMI, FFMI, WC, BMI, WHR, and ABSI, considering potential confounding factors such as sex, age, BMI, and physical activity. Results: There was a statistically significant inverse association between GL and WC, BMI, and ABSI found in the adjusted model. GL was inversely associated with WC for both the adjusted model (p-trend D 0.027) and the crude model. Also, an inverse association was seen between GL and BMI (p-trend D 0.019) in the adjusted model but a marginal association in the crude model. GL was also inversely associated with ABSI (p-trend D 0.089) in the highest tertile. Conclusion: Dietary GL but not GI is inversely associated with fat distribution measures such as WC, BMI, and ABSI in the study population. This result may suggest a beneficial role of higher-GL diets in the prevention of obesity.

Introduction prevalence of obesity, a significant risk factor for several chronic According the World Health Organization (WHO), more than diseases, in the United States has more than doubled during the 1.9 billion adults (18 years and older) are , with over past 25 years (6). For instance, in 2008, more than 60% of adults in 600 million obese (1). Obesity has been associated with increased the United States were categorized as either overweight or obese morbidity and mortality attributable to , , and (7). Popkin in his article stated that adult obesity levels, adult-onset many other conditions (2,3). Obesity has gain global attention; diabetes, and many other noncommunicable diseases are increas- however, most researchers now focus on the pattern of body fat dis- ing far more rapidly in developing countries than in the developed tribution, which seems more important than overall body mass. A countries and further stated that overweight and obesity levels of research conducted among Tehranian adults revealed that 67% of some lower-income countries match, or exceed, those of the United women and 33% of men older than 20 years are suffering from States (8). Also, the trends of obesity are so rapid that many abdominal obesity (4). The epidemic obtrudes not only a decline in researchers predict countries such as China will record a marked the standard of living and high health care burden but also increase increase in total adult mortality rates over the next several decades in mortality rate. Studies have shown that risk factors such as genet- (9), with same manifestation in Iran, since they all fall in the same ics and environment play a crucial role in the development of this geographic location and seem to have similar dietary patterns. condition, with epidemiologic studies revealing habitual as one Most of these studies have indicated that for obesity epidemics to of the most significant factors. be tackled, a change in diet and food systems must be first Studies have shown that obesity prevalence and morbidity and addressed, since it is one of the strategies that are simple and cost- mortality rates have increased significantly worldwide (5).The effective.

CONTACT Khadijeh Mirzaei [email protected] Department of Community Nutrition, School of Nutritional Sciences and Dietetics, Tehran University of Medical Sciences, No. 44, Hojjat-dost Alley, Naderi Street, Keshavarz Boulevard, Tehran, Iran. © 2018 American College of Nutrition 2 A.-M. FUSEINI ET AL.

Carbohydrates are primary nutrient, thus the glycemic index Waist-to-hip ratio (WHR) and A Body Shape Index (ABSI) (GI) and glycemic load (GL) represent different aspects of this were then calculated. macronutrient. GI represents the increment of the area under the glycemic curve after ingestion of 50 g of available carbohy- Complete body composition analysis drate from a test food compared with a standard food (10).GL is the product of the GI and its actual carbohydrate content A body composition analyzer (model BC-418 MA; Tanita, xx, (11). The GI/GL ranks foods according to their effect on the UK) was used to assess the body composition of all the partici- glycemic response, that is, high glucose response (high GI/GL) pants. This equipment is designed to send out a very weak elec- or low glucose response (low GI/GL). Low-GI/GL diets have tric current for measuring the impedance (electrical resistance) been demonstrated to reduce postprandial spikes of blood glu- of the body. Participants were barefoot when they were assessed cose level in healthy individuals (12). A diet with high GI/GL by this device. Taking measurements after vigorous may lead to hyperinsulinemia and insulin resistance (13) as a should be avoided; therefore, we waited until the subject was result of increase in the blood glucose concentration. Increase sufficiently rested, for preventing a possible discrepancy in insulin may further affect lipolysis, lipogenesis, or substrate oxi- measured values. Also measurements were performed in the dation. This may affect hunger, satiety, food intake, or energy morning in a fasting condition (e.g., always urinating before expenditure and hence affect energy balance and body compo- taking measurement, etc.). The device calculates body fat per- sition (14). centage, fat mass, and fat-free mass and estimates muscle mass Considering the high consumption of carbohydrates as sta- on the basis of data using bioelectrical impedance analysis ple food, few studies to the best of our knowledge have been (BIA). Therefore, important body composition components conducted in Tehran regarding the effects of GI and GL among reported for our current study are fat percent, fat mass, and apparently healthy adults, and those few studies that have free-fat mass. Fat mass index (FMI) and fat-free mass Index explored these relationships globally so far are inconsistent. In (FFMI) were calculated as well. this regard, the current research is design to determine the association between dietary GI and GL and a body shape and Dietary intake assessment fat distribution among apparently healthy Iranian adults. Participants consumed their usual diet. They were instructed to fill the 147-item food frequency questionnaire (FFQ) that was Materials and methods validated in a previous studies (15). Questionnaires were com- Subjects pleted by a trained dietitian. Data were recorded in household measures servings and then converted into grams and millili- A study population of 265 individuals (126 males and 139 ters. The Nutritionist 4 (First DataBank, San Bruno, CA, USA) – females) aged 18 55 years participated in this cross-sectional food analyzer was used to analyze the dietary intake data. study. Study population was recruited from the communities of central and west of Tehran based on cluster sampling. All par- Glycemic index calculation ticipants gave written informed consent. Subjects were chosen according to some inclusion and exclusion criteria. Participants Dietary GI and GL calculated were adjusted for total caloric – were included if the following applied: ages of 18 55 years, do intake by using the residuals method (16). The subjects partici- not consume alcohol or abuse drug, absence of any acute or pating in the present study were tested during 3–5 discrete fl chronic in ammatory disease, no history of hypertension, and occasions in the morning after they had fasted overnight. On 2 not being pregnant. Participants were excluded if the following occasions, the subjects ate test meals comprising one of the test applied: drink alcohol or abuse drugs, had history of hyperten- foods—the portion size of each test food contained 50 g avail- sion, being pregnant, currently smoking, having thyroid, able carbohydrate. The test meal, on the other occasions, hepatic, renal, or cardiovascular diseases, heart failure, malig- included the reference food, which could be 50 g anhydrous nancies, or diabetes mellitus, being in any acute or chronic glucose, 55 g dextrose (glucose monohydrate), or 50 g available fl fl in ammatory state that affects in ammatory markers, and any carbohydrate from white bread. After a fasting blood sample kind of infection. The study was approved by local ethics com- was drawn on every occasion, the subjects consumed the test mittee of the Endocrinology and Metabolism Research Center meal. Further blood samples were taken at 15, 30, 45, 60, 90, of Tehran University of Medical Sciences (ethic number: 93- and 120 minutes after they began to eat. For each test food, the 04-161-27722-149580). area under the curve (AUC) for each subject was shown as a percentage of the mean AUC obtained with the reference food Anthropometric assessments in the same subject. The mean of these values for all the sub- jects was the food GI. If white bread was applied as the refer- The weight and height of participants were measured in light ence food, the GI values were multiplied by 0.71 to transform clothing and barefoot, respectively, and then used to compute them to the glucose scale (i.e., the GI of glucose D 100). GL val- the body mass index (BMI) of the participants. Waist circum- ues were further calculated by multiplying the available carbo- ference (WC) was measured in the slimmest area while subjects hydrate content of each food by its GI value and then were at the end of a normal exhalation by a nonelastic tape multiplying the resultant value with the amount of consump- with accuracy of 0.1 cm. Also, hip circumference (HC) was tion (divided by 100) and then summing the values from all measured in the largest part of the hip over light clothing. food items (17). JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION 3

Statistical analysis between the lowest tertiles of both GI and GL comparing with the highest tertiles. Further, also no statistically differences All statistical analysis was performed using Statistical Package were observed between the highest tertiles of GI when compar- for Social Sciences (version 20; IBM, Armonk, NY, USA). ing with the lowest tertiles for both WC and BMI. However, Results are presented as mean § SE, and estimates of effects as there was a decrease in mean values from the lowest tertiles to means and 95% confidence intervals. Statistical significance the highest tertiles for both WC and BMI. Our study also was set at p < 0.05. Participants were categorized on the basis revealed that ABSI values were smaller than those in Krakaur’s of energy-adjusted GI and GL tertiles. Tertiles were chosen to study ([mean § SE?] 0.0784 § 0.0047 vs 0.0816 § 0.0053), illustrate low, medium, and high GI and GL groups. Variables although the difference seems insignificant (Table 1). were compared across tertile categories of GI and GL by 1-way Multivariate logistic regression was used to quantify the analysis of variance (ANOVA) with Tukey’s post hoc compari- strength of association between energy-adjusted GI/GL and fat sons for quantitative variables and chi-square test for categori- distribution parameters. After adjustment for age, sex, BMI, cal variables. and physical activity, GL was inversely associated with ABSI Logistic regression models were used to estimate odds (OR D 0.496, 95% CI D 0.25–0.99; p-trend D 0.089) in the ratios (ORs) and 95% confidence intervals (CIs) for the vari- highest tertile but with a reduced odds ratio. Also, a borderline ous categories of FMI, FFMI, WC, WHR, BMI, and ABSI significant association was seen between GL and ABSI (OR D across tertiles of GI/GL in crude and multivariable-adjusted 0.537, 95% CI D 0.26–1.09) in the second tertile. Although GI models of each category, while controlling for potential con- was not significantly associated with ABSI, higher GI levels founding variables such as age, sex, physical activity, and were associated with increased odds of ABSI (OR D 1.033, 95% BMI. CI D 0.53–2.02 and OR D 1.086, 95% CI D 0.56–2.11) (Table 2). The model also revealed an inverse association between GL Results and WC for both the crude and adjusted models. In the The study sample consisted of 126 (47.5%) men and 139 adjusted model, GL was inversely associated with WC in (52.5%) women. The mean values § SE of dietary GI and die- both the highest tertile (OR D 0.345, 95% CI D 0.13–0.92) and tary GL adjusted for total energy intake were 61.5 § 0.2 and the lowest tertile (OR D 0.307, 95% CI D 0.11–0.90), with 197.8 § 3.4, respectively. Forty percent (40.0%) of men and p-trend D 0.027 (Table 3). 38.2% of women were overweight (30 kg/m2 > BMI 25 kg/ Also, in the crude model, there was an inverse association m2). Also, 16.8% of the participants were obese (BMI 30 kg/ between tertiles of GL and FFMI (OR D 0.434, 95% CI D 0.23– m2) and 59% had abdominal obesity (WC 94 cm for men or 0.82 and OR D 0.457, 95% CI D 0.24–0.86; p-trend D 0.012). 80 cm for women). There was also a significant difference in However, after adjusting for possible confounding factors, no BMI between the 1st and 3rd tertiles of GL (24.7 vs 27.0) intake. association was observed between GL and FFMI (OR D 0.362, For ABSI, no statistically significant differences were seen CI D 0.11–1.18) in the highest tertile (Table 4). Further, an

Table 1. Energy-adjusted characteristics by tertiles of GI/GL. Energy-adjusted GI Energy-adjusted GL

Variable T1 T2 T3 p valuey T1 T2 T3 p valuey

Characteristics GI 57.53 § 0.28 62.00 § 0.09 65.01 § 0.11 <0.001 199.96 § 4.70 191.04 § 5.57 212.87 § 3.35 0.010 GL 61.71 § 0.25 60.91 § 0.38 61.87 § 0.47 0.157 141.95 § 3.11 192.53 § 1.30 258.94 § 3.65 <0.001 n (%) 86(33.6) 85(33.2) 85(33.2) 0.503 86(33.5) 85(33.1) 86(33.5) 0.115 Age (years) 35.8 § 8.7# 34.3 § 9.1 34.7 § 8.7 36.6 § 8.3 34.3 § 8.7 34.0 § 9.2 Sex Male, n (%) 28(32.6) 44(51.8) 51(60.0) 0.001 35(40.7) 38(44.7) 50(58.1) 0.056 Female, n (%) 58 (67.4) 41(48.2) 34(40.0) 51(59.3) 47(55.3) 36(41.9) PA Low, n (%) 38(46.3) 29(35.4) 35(42.2) 0.554 36(43.9) 34(42.0) 32(37.6) 0.553 Medium, n (%) 39(47.6) 47(57.3) 45(54.2) 40(48.8) 45(55.6) 47(55.3) High, n (%) 5(6.1) 6(7.3) 3(3.6) 6(7.3) 2(2.5) 6(7.1) Fat distribution indices WC (cm) 89.2 § 13.8 88.6 § 13.4 87.9 § 10.0 0.802 85.3 § 13.1a 87.8 § 11.3b 92.5 § 12.1a,b 0.001 WHR 0.86 § 0.07 0.87 § 0.08 0.86 § 0.07 0.406 0.85 § 0.08a 0.86 § 0.07 0.88 § 0.07b 0.022 FM (kg) 21.0 § 9.2a 18.7 § 8.8 17.2 § 7.3a 0.020 17.4 § 8.6 19.0 § 8.0 20.5 § 8.9 0.077 FMI (kg/m2) 7.6 § 3.5a 6.7 § 3.3 6.2 § 2.9a 0.015 6.4 § 3.3 6.9 § 3.2 7.2 § 3.4 0.357 FFM (kg) 53.0 § 12.9 54.4 § 12.8 55.4 § 10.8 0.454 51.1 § 11.0a 53.5 § 11.4b 58.6 § 12.9a,b <0.001 FFMI (kg/m2) 18.8 § 3.1 19.0 § 3.0 19.2 § 2.6 0.720 18.3 § 2.6a 18.8 § 2.6 19.9 § 3.2a 0.002 ABSI 0.08 § 0.005 0.08 § 0.005 0.08 § 0.004 0.779 0.08 § 0.005 0.08 § 0.005 0.08 § 0.004 0.244 BMI (kg/m2) 26.5 § 5.4 25.7 § 5.1 25.4 § 3.9 0.134 24.7 § 4.8a 25.9 § 4.7 27.0 § 5.0a 0.010

Note.GID glycemic index; GL D glycemic load; T D tertile; PA D physical activity; WC D waist circumference; WHR D waist-to-hip ratio; FM D fat mass; FMI D fat mass index; FFM D fat-free mass; FFMI D fat-free mass index; ABSI D A Body Shape Index; BMI D body mass index. N D 254 (men and women). Values not sharing common superscripts (a, b) are significantly different in a row using post hoc (Tukey) procedure. yValues were determined by analysis of variance for quantitative variables. 4 A.-M. FUSEINI ET AL.

Table 2. Correlations of GI/GL and ABSI in logistic regression models.a

Crude model Adjusted modelb

Variable OR 95% CI p value p-trendc OR 95% CI p value p-trendc

GI tertiles Tertile 2 1.080 0.59–1.99 0.806 0.964 1.508 0.76–3.01 0.243 0.245 Tertile 3 1.165 0.62–2.18 0.634 1.454 0.73–2.90 0.287 GL tertiles 0.156 0.089 Tertile 2 0.635 0.34–1.19d 0.155 0.537 0.26–1.09 0.086 Tertile 3 0.508 0.27–0.95 0.034 0.496 0.25–0.99 0.046

Note.ORD odds ratio; CI D confidence interval; GI D glycemic index; GL D glycemic load. ABSI entered in the model as low or high (based on median). aOdds ratios and 95% CIs were calculated with the use of logistic regression model for a body shape index categorization in 2 groups of GI tertiles, with tertile 1 as the ref- erence point. bAdjusted for age, sex, BMI, and PA. cTest for trend based on variable containing median value for each tertile. d95% confidence interval. inverse association was observed between GL and BMI (OR D high consumption of GL comprising fruits, vegetables, and 0.437, 95% CI D 0.22–0.87 and OR D 0.518, 95% CI D 0.54– legumes seems to reduce hyperinsulinemia and increase satiety to 1.02; p-trend D 0.019) in the adjusted model but a marginal reduce food intake. This will further increase lipolysis activity to association in the crude model (Table 5). No significant associ- reduce postprandial insulin response, since high levels of insulin ation was seen between GL and FMI and WHR. There was no inhibits lipolysis activity. Even though the molecular details of the significant association observed between GI and FMI, FFMI, lipolytic reaction are not fully understood, it has been stated that WC, ABSI, BMI, and WHR in the adjusted model, but a signifi- lipolysis is involved in the oxidation of the released fatty acids in cant association was observed in the crude model between GI order to decrease the stored fat. This property makes it a drug tar- and FMI (data not shown). get for the treatment of obesity and through the proteins involved in this process. Therefore, a high-GL diet may contain nutritional characteristics that will encourage lipoly- Discussion sisorhaveafat-burningeffect.Hence,alipolyticdietplancon- The results obtained in our current study revealed that both the taining high amount of vegetables, fruits, whole-grain cereals, and quality and quantity of the carbohydrate rather than the quality legumes may be beneficial for a healthy body composition. alone may be associated with reducing fat accumulation among The inverse association seen between GL and the fat distribu- Iranian adults. tion parameters may be a result of the nature of GL, which Some studies have indicated that a potential connection reflects both the quantity and quality of carbohydrate in the diet between GI and obesity is a result of lipogenic effects of hyperin- of the participants. This result could mean that a higher-GL diet sulinemia (18) where high GI/GL is noted to suppress lipolysis. made up of whole-grain cereals, fruits, vegetables, as well as Hence, a low-GI/GL diet has been hypothesized to elicit a low legumes is healthier than a low-GI diet. According to the interna- insulin response and may be beneficial for manage- tional dietary guidelines, people who fellow a high-GL diet may ment. A possible mechanism suggested by researchers is that the be more health conscious than those who do not follow such rec- higher postprandial insulin response following a high-GI/GL diet ommendation (21). This suggests that a high-GL diet may protect consumption may lead to a quicker hunger response and over- against abdominal obesity and could be of clinical relevance. eating by depleting the metabolic fuels in the body (19).Also, The study revealed an inverse association between GL and another possible mechanism of action may be through increased ABSI, WC, and BMI in a multivariate logistic regression model. satiety and decreased voluntary food intake (20).Inthisstudy, An inverse association between dietary GL and BMI was seen

Table 3. Correlations of GI/GL and WC in logistic regression models.a

Crude model Adjusted modelb

Variable OR 95% CI p value p-trendc OR 95% CI p value p-trendc

GI tertiles Tertile 2 1.175 0.64–2.17d 0.607 0.610 0.587 0.22–1.58 0.292 0.300 Tertile 3 1.288 0.69–2.42 0.432 1.760 0.64–4.86 0.275 GL tertiles Tertile 2 0.461 0.24–0.88d 0.018 0.019 0.307 0.11–0.90 0.031 0.027 Tertile 3 0.508 0.27–0.96 0.038 0.345 0.13–0.92 0.034

Note.ORD odds ratio; CI D confidence interval; GI D glycemic index; GL D glycemic load. WC entered in the model based on cutoffs for men and women (men: <94 cm for normal, 94 cm at risk; women: <80 cm for normal, 80 at risk). aOdds ratios and 95% CIs were calculated with the use of logistic regression model for waist circumference categorization in 2 groups of GI tertiles, with tertile 1 as the reference point. bAdjusted for age, sex, BMI, and PA. cTest for trend based on variable containing cutoff values for each tertile. d95% confidence interval. JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION 5

Table 4. Correlations of GI/GL and FFMI in logistic regression models.a

Crude model Adjusted modelb

Variable OR 95% CI p value p-trendc OR 95% CI p value p-trendc

GI tertiles Tertile 2 0.640 0.34–1.19d 0.158 0.158 0.789 0.24–2.64 0.701 0.686 Tertile 3 0.822 0.44–1.53 0.537 0.755 0.25–2.27 0.617 GL Tertiles Tertile 2 0.434 0.23–0.82d 0.011 0.012 1.129 0.35–3.68 0.841 0.809 Tertile 3 0.457 0.24–0.86 0.016 0.362 0.11–1.18 0.092

Note.ORD odds ratio; CI D confidence interval; GI D glycemic index; GL D glycemic load. FFMI entered in the model as low or high (based on median). aOdds ratios and 95% CIs were calculated with the use of logistic regression model for fat-free mass index categorization in 2 groups of GI tertiles, with tertile 1 as the ref- erence point. bAdjusted for age, sex, BMI, and PA. cTest for trend based on variable containing median value for each tertile. d95% confidence interval. and is consistent with the results of Mendez and Colleagues GL and obesity; hence, the likelihood of being obese appeared (22) in a Mediterranean population study and in a univariate lower in the highest quartiles of total carbohydrates and dietary analysis (23) of a total-population study to find the association GL (32). In a review article (33) of 11 cohort studies, the GL of between dietary fiber, GI, GL, and BMI as well as in one other the diet was either not significantly related to BMI or there was Italian study (24). It may be a result of high intake of foods a trend for the two to be inversely related. This suggests that high in fiber, such as fruits, vegetables, legumes, and whole- BMI decreases as GL increases. The studies went further to grain cereals, and also diets low in calories, as stated by 2 stud- indicate how different the food patterns of diets with high GI ies (25,26). The main characteristics of the Iranian diet is con- were from those of high GL, taking into consideration a diet suming large amount of carbohydrate (27), mainly from cereals rich in vegetables, fruits, and legumes was related positively to and potatoes as well as some mono- and disaccharides from high GL but negatively to high GI. milk and fruits. This may imply a beneficial role of higher-GL A Danish cohort study of adults found an inverse associa- diets in the prevention of general obesity due to the combined tion between carbohydrate intake from vegetables and fruits effect of the Iranian eating habits. Also, the study did not reveal (women only) and change in WC (34). Since dietary GL acts any association between dietary GI and BMI, which is consis- as a surrogate for total carbohydrate intake (35), a stronger tent with 2 studies that found no association between dietary correlation therefore exists between total carbohydrate and GI and BMI in both men and women (28,29). One of the rea- GL and hence is consistent with the current study, which sons could be that high-carbohydrate diets may override any also found an inverse association between GL and WC. The effects of GI (30). A number of studies that looked at carbohy- study results indicate that a lower consumption of carbohy- drate intakes also found no significant relationship between GI drates rich in fruits and vegetables may predict abdominal and BMI, and most epidemiologic studies also showed an fat accumulation. Another study that aimed to evaluate the inverse relationship between carbohydrate intake and BMI and relationship between energy-generating nutrients and the hence concluded that a low-fat, fiber-rich carbohydrate may be presence of central and overall obesity after correcting for beneficial for health and weight control (31). In a pooled analy- sociodemographic, lifestyle, and clinical characteristics sis of 3 cross-sectional population-based studies to assess die- among healthy elders revealed that a 1% increase in dietary tary carbohydrate quantity and quality in relation to obesity, an CHO predicted a 12% lower likelihood of central adiposity, inverse association was found between total carbohydrate and but a low-GI diet was not as effective as increase in total

Table 5. Correlations of GI/GL and BMI in logistic regression models.a Crude model Adjusted modelb

Variable OR 95% CI p value p-trendc OR 95% CI p value p-trendc

GI tertiles Tertile 2 0.986 0.54–1.82d 0.964 0.961 1.033 0.53–2.02 0.923 0.925 Tertile 3 1.045 0.56–1.94 0.888 1.086 0.56–2.11 0.808 GL tertiles Tertile 2 0.551 0.30–1.03d 0.062 0.066 0.437 0.22–0.87 0.018 0.019 Tertile 3 0.552 0.29–1.04 0.064 0.518 0.54–1.02 0.058

Note.ORD odds ratio; CI D confidence interval; GI D glycemic index; GL D glycemic load. BMI entered in the model based on cutoffs (<24.9 kg/m2 and 25 kg/m2). aOdds ratios and 95% CIs were calculated with the use of logistic regression model for body mass index categorization in 2 groups of GI tertiles, with tertile 1 as the refer- ence point. bAdjusted for age, sex, and PA. cTest for trend based on variable containing cutoff values for each tertile. d95% confidence interval. 6 A.-M. FUSEINI ET AL.

CHO for reducing the likelihood of central adiposity (36). association was seen in the second tertile but an inverse sig- The authors concluded that a diet higher in carbohydrate is nificant association in the third tertile. This indicates that associated with a lower likelihood of becoming obese as consumption of high-GL diets may be associated with compare with a low-GI diet. It is very obvious that this find- reduced risk of fat accumulation and hence low risk of mor- ing has clinical relevance, since WC is often used as a proxy tality. Again, ABSI seems to more accurately depict the vari- for abdominal fat mass and is associated with cardiometa- ability in circulating insulin. Therefore, a high-GL diet will bolic disease risk (37). In contrary to the study findings, a improve insulin sensitivity and further help insulin-produc- cross-sectional study of 979 adults with normal and impaired ing cells in the pancreas to work effectively in reducing glucose tolerance from the Insulin Resistance hyperinsulinemia. To our knowledge, there are no published study revealed no association for GL with WC (28). A possi- epidemiologic data with which to compare our data on GL ble heterogeneity in the content and the amount of carbohy- and ABSI. drate intake as well as the sample size in various populations The strengths of this research are that studies that have been may explain these differences. Considering the null associa- conducted regarding GI/GL in this population were few, espe- tion between GI and WC, some studies that have seen an cially among healthy individuals, and therefore it will be a good association between GI and WC also suggest that a low-GI contribution to the literature. Also, the new measure of mortal- diet may play a crucial role in the prevention of abdominal ity risk, ABSI, to the best of our knowledge is the first to be obesity (38), in which our current study may accurately used as a fat distribution parameter in finding its association reflect the situation. One reason could be the fact that low- with GI/GL and has provided evidence that it may be one of GI diets are able to induce satiety. For instance, a study of the better predictors of fat accumulation. 22 overweight women revealed that low-GI meals compared Some of the possible limitations to this study are the follow- with high-GI meals reduced hunger as well as food intake ing: it was cross-sectional, hence limiting inference on the time and caused greater satiety (39). One other reason for lack of sequence of the associations. Also, the small sample size from association between GI and the fat distribution parameters the population limited the statistical power, since a greater maybethefactthattheGIofthedietalonemaynotbe power requires a larger sample size considering the overall pop- enough to predict whether the diet will be of good health ulation of Tehran. Further, besides the measurement errors, the effects or not. For instance, a study has shown that a restric- FFQ may introduce as a result of omission or isolation of part tion in consuming some foods in low-GI/GL diets might be of the meal eaten, it was not specifically designed to be used associated with lower intakes of some nutrients, which could also for the calculation of GI values; hence, no validation data be a risk factor for chronic diseases (27).FiberandGI/GL exist regarding its estimation. Another limitation was the use of share some common mechanisms by which they may affect questionnaire to collect data on physical activity, which shows abdominal fat accumulation (40). A prospective study among a limited reliability and validity, since there may be some level adolescence to examine carbohydrate nutrition and develop- of misclassification of participants to low, moderate, and high ment of adiposity revealed that increased dietary GL and physical activity. carbohydrate intake was associated with concurrent develop- ment of adiposity in girls and explained further that the inconsistency with other studies may be a result of differen- Conclusion ces in study characteristics, dietary assessment methods In conclusion, our findings indicated an inverse association used, and potential confounding factors considered (41). between GL and measures of central (WC) and general (BMI) The ABSI, which has been proposed as alternative to BMI obesity as well as ABSI. This may imply a beneficial role of and WC, is a more effective predictor of mortality. A num- higher-GL diets in the prevention of both general and central ber of criticism has been leveled against the use of BMI, for obesity. It is also crucial to note that both the quality and quan- not being able to distinguish between fat mass and muscle tity of the carbohydrate rather than the quality alone may be mass or reflectanindividual’s fat distribution (42,43). Also, necessary as a recommendation for overweight individuals who most studies have shown a very high simple correlation intend to lose weight. between BMI and WC (44-46), with the current study not excepted (p D 0.016). On this note, using WC and BMI alone for the assessment of fat distribution may still cast Acknowledgments doubt considering the outcome due to their correlation and The authors thank the directors of the Endocrine Diseases and Metabolism therefore the need to use the newly introduced measurement, Research Institute, and Tehran University of Medical Sciences, for allowing ABSI, in order to address some of these challenges. The ABSI them to conduct current cross-sectional study. seems to be superior over the traditional methods, since it can predict premature mortality (46),althoughsomestudies Funding found ABSI not to be a suitable measurement to identify car- diovascular disease (CVD) risk compared with BMI and WC This study was supported by grants 93-04-159-28031 and 93-04-161- (47), others found a similar ability with BMI as well as WC 27722 from Tehran University of Medical Sciences. for new onset of diabetes mellitus (48). However, some rea- sons for these varied results could be the difference in the ORCID study population as well as study designs used, hence creat- ing a gap for further research. For ABSI, a marginal Mir Saeed Yekaninejad http://orcid.org/0000-0003-3648-5276 JOURNAL OF THE AMERICAN COLLEGE OF NUTRITION 7

References 21. Alexander J. Pohjoismaiden m: Nordic nutrition recommendations, NNR 2004: integrating nutrition and physical activity. Copenhagen 1. World Health Organization. Obesity and overweight. In. Obesity and (Denmark): Nordic Council of Ministers; 2005. xx p. overweight; june. 2014:2016. [Publication date] [cited YYYY MMM 22. Mendez MA, Covas MI, Marrugat J, Vila J, Schroder H. Glycemic DD]. Available from: [URL] load, glycemic index, and body mass index in spanish adults. The 2. Abete I, Parra MD, Zulet MA, Martinez JA. Different dietary strategies American journal of clinical nutrition. 2009;89(1):316–322. for weight loss in obesity: Role of energy and macronutrient content. doi:10.3945/ajcn.2008.26444. Nutr Res Rev. 2006;19(1):5–17. doi:10.1079/NRR2006112. 23. Lau C, Toft U, Tetens I, Richelsen B, Jorgensen T, Borch-Johnsen 3. Cheah MH, Kam PC. Obesity: Basic science and medical aspects rele- K, Glumer C. Association between dietary glycemic index, glyce- vant to anaesthetists. Anaesthesia. 2005;60(10):1009–1021. doi:10.1111/ mic load, and body mass index in the Inter99 study: Is underre- j.1365-2044.2005.04229.x. porting a problem? The American journal of clinical nutrition. 4. Azadbakht, Mirmiran, Shiva, Azizi. General obesity and central adi- 2006;84(3):641–645. posity in a representative sample of tehranian adults: Prevalence and 24. Rossi M, Bosetti C, Talamini R, Lagiou P, Negri E, Franceschi S, La determinants. International journal for vitamin and nutrition Vecchia C. Glycemic index and glycemic load in relation to body research. 2005;75(4):297–304. doi:10.1024/0300-9831.75.4.297. mass index and waist to hip ratio. European journal of nutrition. 5. World Health Organization Noncommunicable diseases: country pro- 2010;49(8):459–464. doi:10.1007/s00394-010-0104-0. files. 2011 WHO Global Report. Geneva (Switzerland): World Health 25. Ledikwe JH, Blanck HM, Kettel Khan L, Serdula MK, Seymour JD, Organization;2011. xx p. Tohill BC, Rolls BJ. Dietary energy density is associated with energy 6. Manson JE SP, Greenland P, VanItallie TB,. The escalating pandemics intake and weight status in us adults. The American journal of clinical of obesity and . A call to action for clinicians. Arch nutrition. 2006;83(6):1362–1368. Intern Med. 2004;164:249–258. doi:10.1001/archinte.164.3.249. 26. van Dam RM, Seidell JC. Carbohydrate intake and obesity. Eur J Clin 7. Centers for Disease Control and Prevention (CDC). Behavioral risk Nutr. 2007;61(Suppl 1):S75–S99. doi:10.1038/sj.ejcn.1602939. factor surveillance system. Atlanta (GA): Department of Health and 27. Azadbakht L, Mohammadifard N, Akhavanzanjani M, Taheri M, Gol- Human Services,Centers for Disease Control and Prevention;2008. shahi J, Haghighatdoost F. The association between dietary glycemic [cited 2017 Sep 05]. Available from: [URL] index, glycemic load and diet quality indices in Iranian adults: Results 8. Popkin BM. The shift in stages of the nutrition transition in the devel- from Isfahan healthy heart program. International Journal of Food oping world differs from past experiences! Public Health Nutrition. Sciences and Nutrition. 2016;67(2):161–169. doi:10.3109/ 2002;5(1A):205–214. 09637486.2015.1134443. 9. Dong Y, Gao, W., Nan, H., et al. Prevalence of in urban 28. Liese AD, Schulz M, Fang F, Wolever TM, D’Agostino RB, Sparks KC, and rural Chinese populations in Qingdao, China. Diabetic . Mayer-Davis EJ. Dietary glycemic index and glycemic load, carbohy- 2005;22(10):1427–1433. doi:10.1111/j.1464-5491.2005.01658.x. drate and fiber intake, and measures of insulin sensitivity, secretion, 10. Jenkins DJ, Wolever TM, Taylor RH, Barker H, Fielden H, Baldwin and adiposity in the insulin resistance atherosclerosis study. Diabetes JM, Bowling AC, Newman HC, Jenkins AL, Goff DV. Glycemic index care. 2005;28(12):2832–2838. doi:10.2337/diacare.28.12.2832. of foods: A physiological basis for carbohydrate exchange. The Ameri- 29. Murakami K, McCaffrey TA, Livingstone MB. Associations of dietary can journal of clinical nutrition. 1981;34(3):362–366. glycaemic index and glycaemic load with food and nutrient intake and 11. Horan MK, McGowan CA, Gibney ER, Donnelly JM, McAuliffe FM. general and central obesity in British adults. The British journal of Maternal diet and weight at 3 months postpartum following a preg- nutrition. 2013;110(11):2047–2057. doi:10.1017/S0007114513001414. nancy intervention with a low glycaemic index diet: Results from the 30. Hui LL, Nelson EA, Choi KC, Wong GW, Sung R. Twelve-hour glyce- ROLO randomised control trial. Nutrients. 2014;6(7):2946–2955. mic profiles with meals of high, medium, or low glycemic load. Diabe- doi:10.3390/nu6072946. tes care. 2005;28(12):2981–2983. doi:10.2337/diacare.28.12.2981. 12. Food and Agriculture Organization/World Health Organization. Car- 31. Gaesser GA. Carbohydrate quantity and quality in relation to body bohydrates in human nutrition. Rome(Italy): Food and Agriculture mass index. Journal of the American Dietetic Association. 2007;107 Organization; 1998. xx p. (10):1768–1780. doi:10.1016/j.jada.2007.07.011. 13. Liu S. Intake of refined carbohydrates and whole grain foods in rela- 32. Kaartinen NE, Knekt P, Kanerva N, Valsta LM, Eriksson JG, Rissanen tion to risk of type 2 diabetes mellitus and coronary heart disease. H, Jaaskelainen T, Mannisto S. Dietary carbohydrate quantity and Journal of the American College of Nutrition. 2002;21(4):298–306. quality in relation to obesity: A pooled analysis of three Finnish popu- doi:10.1080/07315724.2002.10719227. lation-based studies. Scandinavian journal of public health. 2016;44 14. Goss AM, Goree LL, Ellis AC, Chandler-Laney PC, Casazza K, Lock- (4):385–393. doi:10.1177/1403494815622860. hart ME, Gower BA. Effects of diet macronutrient composition on 33. Jones JM, LN C. Glycemic index. Nutrition Today. 2013;48(1):7–16. body composition and fat distribution during weight maintenance doi:10.1097/NT.0b013e31827d8515. and weight loss. Obesity (Silver Spring, Md). 2013;21(6):1139–1142. 34. Halkjaer J, Tjonneland A, Thomsen BL, Overvad K, Sorensen TI. doi:10.1002/oby.20191. Intake of macronutrients as predictors of 5-y changes in waist circum- 15. Esfahani FH, Asghari G, Mirmiran P, Azizi F. Reproducibility and rel- ference. The American journal of clinical nutrition. 2006;84(4):789– ative validity of food group intake in a food frequency questionnaire 797. developed for the tehran lipid and glucose study. Journal of epidemiol- 35. Du H, Van Der xxx A D., Van Bakel M., Slimani N., Forouhi N., Ware- ogy. 2010;20(2):150–158. doi:10.2188/jea.JE20090083. ham, N., Halkjaer J., Tjønneland A., Jakobsen M., Overvad K. Dietary 16. Willett W. Nutritional epidemiology. xx (xx): Oxford University Press; glycaemic index, glycaemic load andsubsequentchangesofweightand 2012. xx p. waist circumference in European men and women. International journal 17. Willett W, Manson J, Liu S. Glycemic index, glycemic load, and risk of of obesity. 2009;33(11):1280. doi:10.1038/ijo.2009.163. type 2 diabetes. The American journal of clinical nutrition. 2002;76 36. Tyrovolas S, Psaltopoulou T, Pounis G, Papairakleous N, Bountziouka (1):274s–280s. V, Zeimbekis A, Gotsis E, Antonopoulou M, Metallinos G, Polychro- 18. Cusin I, Rohner-Jeanrenaud, F., Terrettaz, J., and Jeanrenaud, B. nopoulos E. Nutrient intake in relation to central and overall obesity Hyperinsulinemia and its impact on obesity and insulin resistance. Int status among elderly people living in the mediterranean islands: The J Obes Relat Metab Disord. 1992;16(Suppl 4):S1–S11. MEDIS study. Nutrition, Metabolism and Cardiovascular Diseases. 19. Ludwig DS. The glycemic index: Physiological mechanisms relating to 2011;21(6):438–445. doi:10.1016/j.numecd.2009.10.012. obesity, diabetes, and . Jama. 2002;287 37.KleinS,AllisonDB,Heymsfield SB, Kelley DE, Leibel RL, Nonas (18):2414–2423. doi:10.1001/jama.287.18.2414. C, Kahn R. Waist circumference and cardiometabolic risk: A con- 20. EsfahaniA,Wong,J.M.,Mirrahimi,A.,Srichaikul,K.,Jenkins,D.J.,and sensus statement from Shaping America’s Health: Association for Kendall, C. W. The glycemic index: Physiological significance. J Am Coll AND Obesity Prevention; NAASO, The Nutr. 2009;28(Suppl);439S–445S. doi:10.1080/07315724.2009.10718109. Obesity Society; the American Society for Nutrition; and the 8 A.-M. FUSEINI ET AL.

American Diabetes Association. Obesity. 2007;15(5):1061–1067. Quantitative contribution of muscularity in a population-based doi:10.1038/oby.2007.632. sample. International journal of obesity. 2009;(2005) 33(12):1363. 38. Du H, van der AD, van Bakel MM, Slimani N, Forouhi NG, Wareham doi:10.1038/ijo.2009.184. NJ, Halkjaer J, Tjonneland A, Jakobsen MU, Overvad K, et al. Dietary 43. Qiao Q, Nyamdorj R. Is the association of type ii diabetes with waist glycaemic index, glycaemic load and subsequent changes of weight circumference or waist-to-hip ratio stronger than that with body mass and waist circumference in European men and women. International index? Eur J Clin Nutr. 2010;64(1):30–34. doi:10.1038/ejcn.2009.93. journal of obesity. 2009;(2005) 33(11):1280–1288. doi:10.1038/ 44. Ahima RS, Lazar MA. The health risk of obesity—better metrics impera- ijo.2009.163. tive. Science. 2013;341(6148):856–858. doi:10.1126/science.1241244. 39. Burton-Freeman B, Keim N. Glycemic index, cholecystokinin, satiety 45. Consultation WE. Waist circumference and waist-hip ratio. Report of and disinhibition: Is there an unappreciated paradox for overweight a WHO Expert Consultation. Geneva (Switzerland): World Health women? International Journal of Obesity. 2008;32(11):1647. doi:10.1038/ Organization; 2008. p. 8–11. ijo.2008.159. 46. Krakauer NY, Krakauer JC. A new body shape index predicts mortal- 40. Romaguera D, Angquist€ L, Du H, Jakobsen MU, Forouhi NG, Halkjær ity hazard independently of body mass index. PloS one. 2012;7(7): J, Feskens EJ, Masala G, Steffen A, Palli D. Dietary determinants of e39504. doi:10.1371/journal.pone.0039504. changes in waist circumference adjusted for body mass index—a 47. Maessen MF, Eijsvogels TM, Verheggen RJ, Hopman MT, Verbeek proxy measure of visceral adiposity. PloS one. 2010;5(7):e11588. AL, de Vegt F. Entering a new era of body indices: The feasibility of a doi:10.1371/journal.pone.0011588. body shape index and body roundness index to identify cardiovascular 41. Gopinath B, Flood VM, Rochtchina E, Baur LA, Louie JCY, Smith W, health status. PloS one. 2014;9(9):e107212. doi:10.1371/journal. Mitchell P. Carbohydrate nutrition and development of adiposity dur- pone.0107212. ing adolescence. Obesity. 2013;21(9):1884–1890. 48. He S, Chen X. Could the new body shape index predict the new onset 42. Heymsfield SB, Scherzer R, Pietrobelli A, Lewis CE, Grunfeld C. of diabetes mellitus in the Chinese population? PloS one. 2013;8(1): Body mass index as a phenotypic expression of adiposity: e50573. doi:10.1371/journal.pone.0050573.